void UpResidual_GMRF_5() {
exchsolution_gmrfData_5(0);
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
/* Statements in this Scop: S141 */
for (int i0 = iterationOffsetBegin[0][1]; (i0<=(iterationOffsetEnd[0][1]+32)); i0 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i0*35)]);
double* fieldData_Residual_GMRF_5_p1 = (&fieldData_Residual_GMRF[5][(i0*35)]);
double* fieldData_Solution_GMRF_5_p1 = (&fieldData_Solution_GMRF[5][(i0*35)]);
double* fieldData_RHS_GMRF_5_p1 = (&fieldData_RHS_GMRF[5][(i0*33)]);
int i1 = (iterationOffsetBegin[0][0]+i0);
for (; (i1<(((iterationOffsetBegin[0][0]+i0)+1)&(~1))); i1 += 1) {
fieldData_Residual_GMRF_5_p1[(i1+38)] = (fieldData_RHS_GMRF_5_p1[i1]-(((((((((fieldData_LaplaceCoeff_GMRF_5_p1[(i1+38)]*fieldData_Solution_GMRF_5_p1[(i1+38)])+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+1298)]*fieldData_Solution_GMRF_5_p1[(i1+39)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+2558)]*fieldData_Solution_GMRF_5_p1[(i1+37)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+3818)]*fieldData_Solution_GMRF_5_p1[(i1+74)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+5078)]*fieldData_Solution_GMRF_5_p1[(i1+2)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+6338)]*fieldData_Solution_GMRF_5_p1[(i1+1)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+7598)]*fieldData_Solution_GMRF_5_p1[(i1+73)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+8858)]*fieldData_Solution_GMRF_5_p1[(i1+3)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+10118)]*fieldData_Solution_GMRF_5_p1[(i1+75)])));
}
for (; (i1<((iterationOffsetEnd[0][0]+i0)+30)); i1 += 4) {
/* fieldData_Residual_GMRF_5_p1[(i1+38)] = (fieldData_RHS_GMRF_5_p1[i1]-(((((((((fieldData_LaplaceCoeff_GMRF_5_p1[(i1+38)]*fieldData_Solution_GMRF_5_p1[(i1+38)])+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+1298)]*fieldData_Solution_GMRF_5_p1[(i1+39)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+2558)]*fieldData_Solution_GMRF_5_p1[(i1+37)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+3818)]*fieldData_Solution_GMRF_5_p1[(i1+74)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+5078)]*fieldData_Solution_GMRF_5_p1[(i1+2)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+6338)]*fieldData_Solution_GMRF_5_p1[(i1+1)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+7598)]*fieldData_Solution_GMRF_5_p1[(i1+73)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+8858)]*fieldData_Solution_GMRF_5_p1[(i1+3)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+10118)]*fieldData_Solution_GMRF_5_p1[(i1+75)]))); */
__m128d vec0 = _mm_loadu_pd((&fieldData_RHS_GMRF_5_p1[i1]));
__m128d vec0_2 = _mm_loadu_pd((&fieldData_RHS_GMRF_5_p1[(i1+2)]));
__m128d vec1 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+10118)]));
__m128d vec1_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+10120)]));
__m128d vec2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+75)]));
__m128d vec2_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+77)]));
__m128d vec3 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+8858)]));
__m128d vec3_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+8860)]));
__m128d vec4 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+3)]));
__m128d vec4_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+5)]));
__m128d vec5 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+7598)]));
__m128d vec5_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+7600)]));
__m128d vec6 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+73)]));
__m128d vec6_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+75)]));
__m128d vec7 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+6338)]));
__m128d vec7_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+6340)]));
__m128d vec8 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+1)]));
__m128d vec8_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+3)]));
__m128d vec9 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+5078)]));
__m128d vec9_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+5080)]));
__m128d vec10 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+2)]));
__m128d vec10_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+4)]));
__m128d vec11 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+3818)]));
__m128d vec11_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+3820)]));
__m128d vec12 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+74)]));
__m128d vec12_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+76)]));
__m128d vec13 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+2558)]));
__m128d vec13_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+2560)]));
__m128d vec14 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+37)]));
__m128d vec14_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+39)]));
__m128d vec15 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+38)]));
__m128d vec15_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+40)]));
__m128d vec16 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+38)]));
__m128d vec16_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+40)]));
__m128d vec17 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+1298)]));
__m128d vec17_2 = _mm_loadu_pd((&fieldData_LaplaceCoeff_GMRF_5_p1[(i1+1300)]));
__m128d vec18 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+39)]));
__m128d vec18_2 = _mm_loadu_pd((&fieldData_Solution_GMRF_5_p1[(i1+41)]));
__m128d vec19;
__m128d vec19_2;
vec19 = _mm_sub_pd(vec0, _mm_add_pd(_mm_mul_pd(vec1, vec2), _mm_add_pd(_mm_mul_pd(vec3, vec4), _mm_add_pd(_mm_mul_pd(vec5, vec6), _mm_add_pd(_mm_mul_pd(vec7, vec8), _mm_add_pd(_mm_mul_pd(vec9, vec10), _mm_add_pd(_mm_mul_pd(vec11, vec12), _mm_add_pd(_mm_mul_pd(vec13, vec14), _mm_add_pd(_mm_mul_pd(vec15, vec16), _mm_mul_pd(vec17, vec18))))))))));
vec19_2 = _mm_sub_pd(vec0_2, _mm_add_pd(_mm_mul_pd(vec1_2, vec2_2), _mm_add_pd(_mm_mul_pd(vec3_2, vec4_2), _mm_add_pd(_mm_mul_pd(vec5_2, vec6_2), _mm_add_pd(_mm_mul_pd(vec7_2, vec8_2), _mm_add_pd(_mm_mul_pd(vec9_2, vec10_2), _mm_add_pd(_mm_mul_pd(vec11_2, vec12_2), _mm_add_pd(_mm_mul_pd(vec13_2, vec14_2), _mm_add_pd(_mm_mul_pd(vec15_2, vec16_2), _mm_mul_pd(vec17_2, vec18_2))))))))));
_mm_storeu_pd((&fieldData_Residual_GMRF_5_p1[(i1+38)]), vec19);
_mm_storeu_pd((&fieldData_Residual_GMRF_5_p1[(i1+40)]), vec19_2);
}
for (; (i1<((iterationOffsetEnd[0][0]+i0)+33)); i1 += 1) {
fieldData_Residual_GMRF_5_p1[(i1+38)] = (fieldData_RHS_GMRF_5_p1[i1]-(((((((((fieldData_LaplaceCoeff_GMRF_5_p1[(i1+38)]*fieldData_Solution_GMRF_5_p1[(i1+38)])+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+1298)]*fieldData_Solution_GMRF_5_p1[(i1+39)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+2558)]*fieldData_Solution_GMRF_5_p1[(i1+37)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+3818)]*fieldData_Solution_GMRF_5_p1[(i1+74)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+5078)]*fieldData_Solution_GMRF_5_p1[(i1+2)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+6338)]*fieldData_Solution_GMRF_5_p1[(i1+1)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+7598)]*fieldData_Solution_GMRF_5_p1[(i1+73)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+8858)]*fieldData_Solution_GMRF_5_p1[(i1+3)]))+(fieldData_LaplaceCoeff_GMRF_5_p1[(i1+10118)]*fieldData_Solution_GMRF_5_p1[(i1+75)])));
}
}
}
}
}
void m2l_along_z(long long nmultipoles, double *scr1, double *scr2, double *d2, double *fr, double *sg)
{
int mmmm,mmm,mm,m;
int i,j,k,l,n,nn;
__m128d reg00,reg01,reg02,reg03;
__m128d reg04,reg05,reg06,reg07;
__m128d reg08,reg09,reg10,reg11;
__m128d reg12,reg13,reg14,reg15;
__m128d reg16,reg17; /* register for rotation matrix TODO: rename regdmat1,regdmat2*/
__m128d reg18,reg19; /* register for g,gl,glm */
i = -15;
__m128d regzero = _mm_setzero_pd();
reg08 = regzero;
reg09 = regzero;
reg10 = regzero;
reg11 = regzero;
for(j=0;j<=nmultipoles;++j)
{
i += 16;
reg00 = _mm_load_pd(&scr2[i-1]);
reg01 = _mm_load_pd(&scr2[i+1]);
reg04 = _mm_load_pd(&scr2[i+7]);
reg05 = _mm_load_pd(&scr2[i+9]);
reg18 = _mm_load1_pd(&fr[j]);
reg08 = _mm_add_pd(reg08,_mm_mul_pd(reg00,reg18));
reg09 = _mm_add_pd(reg09,_mm_mul_pd(reg01,reg18));
reg12 = _mm_add_pd(reg12,_mm_mul_pd(reg04,reg18));
reg13 = _mm_add_pd(reg13,_mm_mul_pd(reg05,reg18));
}
_mm_store_pd(&scr1[ 0],reg12);
_mm_store_pd(&scr1[ 2],reg13);
_mm_store_pd(&scr1[ 4],regzero);
_mm_store_pd(&scr1[ 6],regzero);
_mm_store_pd(&scr1[ 8],reg08);
_mm_store_pd(&scr1[10],reg09);
_mm_store_pd(&scr1[12],regzero);
_mm_store_pd(&scr1[14],regzero);
i = 1;
for(l=1;l<=nmultipoles;++l)
{
i += 16 * l;
j = -15;
k = nmultipoles+l;
reg08 = regzero;
reg09 = regzero;
reg12 = regzero;
reg13 = regzero;
for(m=l;m<=k;++m)
{
j += 16;
reg00 = _mm_load_pd(&scr2[j-1]);
reg01 = _mm_load_pd(&scr2[j+1]);
reg04 = _mm_load_pd(&scr2[j+7]);
reg05 = _mm_load_pd(&scr2[j+9]);
reg18 = _mm_load1_pd(&fr[m]);
reg08 = _mm_add_pd(reg08,_mm_mul_pd(reg00,reg18));
reg09 = _mm_add_pd(reg09,_mm_mul_pd(reg01,reg18));
reg12 = _mm_add_pd(reg12,_mm_mul_pd(reg04,reg18));
reg13 = _mm_add_pd(reg13,_mm_mul_pd(reg05,reg18));
}
reg18 = _mm_load1_pd(&sg[l]);
reg12 = _mm_mul_pd(reg12,reg18);
_mm_store_pd(&scr1[i- 1],reg12);
reg13 = _mm_mul_pd(reg13,reg18);
_mm_store_pd(&scr1[i+ 1],reg13);
_mm_store_pd(&scr1[i+ 3],regzero);
_mm_store_pd(&scr1[i+ 5],regzero);
reg08 = _mm_mul_pd(reg08,reg18);
_mm_store_pd(&scr1[i+ 7],reg08);
reg09 = _mm_mul_pd(reg09,reg18);
_mm_store_pd(&scr1[i+ 9],reg09);
_mm_store_pd(&scr1[i+11],regzero);
_mm_store_pd(&scr1[i+13],regzero);
}
mm = 16 * nmultipoles;
i = 1;
n = mm+1;
for(m=1;m<=nmultipoles;++m)
{
i += 16 * m;
j = i;
for(l=m;l<=nmultipoles;++l)
{
j += 16 * l;
nn = n;
k = m + l;
mmm = nmultipoles + l;
reg08 = regzero;
reg09 = regzero;
reg10 = regzero;
reg11 = regzero;
reg12 = regzero;
reg13 = regzero;
reg14 = regzero;
reg15 = regzero;
for(mmmm=k;mmmm<=mmm;++mmmm)
{
nn += 16;
reg00 = _mm_load_pd(&scr2[nn- 1]);
reg01 = _mm_load_pd(&scr2[nn+ 1]);
reg02 = _mm_load_pd(&scr2[nn+ 3]);
reg03 = _mm_load_pd(&scr2[nn+ 5]);
reg04 = _mm_load_pd(&scr2[nn+ 7]);
reg05 = _mm_load_pd(&scr2[nn+ 9]);
reg06 = _mm_load_pd(&scr2[nn+11]);
reg07 = _mm_load_pd(&scr2[nn+13]);
reg18 = _mm_load1_pd(&fr[mmmm]);
reg08 = _mm_add_pd(reg08,_mm_mul_pd(reg00,reg18));
reg09 = _mm_add_pd(reg09,_mm_mul_pd(reg01,reg18));
reg10 = _mm_sub_pd(reg10,_mm_mul_pd(reg02,reg18));
reg11 = _mm_sub_pd(reg11,_mm_mul_pd(reg03,reg18));
reg12 = _mm_add_pd(reg12,_mm_mul_pd(reg04,reg18));
reg13 = _mm_add_pd(reg13,_mm_mul_pd(reg05,reg18));
reg14 = _mm_sub_pd(reg14,_mm_mul_pd(reg06,reg18));
reg15 = _mm_sub_pd(reg15,_mm_mul_pd(reg07,reg18));
}
reg18 = _mm_load1_pd(&sg[k]);
reg12 = _mm_mul_pd(reg12,reg18);
_mm_store_pd(&scr1[j- 1],reg12);
reg13 = _mm_mul_pd(reg13,reg18);
_mm_store_pd(&scr1[j+ 1],reg13);
reg14 = _mm_mul_pd(reg14,reg18);
_mm_store_pd(&scr1[j+ 3],reg14);
reg15 = _mm_mul_pd(reg15,reg18);
_mm_store_pd(&scr1[j+ 5],reg15);
reg08 = _mm_mul_pd(reg08,reg18);
_mm_store_pd(&scr1[j+ 7],reg08);
reg09 = _mm_mul_pd(reg09,reg18);
_mm_store_pd(&scr1[j+ 9],reg09);
reg10 = _mm_mul_pd(reg10,reg18);
_mm_store_pd(&scr1[j+11],reg10);
reg11 = _mm_mul_pd(reg11,reg18);
_mm_store_pd(&scr1[j+13],reg11);
}
n += mm;
mm -= 16;
}
}
static inline __m128d
my_invrsq_pd(__m128d x)
{
const __m128d three = (const __m128d) {3.0f, 3.0f};
const __m128d half = (const __m128d) {0.5f, 0.5f};
__m128 t = _mm_rsqrt_ps(_mm_cvtpd_ps(x)); /* Convert to single precision and do _mm_rsqrt_ps() */
__m128d t1 = _mm_cvtps_pd(t); /* Convert back to double precision */
/* First Newton-Rapson step, accuracy is now 24 bits */
__m128d t2 = _mm_mul_pd(half,_mm_mul_pd(t1,_mm_sub_pd(three,_mm_mul_pd(x,_mm_mul_pd(t1,t1)))));
/* Return second Newton-Rapson step, accuracy 48 bits */
return (__m128d) _mm_mul_pd(half,_mm_mul_pd(t2,_mm_sub_pd(three,_mm_mul_pd(x,_mm_mul_pd(t2,t2)))));
}
/* to extract single integers from a __m128i datatype */
#define _mm_extract_epi64(x, imm) \
_mm_cvtsi128_si32(_mm_srli_si128((x), 4 * (imm)))
void nb_kernel400_x86_64_sse2(int * p_nri,
int * iinr,
int * jindex,
int * jjnr,
int * shift,
double * shiftvec,
double * fshift,
int * gid,
double * pos,
double * faction,
double * charge,
double * p_facel,
double * p_krf,
double * p_crf,
double * Vc,
int * type,
int * p_ntype,
double * vdwparam,
double * Vvdw,
double * p_tabscale,
double * VFtab,
double * invsqrta,
double * dvda,
double * p_gbtabscale,
double * GBtab,
int * p_nthreads,
int * count,
void * mtx,
int * outeriter,
int * inneriter,
double * work)
{
int nri,ntype,nthreads,offset;
int n,ii,is3,ii3,k,nj0,nj1,jnr1,jnr2,j13,j23,ggid;
double facel,krf,crf,tabscl,gbtabscl,vct,vgbt;
double shX,shY,shZ,isai_d,dva;
gmx_gbdata_t *gbdata;
float * gpol;
__m128d ix,iy,iz,jx,jy,jz;
__m128d dx,dy,dz,t1,t2,t3;
__m128d fix,fiy,fiz,rsq11,rinv,r,fscal,rt,eps,eps2;
__m128d q,iq,qq,isai,isaj,isaprod,vcoul,gbscale,dvdai,dvdaj;
__m128d Y,F,G,H,Fp,VV,FF,vgb,fijC,dvdatmp,dvdasum,vctot,vgbtot,n0d;
__m128d xmm0,xmm1,xmm2,xmm3,xmm4,xmm5,xmm6,xmm7,xmm8;
__m128d fac,tabscale,gbtabscale;
__m128i n0,nnn;
const __m128d neg = {-1.0f,-1.0f};
const __m128d zero = {0.0f,0.0f};
const __m128d half = {0.5f,0.5f};
const __m128d two = {2.0f,2.0f};
const __m128d three = {3.0f,3.0f};
gbdata = (gmx_gbdata_t *)work;
gpol = gbdata->gpol;
nri = *p_nri;
ntype = *p_ntype;
nthreads = *p_nthreads;
facel = (*p_facel) * (1.0 - (1.0/gbdata->gb_epsilon_solvent));
krf = *p_krf;
crf = *p_crf;
tabscl = *p_tabscale;
gbtabscl = *p_gbtabscale;
nj1 = 0;
/* Splat variables */
fac = _mm_load1_pd(&facel);
tabscale = _mm_load1_pd(&tabscl);
gbtabscale = _mm_load1_pd(&gbtabscl);
/* Keep compiler happy */
dvdatmp = _mm_setzero_pd();
vgb = _mm_setzero_pd();
dvdaj = _mm_setzero_pd();
isaj = _mm_setzero_pd();
vcoul = _mm_setzero_pd();
t1 = _mm_setzero_pd();
t2 = _mm_setzero_pd();
t3 = _mm_setzero_pd();
jnr1=jnr2=0;
j13=j23=0;
for(n=0;n<nri;n++)
{
is3 = 3*shift[n];
shX = shiftvec[is3];
shY = shiftvec[is3+1];
shZ = shiftvec[is3+2];
nj0 = jindex[n];
nj1 = jindex[n+1];
offset = (nj1-nj0)%2;
ii = iinr[n];
ii3 = ii*3;
ix = _mm_set1_pd(shX+pos[ii3+0]);
iy = _mm_set1_pd(shX+pos[ii3+1]);
iz = _mm_set1_pd(shX+pos[ii3+2]);
q = _mm_set1_pd(charge[ii]);
iq = _mm_mul_pd(fac,q);
isai_d = invsqrta[ii];
isai = _mm_load1_pd(&isai_d);
fix = _mm_setzero_pd();
fiy = _mm_setzero_pd();
fiz = _mm_setzero_pd();
dvdasum = _mm_setzero_pd();
vctot = _mm_setzero_pd();
vgbtot = _mm_setzero_pd();
for(k=nj0;k<nj1-offset; k+=2)
{
jnr1 = jjnr[k];
jnr2 = jjnr[k+1];
j13 = jnr1 * 3;
j23 = jnr2 * 3;
/* Load coordinates */
xmm1 = _mm_loadu_pd(pos+j13); /* x1 y1 */
xmm2 = _mm_loadu_pd(pos+j23); /* x2 y2 */
xmm5 = _mm_load_sd(pos+j13+2); /* z1 - */
xmm6 = _mm_load_sd(pos+j23+2); /* z2 - */
/* transpose */
jx = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(0,0));
jy = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(1,1));
jz = _mm_shuffle_pd(xmm5,xmm6,_MM_SHUFFLE2(0,0));
/* distances */
dx = _mm_sub_pd(ix,jx);
dy = _mm_sub_pd(iy,jy);
dz = _mm_sub_pd(iz,jz);
rsq11 = _mm_add_pd( _mm_add_pd( _mm_mul_pd(dx,dx) , _mm_mul_pd(dy,dy) ) , _mm_mul_pd(dz,dz) );
rinv = my_invrsq_pd(rsq11);
/* Load invsqrta */
isaj = _mm_loadl_pd(isaj,invsqrta+jnr1);
isaj = _mm_loadh_pd(isaj,invsqrta+jnr2);
isaprod = _mm_mul_pd(isai,isaj);
/* Load charges */
q = _mm_loadl_pd(q,charge+jnr1);
q = _mm_loadh_pd(q,charge+jnr2);
qq = _mm_mul_pd(iq,q);
vcoul = _mm_mul_pd(qq,rinv);
fscal = _mm_mul_pd(vcoul,rinv);
qq = _mm_mul_pd(isaprod,qq);
qq = _mm_mul_pd(qq,neg);
gbscale = _mm_mul_pd(isaprod,gbtabscale);
/* Load dvdaj */
dvdaj = _mm_loadl_pd(dvdaj, dvda+jnr1);
dvdaj = _mm_loadh_pd(dvdaj, dvda+jnr2);
r = _mm_mul_pd(rsq11,rinv);
rt = _mm_mul_pd(r,gbscale);
n0 = _mm_cvttpd_epi32(rt);
n0d = _mm_cvtepi32_pd(n0);
eps = _mm_sub_pd(rt,n0d);
eps2 = _mm_mul_pd(eps,eps);
nnn = _mm_slli_epi64(n0,2);
xmm1 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,0))); /* Y1 F1 */
xmm2 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,1))); /* Y2 F2 */
xmm3 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,0))+2); /* G1 H1 */
xmm4 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,1))+2); /* G2 H2 */
Y = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(0,0)); /* Y1 Y2 */
F = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(1,1)); /* F1 F2 */
G = _mm_shuffle_pd(xmm3,xmm4,_MM_SHUFFLE2(0,0)); /* G1 G2 */
H = _mm_shuffle_pd(xmm3,xmm4,_MM_SHUFFLE2(1,1)); /* H1 H2 */
G = _mm_mul_pd(G,eps);
H = _mm_mul_pd(H,eps2);
Fp = _mm_add_pd(F,G);
Fp = _mm_add_pd(Fp,H);
VV = _mm_mul_pd(Fp,eps);
VV = _mm_add_pd(Y,VV);
H = _mm_mul_pd(two,H);
FF = _mm_add_pd(Fp,G);
FF = _mm_add_pd(FF,H);
vgb = _mm_mul_pd(qq,VV);
fijC = _mm_mul_pd(qq,FF);
fijC = _mm_mul_pd(fijC,gbscale);
dvdatmp = _mm_mul_pd(fijC,r);
dvdatmp = _mm_add_pd(vgb,dvdatmp);
dvdatmp = _mm_mul_pd(dvdatmp,neg);
dvdatmp = _mm_mul_pd(dvdatmp,half);
dvdasum = _mm_add_pd(dvdasum,dvdatmp);
xmm1 = _mm_mul_pd(dvdatmp,isaj);
xmm1 = _mm_mul_pd(xmm1,isaj);
dvdaj = _mm_add_pd(dvdaj,xmm1);
/* store dvda */
_mm_storel_pd(dvda+jnr1,dvdaj);
_mm_storeh_pd(dvda+jnr2,dvdaj);
vctot = _mm_add_pd(vctot,vcoul);
vgbtot = _mm_add_pd(vgbtot,vgb);
fscal = _mm_sub_pd(fijC,fscal);
fscal = _mm_mul_pd(fscal,neg);
fscal = _mm_mul_pd(fscal,rinv);
/* calculate partial force terms */
t1 = _mm_mul_pd(fscal,dx);
t2 = _mm_mul_pd(fscal,dy);
t3 = _mm_mul_pd(fscal,dz);
/* update the i force */
fix = _mm_add_pd(fix,t1);
fiy = _mm_add_pd(fiy,t2);
fiz = _mm_add_pd(fiz,t3);
/* accumulate forces from memory */
xmm1 = _mm_loadu_pd(faction+j13); /* fx1 fy1 */
xmm2 = _mm_loadu_pd(faction+j23); /* fx2 fy2 */
xmm5 = _mm_load1_pd(faction+j13+2); /* fz1 fz1 */
xmm6 = _mm_load1_pd(faction+j23+2); /* fz2 fz2 */
/* transpose */
xmm7 = _mm_shuffle_pd(xmm5,xmm6,_MM_SHUFFLE2(0,0)); /* fz1 fz2 */
xmm5 = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(0,0)); /* fx1 fx2 */
xmm6 = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(1,1)); /* fy1 fy2 */
/* subtract partial forces */
xmm5 = _mm_sub_pd(xmm5,t1);
xmm6 = _mm_sub_pd(xmm6,t2);
xmm7 = _mm_sub_pd(xmm7,t3);
xmm1 = _mm_shuffle_pd(xmm5,xmm6,_MM_SHUFFLE2(0,0)); /* fx1 fy1 */
xmm2 = _mm_shuffle_pd(xmm5,xmm6,_MM_SHUFFLE2(1,1)); /* fy1 fy2 */
/* store fx and fy */
_mm_storeu_pd(faction+j13,xmm1);
_mm_storeu_pd(faction+j23,xmm2);
/* .. then fz */
_mm_storel_pd(faction+j13+2,xmm7);
_mm_storel_pd(faction+j23+2,xmm7);
}
/* In double precision, offset can only be either 0 or 1 */
if(offset!=0)
{
jnr1 = jjnr[k];
j13 = jnr1*3;
jx = _mm_load_sd(pos+j13);
jy = _mm_load_sd(pos+j13+1);
jz = _mm_load_sd(pos+j13+2);
isaj = _mm_load_sd(invsqrta+jnr1);
isaprod = _mm_mul_sd(isai,isaj);
dvdaj = _mm_load_sd(dvda+jnr1);
q = _mm_load_sd(charge+jnr1);
qq = _mm_mul_sd(iq,q);
dx = _mm_sub_sd(ix,jx);
dy = _mm_sub_sd(iy,jy);
dz = _mm_sub_sd(iz,jz);
rsq11 = _mm_add_pd( _mm_add_pd( _mm_mul_pd(dx,dx) , _mm_mul_pd(dy,dy) ) , _mm_mul_pd(dz,dz) );
rinv = my_invrsq_pd(rsq11);
vcoul = _mm_mul_sd(qq,rinv);
fscal = _mm_mul_sd(vcoul,rinv);
qq = _mm_mul_sd(isaprod,qq);
qq = _mm_mul_sd(qq,neg);
gbscale = _mm_mul_sd(isaprod,gbtabscale);
r = _mm_mul_sd(rsq11,rinv);
rt = _mm_mul_sd(r,gbscale);
n0 = _mm_cvttpd_epi32(rt);
n0d = _mm_cvtepi32_pd(n0);
eps = _mm_sub_sd(rt,n0d);
eps2 = _mm_mul_sd(eps,eps);
nnn = _mm_slli_epi64(n0,2);
xmm1 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,0)));
xmm2 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,1)));
xmm3 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,0))+2);
xmm4 = _mm_load_pd(GBtab+(_mm_extract_epi64(nnn,1))+2);
Y = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(0,0));
F = _mm_shuffle_pd(xmm1,xmm2,_MM_SHUFFLE2(1,1));
G = _mm_shuffle_pd(xmm3,xmm4,_MM_SHUFFLE2(0,0));
H = _mm_shuffle_pd(xmm3,xmm4,_MM_SHUFFLE2(1,1));
G = _mm_mul_sd(G,eps);
H = _mm_mul_sd(H,eps2);
Fp = _mm_add_sd(F,G);
Fp = _mm_add_sd(Fp,H);
VV = _mm_mul_sd(Fp,eps);
VV = _mm_add_sd(Y,VV);
H = _mm_mul_sd(two,H);
FF = _mm_add_sd(Fp,G);
FF = _mm_add_sd(FF,H);
vgb = _mm_mul_sd(qq,VV);
fijC = _mm_mul_sd(qq,FF);
fijC = _mm_mul_sd(fijC,gbscale);
dvdatmp = _mm_mul_sd(fijC,r);
dvdatmp = _mm_add_sd(vgb,dvdatmp);
dvdatmp = _mm_mul_sd(dvdatmp,neg);
dvdatmp = _mm_mul_sd(dvdatmp,half);
dvdasum = _mm_add_sd(dvdasum,dvdatmp);
xmm1 = _mm_mul_sd(dvdatmp,isaj);
xmm1 = _mm_mul_sd(xmm1,isaj);
dvdaj = _mm_add_sd(dvdaj,xmm1);
/* store dvda */
_mm_storel_pd(dvda+jnr1,dvdaj);
vctot = _mm_add_sd(vctot,vcoul);
vgbtot = _mm_add_sd(vgbtot,vgb);
fscal = _mm_sub_sd(fijC,fscal);
fscal = _mm_mul_sd(fscal,neg);
fscal = _mm_mul_sd(fscal,rinv);
/* calculate partial force terms */
t1 = _mm_mul_sd(fscal,dx);
t2 = _mm_mul_sd(fscal,dy);
t3 = _mm_mul_sd(fscal,dz);
/* update the i force */
fix = _mm_add_sd(fix,t1);
fiy = _mm_add_sd(fiy,t2);
fiz = _mm_add_sd(fiz,t3);
/* accumulate forces from memory */
xmm5 = _mm_load_sd(faction+j13); /* fx */
xmm6 = _mm_load_sd(faction+j13+1); /* fy */
xmm7 = _mm_load_sd(faction+j13+2); /* fz */
/* subtract partial forces */
xmm5 = _mm_sub_sd(xmm5,t1);
xmm6 = _mm_sub_sd(xmm6,t2);
xmm7 = _mm_sub_sd(xmm7,t3);
/* store forces */
_mm_store_sd(faction+j13,xmm5);
_mm_store_sd(faction+j13+1,xmm6);
_mm_store_sd(faction+j13+2,xmm7);
}
/* fix/fiy/fiz now contain four partial terms, that all should be
* added to the i particle forces
*/
t1 = _mm_unpacklo_pd(t1,fix);
t2 = _mm_unpacklo_pd(t2,fiy);
t3 = _mm_unpacklo_pd(t3,fiz);
fix = _mm_add_pd(fix,t1);
fiy = _mm_add_pd(fiy,t2);
fiz = _mm_add_pd(fiz,t3);
fix = _mm_shuffle_pd(fix,fix,_MM_SHUFFLE2(1,1));
fiy = _mm_shuffle_pd(fiy,fiy,_MM_SHUFFLE2(1,1));
fiz = _mm_shuffle_pd(fiz,fiz,_MM_SHUFFLE2(1,1));
/* Load i forces from memory */
xmm1 = _mm_load_sd(faction+ii3);
xmm2 = _mm_load_sd(faction+ii3+1);
xmm3 = _mm_load_sd(faction+ii3+2);
/* Add to i force */
fix = _mm_add_sd(fix,xmm1);
fiy = _mm_add_sd(fiy,xmm2);
fiz = _mm_add_sd(fiz,xmm3);
/* store i forces to memory */
_mm_store_sd(faction+ii3,fix);
_mm_store_sd(faction+ii3+1,fiy);
_mm_store_sd(faction+ii3+2,fiz);
/* now do dvda */
dvdatmp = _mm_unpacklo_pd(dvdatmp,dvdasum);
dvdasum = _mm_add_pd(dvdasum,dvdatmp);
_mm_storeh_pd(&dva,dvdasum);
dvda[ii] = dvda[ii] + dva*isai_d*isai_d;
ggid = gid[n];
/* Coulomb potential */
vcoul = _mm_unpacklo_pd(vcoul,vctot);
vctot = _mm_add_pd(vctot,vcoul);
_mm_storeh_pd(&vct,vctot);
Vc[ggid] = Vc[ggid] + vct;
/* GB potential */
vgb = _mm_unpacklo_pd(vgb,vgbtot);
vgbtot = _mm_add_pd(vgbtot,vgb);
_mm_storeh_pd(&vgbt,vgbtot);
gpol[ggid] = gpol[ggid] + vgbt;
}
*outeriter = nri;
*inneriter = nj1;
}
void exchlaplacecoeff_gmrfData_0(unsigned int slot) {
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((!neighbor_isValid[0][0])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1053, S1056, S1059, S1050, S1058, S1052, S1055, S1060, S1054, S1057, S1051 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+26)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+32)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+26)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+146)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+152)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+146)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+98)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+104)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+98)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+74)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+80)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+74)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posBegin[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<0); i1 += 4) {
/* xPos = posBegin[0]; */
__m128d vec0 = _mm_load1_pd((&posBegin[0]));
__m128d vec0_2 = _mm_load1_pd((&posBegin[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<3); i1 += 1) {
xPos = posBegin[0];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+122)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+128)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+122)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+170)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+176)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+170)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+194)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+200)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+194)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(1.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<0); i1 += 4) {
/* yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<3); i1 += 1) {
yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+2)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+8)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+2)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+50)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+56)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+50)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[0][1])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1071, S1065, S1068, S1062, S1070, S1064, S1067, S1061, S1069, S1063, S1066 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+195)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+201)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+195)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+51)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+57)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+51)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+75)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+81)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+75)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+9)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+171)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+177)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+171)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(1.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<0); i1 += 4) {
/* yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<3); i1 += 1) {
yPos = ((((i1-1)/1.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+99)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+105)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+99)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+123)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+129)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+123)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+147)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+153)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+147)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posEnd[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<0); i1 += 4) {
/* xPos = posEnd[0]; */
__m128d vec0 = _mm_load1_pd((&posEnd[0]));
__m128d vec0_2 = _mm_load1_pd((&posEnd[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<3); i1 += 1) {
xPos = posEnd[0];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+27)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+33)] = 0.000000e+00;
}
for (; (i1<=2); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+27)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[0][2])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1080, S1074, S1077, S1082, S1076, S1079, S1073, S1072, S1081, S1075, S1078 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+126)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+127)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+126)] = 0.000000e+00;
}
}
{
int i2 = 2;
for (; (i2<=2); i2 += 2) {
xPos = ((((i2-2)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=3); i2 += 1) {
xPos = ((((i2-2)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+198)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+199)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+198)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=2); i2 += 2) {
yPos = posBegin[1];
yPos = posBegin[1];
}
for (; (i2<=3); i2 += 1) {
yPos = posBegin[1];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+30)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+31)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+30)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+174)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+175)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+174)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+78)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+79)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+78)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+54)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+55)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+54)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+150)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+151)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+150)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+6)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+7)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+6)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+102)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+103)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+102)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[0][3])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1083, S1092, S1086, S1089, S1088, S1091, S1085, S1090, S1093, S1087, S1084 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+12)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+13)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+12)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+60)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+61)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+60)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+204)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+205)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+204)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+132)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+133)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+132)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+84)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+85)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+84)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=2); i2 += 2) {
yPos = posEnd[1];
yPos = posEnd[1];
}
for (; (i2<=3); i2 += 1) {
yPos = posEnd[1];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+36)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+37)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+36)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=2); i2 += 2) {
xPos = ((((i2-2)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=3); i2 += 1) {
xPos = ((((i2-2)/1.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+180)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+181)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+180)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+156)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+157)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+156)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][0]);
int i2 = 2;
for (; (i2<=2); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+108)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+109)] = 0.000000e+00;
}
for (; (i2<=3); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[(i2+108)] = 0.000000e+00;
}
}
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1094 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* buffer_Send_1_p1 = (&buffer_Send[1][(i0*2)]);
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
int i1 = 1;
for (; (i1<=1); i1 += 2) {
buffer_Send_1_p1[(i1-1)] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)];
buffer_Send_1_p1[i1] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+9)];
}
for (; (i1<=2); i1 += 1) {
buffer_Send_1_p1[(i1-1)] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(buffer_Send[1], 18, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(buffer_Recv[0], 18, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1095 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i0*2)]);
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
int i1 = 3;
for (; (i1<=3); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-10)] = buffer_Recv_0_p1[(i1-3)];
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-4)] = buffer_Recv_0_p1[(i1-2)];
}
for (; (i1<=4); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-10)] = buffer_Recv_0_p1[(i1-3)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[0][14], 1, mpiDatatype_9_2_24, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[0][8], 1, mpiDatatype_9_2_24, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1096 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
double* buffer_Send_0_p1 = (&buffer_Send[0][(i0*4)]);
int i1 = 0;
for (; (i1<=2); i1 += 2) {
buffer_Send_0_p1[i1] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)];
buffer_Send_0_p1[(i1+1)] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+9)];
}
for (; (i1<=3); i1 += 1) {
buffer_Send_0_p1[i1] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+3)];
}
}
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1097 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* buffer_Send_1_p1 = (&buffer_Send[1][(i0*4)]);
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
int i1 = 0;
for (; (i1<=2); i1 += 2) {
buffer_Send_1_p1[i1] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+2)];
buffer_Send_1_p1[(i1+1)] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+8)];
}
for (; (i1<=3); i1 += 1) {
buffer_Send_1_p1[i1] = fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+2)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Isend(buffer_Send[0], 36, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][0]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[0]);
reqOutstanding_Send[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(buffer_Send[1], 36, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(buffer_Recv[0], 36, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Irecv(buffer_Recv[1], 36, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)(neighbor_fragCommId[0][1]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
if (reqOutstanding_Recv[1]) {
waitForMPIReq(&mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1098 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i0*4)]);
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
int i1 = 1;
for (; (i1<=3); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-5)] = buffer_Recv_0_p1[(i1-1)];
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)+1)] = buffer_Recv_0_p1[i1];
}
for (; (i1<=4); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-5)] = buffer_Recv_0_p1[(i1-1)];
}
}
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1099 */
for (int i0 = 0; (i0<=8); i0 += 1) {
double* buffer_Recv_1_p1 = (&buffer_Recv[1][(i0*4)]);
double* fieldData_LaplaceCoeff_GMRF_0_p1 = (&fieldData_LaplaceCoeff_GMRF[0][(i0*24)]);
int i1 = 4;
for (; (i1<=6); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-20)] = buffer_Recv_1_p1[(i1-4)];
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-14)] = buffer_Recv_1_p1[(i1-3)];
}
for (; (i1<=7); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_0_p1[((i1*6)-20)] = buffer_Recv_1_p1[(i1-4)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[0]) {
waitForMPIReq(&mpiRequest_Send[0]);
reqOutstanding_Send[0] = false;
}
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[0][13], 1, mpiDatatype_9_4_24, neighbor_remoteRank[0][2], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][2]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[2]);
reqOutstanding_Send[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[0][7], 1, mpiDatatype_9_4_24, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[0][1], 1, mpiDatatype_9_4_24, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[0][19], 1, mpiDatatype_9_4_24, neighbor_remoteRank[0][3], ((unsigned int)(neighbor_fragCommId[0][3]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
if (reqOutstanding_Recv[3]) {
waitForMPIReq(&mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[2]) {
waitForMPIReq(&mpiRequest_Send[2]);
reqOutstanding_Send[2] = false;
}
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
}
inline void vwl_s1( __m128d const vqn, __m128d const vx0, __m128d const vx1, __m128d vwl[] ) {
vwl[0] = _mm_mul_pd( vqn, _mm_sub_pd( vx1, vx0 ) );
}
namespace nt2 { namespace ext
{
template<class Dummy>
struct call< tag::minus_ ( tag::simd_<tag::double_,tag::sse_>
, tag::simd_<tag::double_,tag::sse_>
)
, tag::cpu_, Dummy
>
: callable
{
template<class Sig> struct result;
template<class This,class A> struct result<This(A,A)> : meta::strip<A> {};
NT2_FUNCTOR_CALL(2)
{
A0 that = { _mm_sub_pd(a0,a1) };
return that;
}
};
template<class Dummy>
struct call< tag::minus_ ( tag::simd_<tag::float_,tag::sse_>
, tag::simd_<tag::float_,tag::sse_>
)
, tag::cpu_, Dummy
>
: callable
{
template<class Sig> struct result;
template<class This,class A> struct result<This(A,A)> : meta::strip<A> {};
// only compute the necessary indices of su2_i = subgroup( U*staple^\dagger )
void
only_subgroup( GLU_complex *s0 ,
GLU_complex *s1 ,
double *scale ,
const GLU_complex U[ NCNC ] ,
const GLU_complex staple[ NCNC ] ,
const size_t su2_index )
{
const __m128d *u = (const __m128d*)U ;
const __m128d *s = (const __m128d*)staple ;
register __m128d sm0 ;
register __m128d sm1 ;
#if NC == 3
switch( su2_index%3 ) { // I don't like this
// rotation 1
// | s0 s1 0 |
// | -s1* s0* 0 |
// | 0 0 1 |
case 0 :
sm0 = _mm_add_pd(
// temp0
_mm_add_pd( SSE2_MUL_CONJ( *( u + 0 ) , *( s + 0 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 1 ) , *( s + 1 ) ) ,
SSE2_MUL_CONJ( *( u + 2 ) , *( s + 2 ) ) ) ) ,
// temp3^*
_mm_add_pd( SSE2_MULCONJ( *( u + 3 ) , *( s + 3 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 4 ) , *( s + 4 ) ) ,
SSE2_MULCONJ( *( u + 5 ) , *( s + 5 ) ) ) ) ) ;
sm1 = _mm_sub_pd(
// temp1
_mm_add_pd( SSE2_MUL_CONJ( *( u + 0 ) , *( s + 3 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 1 ) , *( s + 4 ) ) ,
SSE2_MUL_CONJ( *( u + 2 ) , *( s + 5 ) ) ) ) ,
// temp2^*
_mm_add_pd( SSE2_MULCONJ( *( u + 3 ) , *( s + 0 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 4 ) , *( s + 1 ) ) ,
SSE2_MULCONJ( *( u + 5 ) , *( s + 2 ) ) ) ) ) ;
break ;
case 1 :
// rotation 2
// | 1 0 0 |
// | 0 s0 s1 |
// | 0 -s1* s0* |
sm0 = _mm_add_pd(
// temp0
_mm_add_pd( SSE2_MUL_CONJ( *( u + 3 ) , *( s + 3 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 4 ) , *( s + 4 ) ) ,
SSE2_MUL_CONJ( *( u + 5 ) , *( s + 5 ) ) ) ) ,
// temp3^*
_mm_add_pd( SSE2_MULCONJ( *( u + 6 ) , *( s + 6 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 7 ) , *( s + 7 ) ) ,
SSE2_MULCONJ( *( u + 8 ) , *( s + 8 ) ) ) ) ) ;
sm1 = _mm_sub_pd(
// temp1
_mm_add_pd( SSE2_MUL_CONJ( *( u + 3 ) , *( s + 6 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 4 ) , *( s + 7 ) ) ,
SSE2_MUL_CONJ( *( u + 5 ) , *( s + 8 ) ) ) ) ,
// temp2^*
_mm_add_pd( SSE2_MULCONJ( *( u + 6 ) , *( s + 3 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 7 ) , *( s + 4 ) ) ,
SSE2_MULCONJ( *( u + 8 ) , *( s + 5 ) ) ) ) ) ;
break ;
case 2 :
// rotation 3
// | s0* 0 -s1 |
// | 0 1 0 |
// | s1 0 s0 |
sm0 = _mm_add_pd(
// temp3^*
_mm_add_pd( SSE2_MULCONJ( *( u + 0 ) , *( s + 0 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 1 ) , *( s + 1 ) ) ,
SSE2_MULCONJ( *( u + 2 ) , *( s + 2 ) ) ) ) ,
// temp0
_mm_add_pd( SSE2_MUL_CONJ( *( u + 6 ) , *( s + 6 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 7 ) , *( s + 7 ) ) ,
SSE2_MUL_CONJ( *( u + 8 ) , *( s + 8 ) ) ) ) ) ;
sm1 = _mm_sub_pd(
// temp1
_mm_add_pd( SSE2_MUL_CONJ( *( u + 6 ) , *( s + 0 ) ) ,
_mm_add_pd( SSE2_MUL_CONJ( *( u + 7 ) , *( s + 1 ) ) ,
SSE2_MUL_CONJ( *( u + 8 ) , *( s + 2 ) ) ) ) ,
// temp2^*
_mm_add_pd( SSE2_MULCONJ( *( u + 0 ) , *( s + 6 ) ) ,
_mm_add_pd( SSE2_MULCONJ( *( u + 1 ) , *( s + 7 ) ) ,
SSE2_MULCONJ( *( u + 2 ) , *( s + 8 ) ) ) ) ) ;
break ;
}
#elif NC == 2
sm0 = _mm_add_pd(
// temp0
_mm_add_pd( SSE2_MUL_CONJ( *( u + 0 ) , *( s + 0 ) ) ,
SSE2_MUL_CONJ( *( u + 1 ) , *( s + 1 ) ) ) ,
// temp3^*
_mm_add_pd( SSE2_MULCONJ( *( u + 2 ) , *( s + 2 ) ) ,
SSE2_MULCONJ( *( u + 3 ) , *( s + 3 ) ) ) ) ;
sm1 = _mm_sub_pd(
// temp1
_mm_add_pd( SSE2_MUL_CONJ( *( u + 0 ) , *( s + 2 ) ) ,
SSE2_MUL_CONJ( *( u + 1 ) , *( s + 3 ) ) ) ,
// temp2^*
_mm_add_pd( SSE2_MULCONJ( *( u + 2 ) , *( s + 0 ) ) ,
SSE2_MULCONJ( *( u + 3 ) , *( s + 1 ) ) ) ) ;
#else
// su(N) version
const size_t row_a = Latt.su2_data[ su2_index ].idx_a / NC ;
const size_t col_b = Latt.su2_data[ su2_index ].idx_b % NC ;
// prefetch the staple & link indices
const __m128d *S1 = ( s + NC * row_a ) , *S2 = ( s + NC * col_b ) ;
const __m128d *U1 = ( u + NC * row_a ) , *U2 = ( u + NC * col_b ) ;
// initialise to zero & perform multiplication
sm0 = _mm_setzero_pd() ; sm1 = _mm_setzero_pd() ;
size_t i ;
for( i = 0 ; i < NC ; i++ ) {
sm0 = _mm_add_pd( sm0 ,
_mm_add_pd( SSE2_MUL_CONJ( *U1 , *S1 ) ,
SSE2_MULCONJ( *U2 , *S2 ) ) ) ;
sm1 = _mm_add_pd( sm1 ,
_mm_sub_pd( SSE2_MUL_CONJ( *U1 , *S2 ) ,
SSE2_MULCONJ( *U2 , *S1 ) ) ) ;
// increment our pointers
S1++ , S2++ , U1++ , U2++ ;
}
#endif
// puts the norm in both parts
register __m128d z = SSE2_FMA( sm0 , sm0 , _mm_mul_pd( sm1 , sm1 ) ) ;
z = _mm_add_pd( z , _mm_shuffle_pd( z , z , 1 ) ) ;
z = _mm_sqrt_pd( z ) ;
z = _mm_div_pd( _mm_set1_pd( 1.0 ) , z ) ;
sm0 = _mm_mul_pd( sm0 , z ) ;
sm1 = _mm_mul_pd( sm1 , z ) ;
// poke back into *s0 and *s1 and *scale
_mm_store_pd( (void*)s0 , sm0 ) ;
_mm_store_pd( (void*)s1 , sm1 ) ;
_mm_store_sd( (void*)scale , z ) ;
return ;
}
int fft5b_(double *a, double *b, double *w, int *m, int *l)
{
/* static double c51 = .95105651629515357;
static double c52 = .61803398874989485;
static double c53 = .55901699437494742;
static double c54 = .25; */
static __m128d c51, c52, c53, c54;
int i, i0, i1, i2, i3, i4, i5, i6, i7, i8, i9, j, j0;
/* double x0, y0, x1, y1, x2, y2, x3, y3, x4, y4, x5, y5, x6, y6, x7, y7,
x8, y8, x9, y9, x10, y10, wi1, wi2, wi3, wi4, wr1, wr2, wr3, wr4; */
__m128d t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, w1, w2, w3, w4;
c51 = _mm_set1_pd(0.95105651629515357);
c52 = _mm_set1_pd(0.61803398874989485);
c53 = _mm_set1_pd(0.55901699437494742);
c54 = _mm_set1_pd(0.25);
for (i = 0; i < *m; i++) {
i0 = i << 1;
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = i3 + (*m * *l << 1);
i5 = i << 1;
i6 = i5 + (*m << 1);
i7 = i6 + (*m << 1);
i8 = i7 + (*m << 1);
i9 = i8 + (*m << 1);
/* x0 = a[i1] + a[i4];
y0 = a[i1 + 1] + a[i4 + 1];
x1 = a[i2] + a[i3];
y1 = a[i2 + 1] + a[i3 + 1];
x2 = c51 * (a[i1] - a[i4]);
y2 = c51 * (a[i1 + 1] - a[i4 + 1]);
x3 = c51 * (a[i2] - a[i3]);
y3 = c51 * (a[i2 + 1] - a[i3 + 1]);
x4 = x0 + x1;
y4 = y0 + y1;
x5 = c53 * (x0 - x1);
y5 = c53 * (y0 - y1);
x6 = a[i0] - c54 * x4;
y6 = a[i0 + 1] - c54 * y4;
x7 = x6 + x5;
y7 = y6 + y5;
x8 = x6 - x5;
y8 = y6 - y5;
x9 = y2 + c52 * y3;
y9 = -x2 - c52 * x3;
x10 = c52 * y2 - y3;
y10 = x3 - c52 * x2; */
t1 = _mm_load_pd(&a[i1]);
t4 = _mm_load_pd(&a[i4]);
t0 = _mm_add_pd(t1, t4);
t2 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_load_pd(&a[i2]);
t4 = _mm_load_pd(&a[i3]);
t3 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_add_pd(t1, t4);
t4 = _mm_add_pd(t0, t1);
t5 = _mm_mul_pd(c53, _mm_sub_pd(t0, t1));
t0 = _mm_load_pd(&a[i0]);
t6 = _mm_sub_pd(t0, _mm_mul_pd(c54, t4));
t7 = _mm_add_pd(t6, t5);
t8 = _mm_sub_pd(t6, t5);
t9 = _mm_xor_pd(_mm_add_pd(t2, _mm_mul_pd(c52, t3)), _mm_set_sd(-0.0));
t9 = _mm_shuffle_pd(t9, t9, 1);
t10 = _mm_sub_pd(t3, _mm_mul_pd(c52, t2));
t10 = _mm_xor_pd(_mm_shuffle_pd(t10, t10, 1), _mm_set_sd(-0.0));
/* b[i5] = a[i0] + x4;
b[i5 + 1] = a[i0 + 1] + y4;
b[i6] = x7 + x9;
b[i6 + 1] = y7 + y9;
b[i7] = x8 + x10;
b[i7 + 1] = y8 + y10;
b[i8] = x8 - x10;
b[i8 + 1] = y8 - y10;
b[i9] = x7 - x9;
b[i9 + 1] = y7 - y9; */
_mm_store_pd(&b[i5], _mm_add_pd(t0, t4));
_mm_store_pd(&b[i6], _mm_add_pd(t7, t9));
_mm_store_pd(&b[i7], _mm_add_pd(t8, t10));
_mm_store_pd(&b[i8], _mm_sub_pd(t8, t10));
_mm_store_pd(&b[i9], _mm_sub_pd(t7, t9));
}
for (j = 1; j < *l; j++) {
j0 = j << 1;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1;
wr3 = wr1 * wr2 - wi1 * wi2;
wi3 = wr1 * wi2 + wi1 * wr2;
wr4 = wr2 * wr2 - wi2 * wi2;
wi4 = wr2 * wi2 + wr2 * wi2; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
w3 = ZMUL(w1, w2);
w4 = ZMUL(w2, w2);
for (i = 0; i < *m; i++) {
i0 = (i << 1) + (j * *m << 1);
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = i3 + (*m * *l << 1);
i5 = (i << 1) + (j * *m * 10);
i6 = i5 + (*m << 1);
i7 = i6 + (*m << 1);
i8 = i7 + (*m << 1);
i9 = i8 + (*m << 1);
/* x0 = a[i1] + a[i4];
y0 = a[i1 + 1] + a[i4 + 1];
x1 = a[i2] + a[i3];
y1 = a[i2 + 1] + a[i3 + 1];
x2 = c51 * (a[i1] - a[i4]);
y2 = c51 * (a[i1 + 1] - a[i4 + 1]);
x3 = c51 * (a[i2] - a[i3]);
y3 = c51 * (a[i2 + 1] - a[i3 + 1]);
x4 = x0 + x1;
y4 = y0 + y1;
x5 = c53 * (x0 - x1);
y5 = c53 * (y0 - y1);
x6 = a[i0] - c54 * x4;
y6 = a[i0 + 1] - c54 * y4;
x7 = x6 + x5;
y7 = y6 + y5;
x8 = x6 - x5;
y8 = y6 - y5;
x9 = y2 + c52 * y3;
y9 = -x2 - c52 * x3;
x10 = c52 * y2 - y3;
y10 = x3 - c52 * x2; */
t1 = _mm_load_pd(&a[i1]);
t4 = _mm_load_pd(&a[i4]);
t0 = _mm_add_pd(t1, t4);
t2 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_load_pd(&a[i2]);
t4 = _mm_load_pd(&a[i3]);
t3 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_add_pd(t1, t4);
t4 = _mm_add_pd(t0, t1);
t5 = _mm_mul_pd(c53, _mm_sub_pd(t0, t1));
t0 = _mm_load_pd(&a[i0]);
t6 = _mm_sub_pd(t0, _mm_mul_pd(c54, t4));
t7 = _mm_add_pd(t6, t5);
t8 = _mm_sub_pd(t6, t5);
t9 = _mm_xor_pd(_mm_add_pd(t2, _mm_mul_pd(c52, t3)), _mm_set_sd(-0.0));
t9 = _mm_shuffle_pd(t9, t9, 1);
t10 = _mm_sub_pd(t3, _mm_mul_pd(c52, t2));
t10 = _mm_xor_pd(_mm_shuffle_pd(t10, t10, 1), _mm_set_sd(-0.0));
/* b[i5] = a[i0] + x4;
b[i5 + 1] = a[i0 + 1] + y4;
b[i6] = wr1 * (x7 + x9) - wi1 * (y7 + y9);
b[i6 + 1] = wr1 * (y7 + y9) + wi1 * (x7 + x9);
b[i7] = wr2 * (x8 + x10) - wi2 * (y8 + y10);
b[i7 + 1] = wr2 * (y8 + y10) + wi2 * (x8 + x10);
b[i8] = wr3 * (x8 - x10) - wi3 * (y8 - y10);
b[i8 + 1] = wr3 * (y8 - y10) + wi3 * (x8 - x10);
b[i9] = wr4 * (x7 - x9) - wi4 * (y7 - y9);
b[i9 + 1] = wr4 * (y7 - y9) + wi4 * (x7 - x9); */
_mm_store_pd(&b[i5], _mm_add_pd(t0, t4));
_mm_store_pd(&b[i6], ZMUL(w1, _mm_add_pd(t7, t9)));
_mm_store_pd(&b[i7], ZMUL(w2, _mm_add_pd(t8, t10)));
_mm_store_pd(&b[i8], ZMUL(w3, _mm_sub_pd(t8, t10)));
_mm_store_pd(&b[i9], ZMUL(w4, _mm_sub_pd(t7, t9)));
}
}
return 0;
}
int fft8a_(double *a, double *b, double *w, int *l)
{
/* static double c81 = .70710678118654752; */
static __m128d c81;
int j, j0, j1, j2, j3, j4, j5, j6, j7, j8, j9, j10, j11, j12, j13, j14, j15;
/* double u0, v0, u1, x0, y0, x1, y1, x2, y2, x3, y3, v1, x4, y4, x5, y5,
x6, y6, x7, y7, u2, v2, u3, v3, wi1, wi2, wi3, wi4, wi5, wi6,
wi7, wr1, wr2, wr3, wr4, wr5, wr6, wr7; */
__m128d t0, t1, t2, t3, t4, t5, t6, t7, t8, u0, u1, u2, u3, w1, w2, w3, w4, w5, w6, w7;
c81 = _mm_set1_pd(0.70710678118654752);
for (j = 0; j < *l; j++) {
j0 = j << 1;
j1 = j0 + (*l << 1);
j2 = j1 + (*l << 1);
j3 = j2 + (*l << 1);
j4 = j3 + (*l << 1);
j5 = j4 + (*l << 1);
j6 = j5 + (*l << 1);
j7 = j6 + (*l << 1);
j8 = j << 4;
j9 = j8 + 2;
j10 = j9 + 2;
j11 = j10 + 2;
j12 = j11 + 2;
j13 = j12 + 2;
j14 = j13 + 2;
j15 = j14 + 2;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1;
wr3 = wr1 * wr2 - wi1 * wi2;
wi3 = wr1 * wi2 + wi1 * wr2;
wr4 = wr2 * wr2 - wi2 * wi2;
wi4 = wr2 * wi2 + wr2 * wi2;
wr5 = wr2 * wr3 - wi2 * wi3;
wi5 = wr2 * wi3 + wi2 * wr3;
wr6 = wr3 * wr3 - wi3 * wi3;
wi6 = wr3 * wi3 + wr3 * wi3;
wr7 = wr3 * wr4 - wi3 * wi4;
wi7 = wr3 * wi4 + wi3 * wr4; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
w3 = ZMUL(w1, w2);
w4 = ZMUL(w2, w2);
w5 = ZMUL(w2, w3);
w6 = ZMUL(w3, w3);
w7 = ZMUL(w3, w4);
/* x0 = a[j0] + a[j4];
y0 = a[j0 + 1] + a[j4 + 1];
x1 = a[j0] - a[j4];
y1 = a[j0 + 1] - a[j4 + 1];
x2 = a[j2] + a[j6];
y2 = a[j2 + 1] + a[j6 + 1];
x3 = a[j2 + 1] - a[j6 + 1];
y3 = a[j6] - a[j2]; */
t0 = _mm_load_pd(&a[j0]);
t2 = _mm_load_pd(&a[j4]);
t1 = _mm_sub_pd(t0, t2);
t0 = _mm_add_pd(t0, t2);
t3 = _mm_load_pd(&a[j2]);
t4 = _mm_load_pd(&a[j6]);
t2 = _mm_add_pd(t3, t4);
t3 = _mm_xor_pd(_mm_sub_pd(t3, t4), _mm_set_sd(-0.0));
t3 = _mm_shuffle_pd(t3, t3, 1);
/* u0 = x0 + x2;
v0 = y0 + y2;
u1 = x0 - x2;
v1 = y0 - y2; */
u0 = _mm_add_pd(t0, t2);
u1 = _mm_sub_pd(t0, t2);
/* x4 = a[j1] + a[j5];
y4 = a[j1 + 1] + a[j5 + 1];
x5 = a[j1] - a[j5];
y5 = a[j1 + 1] - a[j5 + 1];
x6 = a[j3] + a[j7];
y6 = a[j3 + 1] + a[j7 + 1];
x7 = a[j3] - a[j7];
y7 = a[j3 + 1] - a[j7 + 1]; */
t4 = _mm_load_pd(&a[j1]);
t6 = _mm_load_pd(&a[j5]);
t5 = _mm_sub_pd(t4, t6);
t4 = _mm_add_pd(t4, t6);
t7 = _mm_load_pd(&a[j3]);
t8 = _mm_load_pd(&a[j7]);
t6 = _mm_add_pd(t7, t8);
t7 = _mm_sub_pd(t7, t8);
/* u2 = x4 + x6;
v2 = y4 + y6;
u3 = y4 - y6;
v3 = x6 - x4; */
u2 = _mm_add_pd(t4, t6);
u3 = _mm_xor_pd(_mm_sub_pd(t4, t6), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
/* b[j8] = u0 + u2;
b[j8 + 1] = v0 + v2;
b[j12] = wr4 * (u0 - u2) - wi4 * (v0 - v2);
b[j12 + 1] = wr4 * (v0 - v2) + wi4 * (u0 - u2);
b[j10] = wr2 * (u1 + u3) - wi2 * (v1 + v3);
b[j10 + 1] = wr2 * (v1 + v3) + wi2 * (u1 + u3);
b[j14] = wr6 * (u1 - u3) - wi6 * (v1 - v3);
b[j14 + 1] = wr6 * (v1 - v3) + wi6 * (u1 - u3); */
_mm_store_pd(&b[j8], _mm_add_pd(u0, u2));
_mm_store_pd(&b[j12], ZMUL(w4, _mm_sub_pd(u0, u2)));
_mm_store_pd(&b[j10], ZMUL(w2, _mm_add_pd(u1, u3)));
_mm_store_pd(&b[j14], ZMUL(w6, _mm_sub_pd(u1, u3)));
/* u0 = x1 + c81 * (x5 - x7);
v0 = y1 + c81 * (y5 - y7);
u1 = x1 - c81 * (x5 - x7);
v1 = y1 - c81 * (y5 - y7);
u2 = x3 + c81 * (y5 + y7);
v2 = y3 - c81 * (x5 + x7);
u3 = x3 - c81 * (y5 + y7);
v3 = y3 + c81 * (x5 + x7); */
u1 = _mm_mul_pd(c81, _mm_sub_pd(t5, t7));
u0 = _mm_add_pd(t1, u1);
u1 = _mm_sub_pd(t1, u1);
u3 = _mm_xor_pd(_mm_mul_pd(c81, _mm_add_pd(t5, t7)), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
u2 = _mm_add_pd(t3, u3);
u3 = _mm_sub_pd(t3, u3);
/* b[j9] = wr1 * (u0 + u2) - wi1 * (v0 + v2);
b[j9 + 1] = wr1 * (v0 + v2) + wi1 * (u0 + u2);
b[j13] = wr5 * (u1 + u3) - wi5 * (v1 + v3);
b[j13 + 1] = wr5 * (v1 + v3) + wi5 * (u1 + u3);
b[j11] = wr3 * (u1 - u3) - wi3 * (v1 - v3);
b[j11 + 1] = wr3 * (v1 - v3) + wi3 * (u1 - u3);
b[j15] = wr7 * (u0 - u2) - wi7 * (v0 - v2);
b[j15 + 1] = wr7 * (v0 - v2) + wi7 * (u0 - u2); */
_mm_store_pd(&b[j9], ZMUL(w1, _mm_add_pd(u0, u2)));
_mm_store_pd(&b[j13], ZMUL(w5, _mm_add_pd(u1, u3)));
_mm_store_pd(&b[j11], ZMUL(w3, _mm_sub_pd(u1, u3)));
_mm_store_pd(&b[j15], ZMUL(w7, _mm_sub_pd(u0, u2)));
}
return 0;
}
int fft4b_(double *a, double *b, double *w, int *m, int *l)
{
int i, i0, i1, i2, i3, i4, i5, i6, i7, j, j0;
/* double x0, y0, x1, y1, x2, y2, x3, y3, wi1, wi2, wi3, wr1, wr2, wr3; */
__m128d t0, t1, t2, t3, t4, w1, w2, w3;
for (i = 0; i < *m; i++) {
i0 = i << 1;
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = i << 1;
i5 = i4 + (*m << 1);
i6 = i5 + (*m << 1);
i7 = i6 + (*m << 1);
/* x0 = a[i0] + a[i2];
y0 = a[i0 + 1] + a[i2 + 1];
x1 = a[i0] - a[i2];
y1 = a[i0 + 1] - a[i2 + 1];
x2 = a[i1] + a[i3];
y2 = a[i1 + 1] + a[i3 + 1];
x3 = a[i1 + 1] - a[i3 + 1];
y3 = a[i3] - a[i1]; */
t0 = _mm_load_pd(&a[i0]);
t2 = _mm_load_pd(&a[i2]);
t1 = _mm_sub_pd(t0, t2);
t0 = _mm_add_pd(t0, t2);
t3 = _mm_load_pd(&a[i1]);
t4 = _mm_load_pd(&a[i3]);
t2 = _mm_add_pd(t3, t4);
t3 = _mm_xor_pd(_mm_sub_pd(t3, t4), _mm_set_sd(-0.0));
t3 = _mm_shuffle_pd(t3, t3, 1);
/* b[i4] = x0 + x2;
b[i4 + 1] = y0 + y2;
b[i6] = x0 - x2;
b[i6 + 1] = y0 - y2;
b[i5] = x1 + x3;
b[i5 + 1] = y1 + y3;
b[i7] = x1 - x3;
b[i7 + 1] = y1 - y3; */
_mm_store_pd(&b[i4], _mm_add_pd(t0, t2));
_mm_store_pd(&b[i6], _mm_sub_pd(t0, t2));
_mm_store_pd(&b[i5], _mm_add_pd(t1, t3));
_mm_store_pd(&b[i7], _mm_sub_pd(t1, t3));
}
for (j = 1; j < *l; j++) {
j0 = j << 1;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1;
wr3 = wr1 * wr2 - wi1 * wi2;
wi3 = wr1 * wi2 + wi1 * wr2; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
w3 = ZMUL(w1, w2);
for (i = 0; i < *m; i++) {
i0 = (i << 1) + (j * *m << 1);
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = (i << 1) + (j * *m << 3);
i5 = i4 + (*m << 1);
i6 = i5 + (*m << 1);
i7 = i6 + (*m << 1);
/* x0 = a[i0] + a[i2];
y0 = a[i0 + 1] + a[i2 + 1];
x1 = a[i0] - a[i2];
y1 = a[i0 + 1] - a[i2 + 1];
x2 = a[i1] + a[i3];
y2 = a[i1 + 1] + a[i3 + 1];
x3 = a[i1 + 1] - a[i3 + 1];
y3 = a[i3] - a[i1]; */
t0 = _mm_load_pd(&a[i0]);
t2 = _mm_load_pd(&a[i2]);
t1 = _mm_sub_pd(t0, t2);
t0 = _mm_add_pd(t0, t2);
t3 = _mm_load_pd(&a[i1]);
t4 = _mm_load_pd(&a[i3]);
t2 = _mm_add_pd(t3, t4);
t3 = _mm_xor_pd(_mm_sub_pd(t3, t4), _mm_set_sd(-0.0));
t3 = _mm_shuffle_pd(t3, t3, 1);
/* b[i4] = x0 + x2;
b[i4 + 1] = y0 + y2;
b[i6] = wr2 * (x0 - x2) - wi2 * (y0 - y2);
b[i6 + 1] = wr2 * (y0 - y2) + wi2 * (x0 - x2);
b[i5] = wr1 * (x1 + x3) - wi1 * (y1 + y3);
b[i5 + 1] = wr1 * (y1 + y3) + wi1 * (x1 + x3);
b[i7] = wr3 * (x1 - x3) - wi3 * (y1 - y3);
b[i7 + 1] = wr3 * (y1 - y3) + wi3 * (x1 - x3); */
_mm_store_pd(&b[i4], _mm_add_pd(t0, t2));
_mm_store_pd(&b[i6], ZMUL(w2, _mm_sub_pd(t0, t2)));
_mm_store_pd(&b[i5], ZMUL(w1, _mm_add_pd(t1, t3)));
_mm_store_pd(&b[i7], ZMUL(w3, _mm_sub_pd(t1, t3)));
}
}
return 0;
}
int fft5a_(double *a, double *b, double *w, int *l)
{
/* static double c51 = .95105651629515357;
static double c52 = .61803398874989485;
static double c53 = .55901699437494742;
static double c54 = .25; */
static __m128d c51, c52, c53, c54;
int j, j0, j1, j2, j3, j4, j5, j6, j7, j8, j9;
/* double x0, y0, x1, y1, x2, y2, x3, y3, x4, y4, x5, y5, x6, y6, x7, y7,
x8, y8, x9, y9, x10, y10, wi1, wi2, wi3, wi4, wr1, wr2, wr3, wr4; */
__m128d t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, w1, w2, w3, w4;
c51 = _mm_set1_pd(0.95105651629515357);
c52 = _mm_set1_pd(0.61803398874989485);
c53 = _mm_set1_pd(0.55901699437494742);
c54 = _mm_set1_pd(0.25);
for (j = 0; j < *l; j++) {
j0 = j << 1;
j1 = j0 + (*l << 1);
j2 = j1 + (*l << 1);
j3 = j2 + (*l << 1);
j4 = j3 + (*l << 1);
j5 = j * 10;
j6 = j5 + 2;
j7 = j6 + 2;
j8 = j7 + 2;
j9 = j8 + 2;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1;
wr3 = wr1 * wr2 - wi1 * wi2;
wi3 = wr1 * wi2 + wi1 * wr2;
wr4 = wr2 * wr2 - wi2 * wi2;
wi4 = wr2 * wi2 + wr2 * wi2; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
w3 = ZMUL(w1, w2);
w4 = ZMUL(w2, w2);
/* x0 = a[j1] + a[j4];
y0 = a[j1 + 1] + a[j4 + 1];
x1 = a[j2] + a[j3];
y1 = a[j2 + 1] + a[j3 + 1];
x2 = c51 * (a[j1] - a[j4]);
y2 = c51 * (a[j1 + 1] - a[j4 + 1]);
x3 = c51 * (a[j2] - a[j3]);
y3 = c51 * (a[j2 + 1] - a[j3 + 1]);
x4 = x0 + x1;
y4 = y0 + y1;
x5 = c53 * (x0 - x1);
y5 = c53 * (y0 - y1);
x6 = a[j0] - c54 * x4;
y6 = a[j0 + 1] - c54 * y4;
x7 = x6 + x5;
y7 = y6 + y5;
x8 = x6 - x5;
y8 = y6 - y5;
x9 = y2 + c52 * y3;
y9 = -x2 - c52 * x3;
x10 = c52 * y2 - y3;
y10 = x3 - c52 * x2; */
t1 = _mm_load_pd(&a[j1]);
t4 = _mm_load_pd(&a[j4]);
t0 = _mm_add_pd(t1, t4);
t2 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_load_pd(&a[j2]);
t4 = _mm_load_pd(&a[j3]);
t3 = _mm_mul_pd(c51, _mm_sub_pd(t1, t4));
t1 = _mm_add_pd(t1, t4);
t4 = _mm_add_pd(t0, t1);
t5 = _mm_mul_pd(c53, _mm_sub_pd(t0, t1));
t0 = _mm_load_pd(&a[j0]);
t6 = _mm_sub_pd(t0, _mm_mul_pd(c54, t4));
t7 = _mm_add_pd(t6, t5);
t8 = _mm_sub_pd(t6, t5);
t9 = _mm_xor_pd(_mm_add_pd(t2, _mm_mul_pd(c52, t3)), _mm_set_sd(-0.0));
t9 = _mm_shuffle_pd(t9, t9, 1);
t10 = _mm_sub_pd(t3, _mm_mul_pd(c52, t2));
t10 = _mm_xor_pd(_mm_shuffle_pd(t10, t10, 1), _mm_set_sd(-0.0));
/* b[j5] = a[j0] + x4;
b[j5 + 1] = a[j0 + 1] + y4;
b[j6] = wr1 * (x7 + x9) - wi1 * (y7 + y9);
b[j6 + 1] = wr1 * (y7 + y9) + wi1 * (x7 + x9);
b[j7] = wr2 * (x8 + x10) - wi2 * (y8 + y10);
b[j7 + 1] = wr2 * (y8 + y10) + wi2 * (x8 + x10);
b[j8] = wr3 * (x8 - x10) - wi3 * (y8 - y10);
b[j8 + 1] = wr3 * (y8 - y10) + wi3 * (x8 - x10);
b[j9] = wr4 * (x7 - x9) - wi4 * (y7 - y9);
b[j9 + 1] = wr4 * (y7 - y9) + wi4 * (x7 - x9); */
_mm_store_pd(&b[j5], _mm_add_pd(t0, t4));
_mm_store_pd(&b[j6], ZMUL(w1, _mm_add_pd(t7, t9)));
_mm_store_pd(&b[j7], ZMUL(w2, _mm_add_pd(t8, t10)));
_mm_store_pd(&b[j8], ZMUL(w3, _mm_sub_pd(t8, t10)));
_mm_store_pd(&b[j9], ZMUL(w4, _mm_sub_pd(t7, t9)));
}
return 0;
}
int fft3b_(double *a, double *b, double *w, int *m, int *l)
{
/* static double c31 = .86602540378443865;
static double c32 = .5; */
static __m128d c31, c32;
int i, i0, i1, i2, i3, i4, i5, j, j0;
/* double x0, y0, x1, y1, x2, y2, wi1, wi2, wr1, wr2; */
__m128d t0, t1, t2, t3, w1, w2;
c31 = _mm_set1_pd(0.86602540378443865);
c32 = _mm_set1_pd(0.5);
for (i = 0; i < *m; i++) {
i0 = i << 1;
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i << 1;
i4 = i3 + (*m << 1);
i5 = i4 + (*m << 1);
/* x0 = a[i1] + a[i2];
y0 = a[i1 + 1] + a[i2 + 1];
x1 = a[i0] - c32 * x0;
y1 = a[i0 + 1] - c32 * y0;
x2 = c31 * (a[i1 + 1] - a[i2 + 1]);
y2 = c31 * (a[i2] - a[i1]); */
t1 = _mm_load_pd(&a[i1]);
t2 = _mm_load_pd(&a[i2]);
t0 = _mm_add_pd(t1, t2);
t2 = _mm_xor_pd(_mm_sub_pd(t1, t2), _mm_set_sd(-0.0));
t2 = _mm_mul_pd(c31, _mm_shuffle_pd(t2, t2, 1));
t3 = _mm_load_pd(&a[i0]);
t1 = _mm_sub_pd(t3, _mm_mul_pd(c32, t0));
/* b[i3] = a[i0] + x0;
b[i3 + 1] = a[i0 + 1] + y0;
b[i4] = x1 + x2;
b[i4 + 1] = y1 + y2;
b[i5] = x1 - x2;
b[i5 + 1] = y1 - y2; */
_mm_store_pd(&b[i3], _mm_add_pd(t3, t0));
_mm_store_pd(&b[i4], _mm_add_pd(t1, t2));
_mm_store_pd(&b[i5], _mm_sub_pd(t1, t2));
}
for (j = 1; j < *l; j++) {
j0 = j << 1;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
for (i = 0; i < *m; i++) {
i0 = (i << 1) + (j * *m << 1);
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = (i << 1) + (j * *m * 6);
i4 = i3 + (*m << 1);
i5 = i4 + (*m << 1);
/* x0 = a[i1] + a[i2];
y0 = a[i1 + 1] + a[i2 + 1];
x1 = a[i0] - x0 * .5;
y1 = a[i0 + 1] - y0 * .5;
x2 = c31 * (a[i1 + 1] - a[i2 + 1]);
y2 = c31 * (a[i2] - a[i1]); */
t1 = _mm_load_pd(&a[i1]);
t2 = _mm_load_pd(&a[i2]);
t0 = _mm_add_pd(t1, t2);
t2 = _mm_xor_pd(_mm_sub_pd(t1, t2), _mm_set_sd(-0.0));
t2 = _mm_mul_pd(c31, _mm_shuffle_pd(t2, t2, 1));
t3 = _mm_load_pd(&a[i0]);
t1 = _mm_sub_pd(t3, _mm_mul_pd(c32, t0));
/* b[i3] = a[i0] + x0;
b[i3 + 1] = a[i0 + 1] + y0;
b[i4] = wr1 * (x1 + x2) - wi1 * (y1 + y2);
b[i4 + 1] = wr1 * (y1 + y2) + wi1 * (x1 + x2);
b[i5] = wr2 * (x1 - x2) - wi2 * (y1 - y2);
b[i5 + 1] = wr2 * (y1 - y2) + wi2 * (x1 - x2); */
_mm_store_pd(&b[i3], _mm_add_pd(t3, t0));
_mm_store_pd(&b[i4], ZMUL(w1, _mm_add_pd(t1, t2)));
_mm_store_pd(&b[i5], ZMUL(w2, _mm_sub_pd(t1, t2)));
}
}
return 0;
}
// =============================================================================
//
// sse2_vChirpData
// version by: Alex Kan - SSE2 mods (haddsum removal) BH
// http://tbp.berkeley.edu/~alexkan/seti/
//
int sse2_ChirpData_ak(
sah_complex * cx_DataArray,
sah_complex * cx_ChirpDataArray,
int chirp_rate_ind,
double chirp_rate,
int ul_NumDataPoints,
double sample_rate
) {
int i;
if (chirp_rate_ind == 0) {
memcpy(cx_ChirpDataArray, cx_DataArray, (int)ul_NumDataPoints * sizeof(sah_complex) );
return 0;
}
int vEnd;
double srate = chirp_rate * 0.5 / (sample_rate * sample_rate);
__m128d rate = _mm_set1_pd(chirp_rate * 0.5 / (sample_rate * sample_rate));
__m128d roundVal = _mm_set1_pd(srate >= 0.0 ? TWO_TO_52 : -TWO_TO_52);
// main vectorised loop
vEnd = ul_NumDataPoints - (ul_NumDataPoints & 3);
for (i = 0; i < vEnd; i += 4) {
const float *data = (const float *) (cx_DataArray + i);
float *chirped = (float *) (cx_ChirpDataArray + i);
__m128d di = _mm_set1_pd(i);
__m128d a1 = _mm_add_pd(_mm_set_pd(1.0, 0.0), di);
__m128d a2 = _mm_add_pd(_mm_set_pd(3.0, 2.0), di);
__m128d x1, y1;
__m128 d1, d2;
__m128 cd1, cd2;
__m128 td1, td2;
__m128 x;
__m128 y;
__m128 s;
__m128 c;
__m128 m;
// load the signal to be chirped
prefetchnta((const void *)( data+32 ));
d1 = _mm_load_ps(data);
d2 = _mm_load_ps(data+4);
// calculate the input angle
a1 = _mm_mul_pd(a1, a1);
a2 = _mm_mul_pd(a2, a2);
a1 = _mm_mul_pd(a1, rate);
a2 = _mm_mul_pd(a2, rate);
// reduce the angle to the range (-0.5, 0.5)
x1 = _mm_add_pd(a1, roundVal);
y1 = _mm_add_pd(a2, roundVal);
x1 = _mm_sub_pd(x1, roundVal);
y1 = _mm_sub_pd(y1, roundVal);
a1 = _mm_sub_pd(a1, x1);
a2 = _mm_sub_pd(a2, y1);
// convert pair of packed double into packed single
x = _mm_movelh_ps(_mm_cvtpd_ps(a1), _mm_cvtpd_ps(a2));
// square to the range [0, 0.25)
y = _mm_mul_ps(x, x);
// perform the initial polynomial approximations
s = _mm_mul_ps(y, SS4);
c = _mm_mul_ps(y, CC3);
s = _mm_add_ps(s, SS3);
c = _mm_add_ps(c, CC2);
s = _mm_mul_ps(s, y);
c = _mm_mul_ps(c, y);
s = _mm_add_ps(s, SS2);
c = _mm_add_ps(c, CC1);
s = _mm_mul_ps(s, y);
c = _mm_mul_ps(c, y);
s = _mm_add_ps(s, SS1);
s = _mm_mul_ps(s, x);
c = _mm_add_ps(c, ONE);
// perform first angle doubling
x = _mm_sub_ps(_mm_mul_ps(c, c), _mm_mul_ps(s, s));
y = _mm_mul_ps(_mm_mul_ps(s, c), TWO);
// calculate scaling factor to correct the magnitude
// m1 = vec_nmsub(y1, y1, vec_nmsub(x1, x1, TWO));
// m2 = vec_nmsub(y2, y2, vec_nmsub(x2, x2, TWO));
m = vec_recip2(_mm_add_ps(_mm_mul_ps(x, x), _mm_mul_ps(y, y)));
// perform second angle doubling
c = _mm_sub_ps(_mm_mul_ps(x, x), _mm_mul_ps(y, y));
s = _mm_mul_ps(_mm_mul_ps(y, x), TWO);
// correct the magnitude (final sine / cosine approximations)
c = _mm_mul_ps(c, m);
s = _mm_mul_ps(s, m);
/* c1 c2 c3 c4
s1 s2 s3 s4
R1 i1 R2 I2 R3 i3 R4 i4
R1 * c1 + (i1 * s1 * -1)
i1 * c1 + R1 * s1
R2 * c2 + (i2 * s2 * -1)
i2 * c2 + R2 * s2
*/
x = d1;
y = d2;
x = _mm_shuffle_ps(x, x, 0xB1);
y = _mm_shuffle_ps(y, y, 0xB1);
x = _mm_mul_ps(x, R_NEG);
y = _mm_mul_ps(y, R_NEG);
cd1 = _mm_shuffle_ps(c, c, 0x50); // 01 01 00 00 AaBb => BBbb => c3c3c4c4
cd2 = _mm_shuffle_ps(c, c, 0xfa); // 11 11 10 10 AaBb => AAaa => c1c1c2c2
td1 = _mm_shuffle_ps(s, s, 0x50);
td2 = _mm_shuffle_ps(s, s, 0xfa);
cd1 = _mm_mul_ps(cd1, d1);
cd2 = _mm_mul_ps(cd2, d2);
td1 = _mm_mul_ps(td1, x);
td2 = _mm_mul_ps(td2, y);
cd1 = _mm_add_ps(cd1, td1);
cd2 = _mm_add_ps(cd2, td2);
// store chirped values
_mm_stream_ps(chirped+0, cd1);
_mm_stream_ps(chirped+4, cd2);
}
_mm_sfence();
if( i < ul_NumDataPoints) {
// use original routine to finish up any tailings (max stride-1 elements)
v_ChirpData(cx_DataArray+i, cx_ChirpDataArray+i
, chirp_rate_ind, chirp_rate, ul_NumDataPoints-i, sample_rate);
}
analysis_state.FLOP_counter+=12.0*ul_NumDataPoints;
return 0;
}
void CalcGravity(int sp, PSpot* allSpot,int* length)
{
__m128d force1 = _mm_set1_pd(0);
__m128d force2 = _mm_set1_pd(0);
PSpot* spotSp = &allSpot[sp];
for(int i=0;i<sp;i++)
{
__m128d diff1 = _mm_sub_pd(allSpot[i].pos1, spotSp->pos1);
__m128d diff2 = _mm_sub_sd(allSpot[i].pos2, spotSp->pos2);
__m128d r = Length(diff1, diff2);
if (r.m128d_f64[0]*2 < (spotSp->qmass + allSpot[i].qmass))
{
if (allSpot[i].mass > spotSp->mass)
{
allSpot[i].heading1 =
_mm_add_pd(
allSpot[i].heading1,
_mm_mul_pd(
_mm_sub_pd(spotSp->heading1, allSpot[i].heading1),
_mm_set1_pd(spotSp->mass / (spotSp->mass + allSpot[i].mass))
)
);
allSpot[i].heading2 =
_mm_add_sd(
allSpot[i].heading2,
_mm_mul_sd(
_mm_sub_sd(spotSp->heading2, allSpot[i].heading2),
_mm_set1_pd(spotSp->mass / (spotSp->mass + allSpot[i].mass))
)
);
allSpot[i].mass += spotSp->mass;
allSpot[i].qmass = pow(allSpot[i].mass, 0.33333);
spotSp->mass = 0;
(*length)--;
PSpot temp;
temp = allSpot[sp];
allSpot[sp] = allSpot[*length];
allSpot[*length] = temp;
return;
}
else
{
spotSp->heading1 =
_mm_add_pd(
spotSp->heading1,
_mm_mul_pd(
_mm_sub_pd(allSpot[i].heading1, spotSp->heading1),
_mm_set1_pd(allSpot[i].mass / (spotSp->mass + allSpot[i].mass))
)
);
spotSp->heading2 =
_mm_add_sd(
spotSp->heading2,
_mm_mul_sd(
_mm_sub_sd(allSpot[i].heading2, spotSp->heading2),
_mm_set1_pd(allSpot[i].mass / (spotSp->mass + allSpot[i].mass))
)
);
spotSp->mass += allSpot[i].mass;
spotSp->qmass = pow(spotSp->mass, 0.33333);
allSpot[i].mass = 0;
(*length)--;
PSpot temp;
temp = allSpot[i];
allSpot[i] = allSpot[*length];
allSpot[*length] = temp;
return;
}
}
//float f = (G * spotSp->mass * allSpot[i].mass) / (r.m128d_f64[0] * r.m128d_f64[0] * r.m128d_f64[0]);
__m128d r1 = r;
r1.m128d_f64[1] = G;
__m128d r2 = r;
r2.m128d_f64[1] = spotSp->mass;
__m128d r3 = r;
r3.m128d_f64[1] = allSpot[i].mass;
__m128d r4 = _mm_mul_pd(_mm_mul_pd(r1, r2), r3);
__m128d r5 = _mm_shuffle_pd(r4, r4, 3);
r4 = _mm_shuffle_pd(r4, r4, 0);
__m128d r6 = _mm_div_pd(r5, r4);
force1 = _mm_add_pd(force1,_mm_mul_pd(diff1, r6));
force2 = _mm_add_sd(force2,_mm_mul_sd(diff2, r6));
}
for(int i=sp+1;i<*length;i++)
{
__m128d diff1 = _mm_sub_pd(allSpot[i].pos1, spotSp->pos1);
__m128d diff2 = _mm_sub_sd(allSpot[i].pos2, spotSp->pos2);
__m128d r = Length(diff1, diff2);
if (r.m128d_f64[0]*2 < (spotSp->qmass + allSpot[i].qmass))
{
if (allSpot[i].mass > spotSp->mass)
{
allSpot[i].heading1 =
_mm_add_pd(
allSpot[i].heading1,
_mm_mul_pd(
_mm_sub_pd(spotSp->heading1, allSpot[i].heading1),
_mm_set1_pd(spotSp->mass / (spotSp->mass + allSpot[i].mass))
)
);
allSpot[i].heading2 =
_mm_add_sd(
allSpot[i].heading2,
_mm_mul_sd(
_mm_sub_sd(spotSp->heading2, allSpot[i].heading2),
_mm_set1_pd(spotSp->mass / (spotSp->mass + allSpot[i].mass))
)
);
allSpot[i].mass += spotSp->mass;
allSpot[i].qmass = pow(allSpot[i].mass, 0.33333);
spotSp->mass = 0;
(*length)--;
PSpot temp;
temp = allSpot[sp];
allSpot[sp] = allSpot[*length];
allSpot[*length] = temp;
return;
}
else
{
spotSp->heading1 =
_mm_add_pd(
spotSp->heading1,
_mm_mul_pd(
_mm_sub_pd(allSpot[i].heading1, spotSp->heading1),
_mm_set1_pd(allSpot[i].mass / (spotSp->mass + allSpot[i].mass))
)
);
spotSp->heading2 =
_mm_add_sd(
spotSp->heading2,
_mm_mul_sd(
_mm_sub_sd(allSpot[i].heading2, spotSp->heading2),
_mm_set1_pd(allSpot[i].mass / (spotSp->mass + allSpot[i].mass))
)
);
spotSp->mass += allSpot[i].mass;
spotSp->qmass = pow(spotSp->mass, 0.33333);
allSpot[i].mass = 0;
(*length)--;
PSpot temp;
temp = allSpot[i];
allSpot[i] = allSpot[*length];
allSpot[*length] = temp;
return;
}
}
//float f = (G * spotSp->mass * allSpot[i].mass) / (r.m128d_f64[0] * r.m128d_f64[0] * r.m128d_f64[0]);
__m128d r1 = r;
r1.m128d_f64[1] = G;
__m128d r2 = r;
r2.m128d_f64[1] = spotSp->mass;
__m128d r3 = r;
r3.m128d_f64[1] = allSpot[i].mass;
__m128d r4 = _mm_mul_pd(_mm_mul_pd(r1, r2), r3);
__m128d r5 = _mm_shuffle_pd(r4, r4, 3);
r4 = _mm_shuffle_pd(r4, r4, 0);
__m128d r6 = _mm_div_pd(r5, r4);
force1 = _mm_add_pd(force1,_mm_mul_pd(diff1, r6));
force2 = _mm_add_sd(force2,_mm_mul_sd(diff2, r6));
}
force1 = _mm_div_pd(force1, _mm_set1_pd(spotSp->mass));
force2 = _mm_div_sd(force2, _mm_set1_pd(spotSp->mass));
__m128d forcef = Length(force1, force2);
if (forcef.m128d_f64[0] > 0)
{
double gate = 0.001f;
double step = gate / forcef.m128d_f64[0];
if (spotSp->process + step < 1)
{
spotSp->process += step;
}
else
{
step = 1 - spotSp->process;
spotSp->process = 1;
}
__m128d stepd = _mm_set1_pd(step);
spotSp->heading1 = _mm_add_pd(spotSp->heading1,_mm_mul_pd(force1,stepd));
spotSp->heading2 = _mm_add_sd(spotSp->heading2,_mm_mul_sd(force2,stepd));
spotSp->pos1 = _mm_add_pd(spotSp->pos1, _mm_mul_pd(spotSp->heading1,stepd));
spotSp->pos2 = _mm_add_sd(spotSp->pos2, _mm_mul_sd(spotSp->heading2,stepd));
}
else
{
spotSp->pos1 = _mm_add_pd(spotSp->pos1, spotSp->heading1);
spotSp->pos2 = _mm_add_sd(spotSp->pos2, spotSp->heading2);
spotSp->process = 1;
}
}
void exchlaplacecoeffData_6(unsigned int slot) {
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((!neighbor_isValid[1][0])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S902, S905, S908, S907, S901, S910, S904, S903, S906, S909, S900 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+4558)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+4626)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+4558)] = 0.000000e+00;
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(6.400000e+01);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<63); i1 += 4) {
/* yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<66); i1 += 1) {
yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+9114)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+9182)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+9114)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+2)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+70)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+2)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+31894)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+31962)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+31894)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+36450)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+36518)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+36450)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posBegin[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<63); i1 += 4) {
/* xPos = posBegin[0]; */
__m128d vec0 = _mm_load1_pd((&posBegin[0]));
__m128d vec0_2 = _mm_load1_pd((&posBegin[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<66); i1 += 1) {
xPos = posBegin[0];
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+18226)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+18294)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+18226)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+13670)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+13738)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+13670)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+27338)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+27406)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+27338)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+22782)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+22850)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+22782)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[1][1])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S920, S914, S917, S911, S913, S916, S919, S921, S918, S912, S915 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+4622)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+4690)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+4622)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+31958)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+32026)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+31958)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+13734)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+13802)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+13734)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+66)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+134)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+66)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+22846)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+22914)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+22846)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(6.400000e+01);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<63); i1 += 4) {
/* yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<66); i1 += 1) {
yPos = ((((i1-1)/6.400000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+18290)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+18358)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+18290)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+27402)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+27470)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+27402)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+36514)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+36582)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+36514)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i1 = 1;
for (; (i1<=64); i1 += 2) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+9178)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[((i1*68)+9246)] = 0.000000e+00;
}
for (; (i1<=65); i1 += 1) {
fieldData_LaplaceCoeff_6_p1[((i1*68)+9178)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posEnd[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<63); i1 += 4) {
/* xPos = posEnd[0]; */
__m128d vec0 = _mm_load1_pd((&posEnd[0]));
__m128d vec0_2 = _mm_load1_pd((&posEnd[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<66); i1 += 1) {
xPos = posEnd[0];
}
}
}
}
}
if ((!neighbor_isValid[1][2])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S929, S923, S926, S931, S925, S928, S922, S930, S924, S927, S932 */
{
{
{
{
{
{
{
{
{
{
{
int i2 = 2;
for (; (i2<=65); i2 += 2) {
xPos = ((((i2-2)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=66); i2 += 1) {
xPos = ((((i2-2)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+18292)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+18293)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+18292)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+36516)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+36517)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+36516)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+31960)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+31961)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+31960)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+68)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+69)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+68)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+9180)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+9181)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+9180)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+22848)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+22849)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+22848)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+27404)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+27405)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+27404)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+4624)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+4625)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+4624)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+13736)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+13737)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+13736)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=65); i2 += 2) {
yPos = posBegin[1];
yPos = posBegin[1];
}
for (; (i2<=66); i2 += 1) {
yPos = posBegin[1];
}
}
}
}
}
if ((!neighbor_isValid[1][3])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S941, S935, S938, S943, S940, S934, S937, S942, S936, S939, S933 */
{
{
{
{
{
{
{
{
{
{
{
int i2 = 2;
for (; (i2<=65); i2 += 2) {
xPos = ((((i2-2)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=66); i2 += 1) {
xPos = ((((i2-2)/6.400000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+36312)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+36313)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+36312)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+22644)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+22645)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+22644)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+13532)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+13533)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+13532)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+40868)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+40869)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+40868)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+18088)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+18089)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+18088)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+4420)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+4421)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+4420)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+31756)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+31757)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+31756)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+27200)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+27201)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+27200)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=65); i2 += 2) {
yPos = posEnd[1];
yPos = posEnd[1];
}
for (; (i2<=66); i2 += 1) {
yPos = posEnd[1];
}
}
}
{
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][0]);
int i2 = 2;
for (; (i2<=65); i2 += 2) {
fieldData_LaplaceCoeff_6_p1[(i2+8976)] = 0.000000e+00;
fieldData_LaplaceCoeff_6_p1[(i2+8977)] = 0.000000e+00;
}
for (; (i2<=66); i2 += 1) {
fieldData_LaplaceCoeff_6_p1[(i2+8976)] = 0.000000e+00;
}
}
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
/* Statements in this Scop: S944 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
double* buffer_Send_1_p1 = (&buffer_Send[1][(i3*65)]);
int i4 = 1;
for (; (i4<=64); i4 += 2) {
buffer_Send_1_p1[(i4-1)] = fieldData_LaplaceCoeff_6_p1[((i4*68)+66)];
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_6_p1[((i4*68)+134)];
}
for (; (i4<=65); i4 += 1) {
buffer_Send_1_p1[(i4-1)] = fieldData_LaplaceCoeff_6_p1[((i4*68)+66)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Isend(buffer_Send[1], 585, MPI_DOUBLE, neighbor_remoteRank[1][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Irecv(buffer_Recv[0], 585, MPI_DOUBLE, neighbor_remoteRank[1][0], ((unsigned int)(neighbor_fragCommId[1][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
/* Statements in this Scop: S945 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i3*65)]);
int i4 = 3;
for (; (i4<=66); i4 += 2) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-134)] = buffer_Recv_0_p1[(i4-3)];
fieldData_LaplaceCoeff_6_p1[((i4*68)-66)] = buffer_Recv_0_p1[(i4-2)];
}
for (; (i4<=67); i4 += 1) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-134)] = buffer_Recv_0_p1[(i4-3)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Isend(&fieldData_LaplaceCoeff[6][4422], 1, mpiDatatype_9_65_4556, neighbor_remoteRank[1][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff[6][70], 1, mpiDatatype_9_65_4556, neighbor_remoteRank[1][2], ((unsigned int)(neighbor_fragCommId[1][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
/* Statements in this Scop: S946 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
double* buffer_Send_0_p1 = (&buffer_Send[0][(i3*67)]);
int i4 = 0;
for (; (i4<=65); i4 += 2) {
buffer_Send_0_p1[i4] = fieldData_LaplaceCoeff_6_p1[((i4*68)+3)];
buffer_Send_0_p1[(i4+1)] = fieldData_LaplaceCoeff_6_p1[((i4*68)+71)];
}
for (; (i4<=66); i4 += 1) {
buffer_Send_0_p1[i4] = fieldData_LaplaceCoeff_6_p1[((i4*68)+3)];
}
}
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
/* Statements in this Scop: S947 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
double* buffer_Send_1_p1 = (&buffer_Send[1][(i3*67)]);
int i4 = 0;
for (; (i4<=65); i4 += 2) {
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_6_p1[((i4*68)+65)];
buffer_Send_1_p1[(i4+1)] = fieldData_LaplaceCoeff_6_p1[((i4*68)+133)];
}
for (; (i4<=66); i4 += 1) {
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_6_p1[((i4*68)+65)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Isend(buffer_Send[0], 603, MPI_DOUBLE, neighbor_remoteRank[1][0], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][0]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[0]);
reqOutstanding_Send[0] = true;
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Isend(buffer_Send[1], 603, MPI_DOUBLE, neighbor_remoteRank[1][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Irecv(buffer_Recv[0], 603, MPI_DOUBLE, neighbor_remoteRank[1][0], ((unsigned int)(neighbor_fragCommId[1][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Irecv(buffer_Recv[1], 603, MPI_DOUBLE, neighbor_remoteRank[1][1], ((unsigned int)(neighbor_fragCommId[1][1]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
if (reqOutstanding_Recv[1]) {
waitForMPIReq(&mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
/* Statements in this Scop: S948 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i3*67)]);
int i4 = 1;
for (; (i4<=66); i4 += 2) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-67)] = buffer_Recv_0_p1[(i4-1)];
fieldData_LaplaceCoeff_6_p1[((i4*68)+1)] = buffer_Recv_0_p1[i4];
}
for (; (i4<=67); i4 += 1) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-67)] = buffer_Recv_0_p1[(i4-1)];
}
}
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
/* Statements in this Scop: S949 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* buffer_Recv_1_p1 = (&buffer_Recv[1][(i3*67)]);
double* fieldData_LaplaceCoeff_6_p1 = (&fieldData_LaplaceCoeff[6][(i3*4556)]);
int i4 = 67;
for (; (i4<=132); i4 += 2) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-4489)] = buffer_Recv_1_p1[(i4-67)];
fieldData_LaplaceCoeff_6_p1[((i4*68)-4421)] = buffer_Recv_1_p1[(i4-66)];
}
for (; (i4<=133); i4 += 1) {
fieldData_LaplaceCoeff_6_p1[((i4*68)-4489)] = buffer_Recv_1_p1[(i4-67)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[0]) {
waitForMPIReq(&mpiRequest_Send[0]);
reqOutstanding_Send[0] = false;
}
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Isend(&fieldData_LaplaceCoeff[6][137], 1, mpiDatatype_9_67_4556, neighbor_remoteRank[1][2], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][2]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[2]);
reqOutstanding_Send[2] = true;
}
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Isend(&fieldData_LaplaceCoeff[6][4353], 1, mpiDatatype_9_67_4556, neighbor_remoteRank[1][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff[6][1], 1, mpiDatatype_9_67_4556, neighbor_remoteRank[1][2], ((unsigned int)(neighbor_fragCommId[1][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Irecv(&fieldData_LaplaceCoeff[6][4489], 1, mpiDatatype_9_67_4556, neighbor_remoteRank[1][3], ((unsigned int)(neighbor_fragCommId[1][3]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
if (reqOutstanding_Recv[3]) {
waitForMPIReq(&mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[2]) {
waitForMPIReq(&mpiRequest_Send[2]);
reqOutstanding_Send[2] = false;
}
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
}
int fft8b_(double *a, double *b, double *w, int *m, int *l)
{
/* static double c81 = .70710678118654752; */
static __m128d c81;
int i, i0, i1, i2, i3, i4, i5, i6, i7, i8, i9, i10, i11, i12, i13, i14, i15, j, j0;
/* double u0, v0, u1, x0, y0, x1, y1, x2, y2, x3, y3, v1, x4, y4, x5, y5,
x6, y6, x7, y7, u2, v2, u3, v3, wi1, wi2, wi3, wi4, wi5, wi6,
wi7, wr1, wr2, wr3, wr4, wr5, wr6, wr7; */
__m128d t0, t1, t2, t3, t4, t5, t6, t7, t8, u0, u1, u2, u3, w1, w2, w3, w4, w5, w6, w7;
c81 = _mm_set1_pd(0.70710678118654752);
for (i = 0; i < *m; i++) {
i0 = i << 1;
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = i3 + (*m * *l << 1);
i5 = i4 + (*m * *l << 1);
i6 = i5 + (*m * *l << 1);
i7 = i6 + (*m * *l << 1);
i8 = i << 1;
i9 = i8 + (*m << 1);
i10 = i9 + (*m << 1);
i11 = i10 + (*m << 1);
i12 = i11 + (*m << 1);
i13 = i12 + (*m << 1);
i14 = i13 + (*m << 1);
i15 = i14 + (*m << 1);
/* x0 = a[i0] + a[i4];
y0 = a[i0 + 1] + a[i4 + 1];
x1 = a[i0] - a[i4];
y1 = a[i0 + 1] - a[i4 + 1];
x2 = a[i2] + a[i6];
y2 = a[i2 + 1] + a[i6 + 1];
x3 = a[i2 + 1] - a[i6 + 1];
y3 = a[i6] - a[i2]; */
t0 = _mm_load_pd(&a[i0]);
t2 = _mm_load_pd(&a[i4]);
t1 = _mm_sub_pd(t0, t2);
t0 = _mm_add_pd(t0, t2);
t3 = _mm_load_pd(&a[i2]);
t4 = _mm_load_pd(&a[i6]);
t2 = _mm_add_pd(t3, t4);
t3 = _mm_xor_pd(_mm_sub_pd(t3, t4), _mm_set_sd(-0.0));
t3 = _mm_shuffle_pd(t3, t3, 1);
/* u0 = x0 + x2;
v0 = y0 + y2;
u1 = x0 - x2;
v1 = y0 - y2; */
u0 = _mm_add_pd(t0, t2);
u1 = _mm_sub_pd(t0, t2);
/* x4 = a[i1] + a[i5];
y4 = a[i1 + 1] + a[i5 + 1];
x5 = a[i1] - a[i5];
y5 = a[i1 + 1] - a[i5 + 1];
x6 = a[i3] + a[i7];
y6 = a[i3 + 1] + a[i7 + 1];
x7 = a[i3] - a[i7];
y7 = a[i3 + 1] - a[i7 + 1]; */
t4 = _mm_load_pd(&a[i1]);
t6 = _mm_load_pd(&a[i5]);
t5 = _mm_sub_pd(t4, t6);
t4 = _mm_add_pd(t4, t6);
t7 = _mm_load_pd(&a[i3]);
t8 = _mm_load_pd(&a[i7]);
t6 = _mm_add_pd(t7, t8);
t7 = _mm_sub_pd(t7, t8);
/* u2 = x4 + x6;
v2 = y4 + y6;
u3 = y4 - y6;
v3 = x6 - x4; */
u2 = _mm_add_pd(t4, t6);
u3 = _mm_xor_pd(_mm_sub_pd(t4, t6), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
/* b[i8] = u0 + u2;
b[i8 + 1] = v0 + v2;
b[i12] = u0 - u2;
b[i12 + 1] = v0 - v2;
b[i10] = u1 + u3;
b[i10 + 1] = v1 + v3;
b[i14] = u1 - u3;
b[i14 + 1] = v1 - v3; */
_mm_store_pd(&b[i8], _mm_add_pd(u0, u2));
_mm_store_pd(&b[i12], _mm_sub_pd(u0, u2));
_mm_store_pd(&b[i10], _mm_add_pd(u1, u3));
_mm_store_pd(&b[i14], _mm_sub_pd(u1, u3));
/* u0 = x1 + c81 * (x5 - x7);
v0 = y1 + c81 * (y5 - y7);
u1 = x1 - c81 * (x5 - x7);
v1 = y1 - c81 * (y5 - y7);
u2 = x3 + c81 * (y5 + y7);
v2 = y3 - c81 * (x5 + x7);
u3 = x3 - c81 * (y5 + y7);
v3 = y3 + c81 * (x5 + x7); */
u1 = _mm_mul_pd(c81, _mm_sub_pd(t5, t7));
u0 = _mm_add_pd(t1, u1);
u1 = _mm_sub_pd(t1, u1);
u3 = _mm_xor_pd(_mm_mul_pd(c81, _mm_add_pd(t5, t7)), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
u2 = _mm_add_pd(t3, u3);
u3 = _mm_sub_pd(t3, u3);
/* b[i9] = u0 + u2;
b[i9 + 1] = v0 + v2;
b[i13] = u1 + u3;
b[i13 + 1] = v1 + v3;
b[i11] = u1 - u3;
b[i11 + 1] = v1 - v3;
b[i15] = u0 - u2;
b[i15 + 1] = v0 - v2; */
_mm_store_pd(&b[i9], _mm_add_pd(u0, u2));
_mm_store_pd(&b[i13], _mm_add_pd(u1, u3));
_mm_store_pd(&b[i11], _mm_sub_pd(u1, u3));
_mm_store_pd(&b[i15], _mm_sub_pd(u0, u2));
}
for (j = 1; j < *l; j++) {
j0 = j << 1;
/* wr1 = w[j0];
wi1 = w[j0 + 1];
wr2 = wr1 * wr1 - wi1 * wi1;
wi2 = wr1 * wi1 + wr1 * wi1;
wr3 = wr1 * wr2 - wi1 * wi2;
wi3 = wr1 * wi2 + wi1 * wr2;
wr4 = wr2 * wr2 - wi2 * wi2;
wi4 = wr2 * wi2 + wr2 * wi2;
wr5 = wr2 * wr3 - wi2 * wi3;
wi5 = wr2 * wi3 + wi2 * wr3;
wr6 = wr3 * wr3 - wi3 * wi3;
wi6 = wr3 * wi3 + wr3 * wi3;
wr7 = wr3 * wr4 - wi3 * wi4;
wi7 = wr3 * wi4 + wi3 * wr4; */
w1 = _mm_load_pd(&w[j0]);
w2 = ZMUL(w1, w1);
w3 = ZMUL(w1, w2);
w4 = ZMUL(w2, w2);
w5 = ZMUL(w2, w3);
w6 = ZMUL(w3, w3);
w7 = ZMUL(w3, w4);
for (i = 0; i < *m; i++) {
i0 = (i << 1) + (j * *m << 1);
i1 = i0 + (*m * *l << 1);
i2 = i1 + (*m * *l << 1);
i3 = i2 + (*m * *l << 1);
i4 = i3 + (*m * *l << 1);
i5 = i4 + (*m * *l << 1);
i6 = i5 + (*m * *l << 1);
i7 = i6 + (*m * *l << 1);
i8 = (i << 1) + (j * *m << 4);
i9 = i8 + (*m << 1);
i10 = i9 + (*m << 1);
i11 = i10 + (*m << 1);
i12 = i11 + (*m << 1);
i13 = i12 + (*m << 1);
i14 = i13 + (*m << 1);
i15 = i14 + (*m << 1);
/* x0 = a[i0] + a[i4];
y0 = a[i0 + 1] + a[i4 + 1];
x1 = a[i0] - a[i4];
y1 = a[i0 + 1] - a[i4 + 1];
x2 = a[i2] + a[i6];
y2 = a[i2 + 1] + a[i6 + 1];
x3 = a[i2 + 1] - a[i6 + 1];
y3 = a[i6] - a[i2]; */
t0 = _mm_load_pd(&a[i0]);
t2 = _mm_load_pd(&a[i4]);
t1 = _mm_sub_pd(t0, t2);
t0 = _mm_add_pd(t0, t2);
t3 = _mm_load_pd(&a[i2]);
t4 = _mm_load_pd(&a[i6]);
t2 = _mm_add_pd(t3, t4);
t3 = _mm_xor_pd(_mm_sub_pd(t3, t4), _mm_set_sd(-0.0));
t3 = _mm_shuffle_pd(t3, t3, 1);
/* u0 = x0 + x2;
v0 = y0 + y2;
u1 = x0 - x2;
v1 = y0 - y2; */
u0 = _mm_add_pd(t0, t2);
u1 = _mm_sub_pd(t0, t2);
/* x4 = a[i1] + a[i5];
y4 = a[i1 + 1] + a[i5 + 1];
x5 = a[i1] - a[i5];
y5 = a[i1 + 1] - a[i5 + 1];
x6 = a[i3] + a[i7];
y6 = a[i3 + 1] + a[i7 + 1];
x7 = a[i3] - a[i7];
y7 = a[i3 + 1] - a[i7 + 1]; */
t4 = _mm_load_pd(&a[i1]);
t6 = _mm_load_pd(&a[i5]);
t5 = _mm_sub_pd(t4, t6);
t4 = _mm_add_pd(t4, t6);
t7 = _mm_load_pd(&a[i3]);
t8 = _mm_load_pd(&a[i7]);
t6 = _mm_add_pd(t7, t8);
t7 = _mm_sub_pd(t7, t8);
/* u2 = x4 + x6;
v2 = y4 + y6;
u3 = y4 - y6;
v3 = x6 - x4; */
u2 = _mm_add_pd(t4, t6);
u3 = _mm_xor_pd(_mm_sub_pd(t4, t6), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
/* b[i8] = u0 + u2;
b[i8 + 1] = v0 + v2;
b[i12] = wr4 * (u0 - u2) - wi4 * (v0 - v2);
b[i12 + 1] = wr4 * (v0 - v2) + wi4 * (u0 - u2);
b[i10] = wr2 * (u1 + u3) - wi2 * (v1 + v3);
b[i10 + 1] = wr2 * (v1 + v3) + wi2 * (u1 + u3);
b[i14] = wr6 * (u1 - u3) - wi6 * (v1 - v3);
b[i14 + 1] = wr6 * (v1 - v3) + wi6 * (u1 - u3); */
_mm_store_pd(&b[i8], _mm_add_pd(u0, u2));
_mm_store_pd(&b[i12], ZMUL(w4, _mm_sub_pd(u0, u2)));
_mm_store_pd(&b[i10], ZMUL(w2, _mm_add_pd(u1, u3)));
_mm_store_pd(&b[i14], ZMUL(w6, _mm_sub_pd(u1, u3)));
/* u0 = x1 + c81 * (x5 - x7);
v0 = y1 + c81 * (y5 - y7);
u1 = x1 - c81 * (x5 - x7);
v1 = y1 - c81 * (y5 - y7);
u2 = x3 + c81 * (y5 + y7);
v2 = y3 - c81 * (x5 + x7);
u3 = x3 - c81 * (y5 + y7);
v3 = y3 + c81 * (x5 + x7); */
u1 = _mm_mul_pd(c81, _mm_sub_pd(t5, t7));
u0 = _mm_add_pd(t1, u1);
u1 = _mm_sub_pd(t1, u1);
u3 = _mm_xor_pd(_mm_mul_pd(c81, _mm_add_pd(t5, t7)), _mm_set_sd(-0.0));
u3 = _mm_shuffle_pd(u3, u3, 1);
u2 = _mm_add_pd(t3, u3);
u3 = _mm_sub_pd(t3, u3);
/* b[i9] = wr1 * (u0 + u2) - wi1 * (v0 + v2);
b[i9 + 1] = wr1 * (v0 + v2) + wi1 * (u0 + u2);
b[i13] = wr5 * (u1 + u3) - wi5 * (v1 + v3);
b[i13 + 1] = wr5 * (v1 + v3) + wi5 * (u1 + u3);
b[i11] = wr3 * (u1 - u3) - wi3 * (v1 - v3);
b[i11 + 1] = wr3 * (v1 - v3) + wi3 * (u1 - u3);
b[i15] = wr7 * (u0 - u2) - wi7 * (v0 - v2);
b[i15 + 1] = wr7 * (v0 - v2) + wi7 * (u0 - u2); */
_mm_store_pd(&b[i9], ZMUL(w1, _mm_add_pd(u0, u2)));
_mm_store_pd(&b[i13], ZMUL(w5, _mm_add_pd(u1, u3)));
_mm_store_pd(&b[i11], ZMUL(w3, _mm_sub_pd(u1, u3)));
_mm_store_pd(&b[i15], ZMUL(w7, _mm_sub_pd(u0, u2)));
}
}
return 0;
}
void nb_kernel430_ia32_sse2(int * p_nri,
int * iinr,
int * jindex,
int * jjnr,
int * shift,
double * shiftvec,
double * fshift,
int * gid,
double * pos,
double * faction,
double * charge,
double * p_facel,
double * p_krf,
double * p_crf,
double * vc,
int * type,
int * p_ntype,
double * vdwparam,
double * vvdw,
double * p_tabscale,
double * VFtab,
double * invsqrta,
double * dvda,
double * p_gbtabscale,
double * GBtab,
int * p_nthreads,
int * count,
void * mtx,
int * outeriter,
int * inneriter,
double * work)
{
int nri,ntype,nthreads;
int n,ii,is3,ii3,k,nj0,nj1,ggid;
double shX,shY,shZ;
int offset,nti;
int jnrA,jnrB;
int j3A,j3B;
int tjA,tjB;
gmx_gbdata_t *gbdata;
double * gpol;
__m128d iq,qq,jq,isai;
__m128d ix,iy,iz;
__m128d jx,jy,jz;
__m128d dx,dy,dz;
__m128d vctot,vvdwtot,vgbtot,dvdasum,gbfactor;
__m128d fix,fiy,fiz,tx,ty,tz,rsq;
__m128d rinv,isaj,isaprod;
__m128d vcoul,fscal,gbscale,c6,c12;
__m128d rinvsq,r,rtab;
__m128d eps,Y,F,G,H;
__m128d VV,FF,Fp;
__m128d vgb,fijGB,dvdatmp;
__m128d rinvsix,vvdw6,vvdw12,vvdwtmp;
__m128d facel,gbtabscale,dvdaj;
__m128d fijD,fijR;
__m128d xmm1,tabscale,eps2;
__m128i n0, nnn;
const __m128d neg = _mm_set1_pd(-1.0);
const __m128d zero = _mm_set1_pd(0.0);
const __m128d minushalf = _mm_set1_pd(-0.5);
const __m128d two = _mm_set1_pd(2.0);
gbdata = (gmx_gbdata_t *)work;
gpol = gbdata->gpol;
nri = *p_nri;
ntype = *p_ntype;
gbfactor = _mm_set1_pd( - ((1.0/gbdata->epsilon_r) - (1.0/gbdata->gb_epsilon_solvent)));
gbtabscale = _mm_load1_pd(p_gbtabscale);
facel = _mm_load1_pd(p_facel);
tabscale = _mm_load1_pd(p_tabscale);
nj1 = 0;
jnrA = jnrB = 0;
j3A = j3B = 0;
jx = _mm_setzero_pd();
jy = _mm_setzero_pd();
jz = _mm_setzero_pd();
c6 = _mm_setzero_pd();
c12 = _mm_setzero_pd();
for(n=0;n<nri;n++)
{
is3 = 3*shift[n];
shX = shiftvec[is3];
shY = shiftvec[is3+1];
shZ = shiftvec[is3+2];
nj0 = jindex[n];
nj1 = jindex[n+1];
ii = iinr[n];
ii3 = 3*ii;
ix = _mm_set1_pd(shX+pos[ii3+0]);
iy = _mm_set1_pd(shY+pos[ii3+1]);
iz = _mm_set1_pd(shZ+pos[ii3+2]);
iq = _mm_load1_pd(charge+ii);
iq = _mm_mul_pd(iq,facel);
isai = _mm_load1_pd(invsqrta+ii);
nti = 2*ntype*type[ii];
vctot = _mm_setzero_pd();
vvdwtot = _mm_setzero_pd();
vgbtot = _mm_setzero_pd();
dvdasum = _mm_setzero_pd();
fix = _mm_setzero_pd();
fiy = _mm_setzero_pd();
fiz = _mm_setzero_pd();
for(k=nj0;k<nj1-1; k+=2)
{
jnrA = jjnr[k];
jnrB = jjnr[k+1];
j3A = jnrA * 3;
j3B = jnrB * 3;
GMX_MM_LOAD_1RVEC_2POINTERS_PD(pos+j3A,pos+j3B,jx,jy,jz);
dx = _mm_sub_pd(ix,jx);
dy = _mm_sub_pd(iy,jy);
dz = _mm_sub_pd(iz,jz);
rsq = gmx_mm_calc_rsq_pd(dx,dy,dz);
rinv = gmx_mm_invsqrt_pd(rsq);
rinvsq = _mm_mul_pd(rinv,rinv);
/***********************************/
/* INTERACTION SECTION STARTS HERE */
/***********************************/
GMX_MM_LOAD_2VALUES_PD(charge+jnrA,charge+jnrB,jq);
GMX_MM_LOAD_2VALUES_PD(invsqrta+jnrA,invsqrta+jnrB,isaj);
/* Lennard-Jones */
tjA = nti+2*type[jnrA];
tjB = nti+2*type[jnrB];
GMX_MM_LOAD_2PAIRS_PD(vdwparam+tjA,vdwparam+tjB,c6,c12);
isaprod = _mm_mul_pd(isai,isaj);
qq = _mm_mul_pd(iq,jq);
vcoul = _mm_mul_pd(qq,rinv);
fscal = _mm_mul_pd(vcoul,rinv);
vctot = _mm_add_pd(vctot,vcoul);
/* Polarization interaction */
qq = _mm_mul_pd(qq,_mm_mul_pd(isaprod,gbfactor));
gbscale = _mm_mul_pd(isaprod,gbtabscale);
/* Calculate GB table index */
r = _mm_mul_pd(rsq,rinv);
rtab = _mm_mul_pd(r,gbscale);
n0 = _mm_cvttpd_epi32(rtab);
eps = _mm_sub_pd(rtab,_mm_cvtepi32_pd(n0));
nnn = _mm_slli_epi32(n0,2);
/* the tables are 16-byte aligned, so we can use _mm_load_pd */
Y = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,0)));
F = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,1)));
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,0))+2);
H = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,1))+2);
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_pd(G,eps);
H = _mm_mul_pd(H, _mm_mul_pd(eps,eps) );
F = _mm_add_pd(F, _mm_add_pd( G , H ) );
Y = _mm_add_pd(Y, _mm_mul_pd(F, eps));
F = _mm_add_pd(F, _mm_add_pd(G , _mm_mul_pd(H,two)));
vgb = _mm_mul_pd(Y, qq);
fijGB = _mm_mul_pd(F, _mm_mul_pd(qq,gbscale));
dvdatmp = _mm_mul_pd(_mm_add_pd(vgb, _mm_mul_pd(fijGB,r)) , minushalf);
vgbtot = _mm_add_pd(vgbtot, vgb);
dvdasum = _mm_add_pd(dvdasum, dvdatmp);
dvdatmp = _mm_mul_pd(dvdatmp, _mm_mul_pd(isaj,isaj));
GMX_MM_INCREMENT_2VALUES_PD(dvda+jnrA,dvda+jnrB,dvdatmp);
/* Calculate VDW table index */
rtab = _mm_mul_pd(r,tabscale);
n0 = _mm_cvttpd_epi32(rtab);
eps = _mm_sub_pd(rtab,_mm_cvtepi32_pd(n0));
eps2 = _mm_mul_pd(eps,eps);
nnn = _mm_slli_epi32(n0,3);
/* Dispersion */
Y = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0)));
F = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,1)));
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+2);
H = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,1))+2);
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_pd(G,eps);
H = _mm_mul_pd(H,eps2);
Fp = _mm_add_pd(F,G);
Fp = _mm_add_pd(Fp,H);
VV = _mm_mul_pd(Fp,eps);
VV = _mm_add_pd(Y,VV);
xmm1 = _mm_mul_pd(two,H);
FF = _mm_add_pd(Fp,G);
FF = _mm_add_pd(FF,xmm1);
vvdw6 = _mm_mul_pd(c6,VV);
fijD = _mm_mul_pd(c6,FF);
/* Dispersion */
Y = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+4);
F = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,1))+4);
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+6);
H = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,1))+6);
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_pd(G,eps);
H = _mm_mul_pd(H,eps2);
Fp = _mm_add_pd(F,G);
Fp = _mm_add_pd(Fp,H);
VV = _mm_mul_pd(Fp,eps);
VV = _mm_add_pd(Y,VV);
xmm1 = _mm_mul_pd(two,H);
FF = _mm_add_pd(Fp,G);
FF = _mm_add_pd(FF,xmm1);
vvdw12 = _mm_mul_pd(c12,VV);
fijR = _mm_mul_pd(c12,FF);
vvdwtmp = _mm_add_pd(vvdw12,vvdw6);
vvdwtot = _mm_add_pd(vvdwtot,vvdwtmp);
xmm1 = _mm_add_pd(fijD,fijR);
xmm1 = _mm_mul_pd(xmm1,tabscale);
xmm1 = _mm_add_pd(xmm1,fijGB);
xmm1 = _mm_sub_pd(xmm1,fscal);
fscal = _mm_mul_pd(xmm1,neg);
fscal = _mm_mul_pd(fscal,rinv);
/***********************************/
/* INTERACTION SECTION ENDS HERE */
/***********************************/
/* Calculate temporary vectorial force */
tx = _mm_mul_pd(fscal,dx);
ty = _mm_mul_pd(fscal,dy);
tz = _mm_mul_pd(fscal,dz);
/* Increment i atom force */
fix = _mm_add_pd(fix,tx);
fiy = _mm_add_pd(fiy,ty);
fiz = _mm_add_pd(fiz,tz);
/* Store j forces back */
GMX_MM_DECREMENT_1RVEC_2POINTERS_PD(faction+j3A,faction+j3B,tx,ty,tz);
}
/* In double precision, offset can only be either 0 or 1 */
if(k<nj1)
{
jnrA = jjnr[k];
j3A = jnrA * 3;
GMX_MM_LOAD_1RVEC_1POINTER_PD(pos+j3A,jx,jy,jz);
dx = _mm_sub_sd(ix,jx);
dy = _mm_sub_sd(iy,jy);
dz = _mm_sub_sd(iz,jz);
rsq = gmx_mm_calc_rsq_pd(dx,dy,dz);
rinv = gmx_mm_invsqrt_pd(rsq);
rinvsq = _mm_mul_sd(rinv,rinv);
/* These reason for zeroing these variables here is for fixing bug 585
* What happens is that __m128d _mm_add_sd(a,b) gives back r0=a[0]+b[0],
* and r1=0, but it should be r1=a[1].
* This might be a compiler issue (tested with gcc-4.1.3 and -O3).
* To work around it, we zero these variables and use _mm_add_pd (**) instead
* Note that the only variables that get affected are the energies since
* the total sum needs to be correct
*/
vgb = _mm_setzero_pd();
vcoul = _mm_setzero_pd();
dvdatmp = _mm_setzero_pd();
vvdw6 = _mm_setzero_pd();
vvdw12 = _mm_setzero_pd();
/***********************************/
/* INTERACTION SECTION STARTS HERE */
/***********************************/
GMX_MM_LOAD_1VALUE_PD(charge+jnrA,jq);
GMX_MM_LOAD_1VALUE_PD(invsqrta+jnrA,isaj);
/* Lennard-Jones */
tjA = nti+2*type[jnrA];
GMX_MM_LOAD_1PAIR_PD(vdwparam+tjA,c6,c12);
isaprod = _mm_mul_sd(isai,isaj);
qq = _mm_mul_sd(jq,iq);
vcoul = _mm_mul_sd(qq,rinv);
fscal = _mm_mul_sd(vcoul,rinv);
vctot = _mm_add_pd(vctot,vcoul); /* (**) */
/* Polarization interaction */
qq = _mm_mul_sd(qq,_mm_mul_sd(isaprod,gbfactor));
gbscale = _mm_mul_sd(isaprod,gbtabscale);
/* Calculate GB table index */
r = _mm_mul_sd(rsq,rinv);
rtab = _mm_mul_sd(r,gbscale);
n0 = _mm_cvttpd_epi32(rtab);
eps = _mm_sub_sd(rtab,_mm_cvtepi32_pd(n0));
nnn = _mm_slli_epi32(n0,2);
/* the tables are 16-byte aligned, so we can use _mm_load_pd */
Y = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,0)));
F = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(GBtab+(gmx_mm_extract_epi32(nnn,0))+2);
H = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_sd(G,eps);
H = _mm_mul_sd(H, _mm_mul_sd(eps,eps) );
F = _mm_add_sd(F, _mm_add_sd( G , H ) );
Y = _mm_add_sd(Y, _mm_mul_sd(F, eps));
F = _mm_add_sd(F, _mm_add_sd(G , _mm_mul_sd(H,two)));
vgb = _mm_mul_sd(Y, qq);
fijGB = _mm_mul_sd(F, _mm_mul_sd(qq,gbscale));
dvdatmp = _mm_mul_sd(_mm_add_sd(vgb, _mm_mul_sd(fijGB,r)) , minushalf);
vgbtot = _mm_add_pd(vgbtot, vgb); /* (**) */
dvdasum = _mm_add_pd(dvdasum, dvdatmp); /* (**) */
dvdatmp = _mm_mul_sd(dvdatmp, _mm_mul_sd(isaj,isaj));
GMX_MM_INCREMENT_1VALUE_PD(dvda+jnrA,dvdatmp);
/* Calculate VDW table index */
rtab = _mm_mul_sd(r,tabscale);
n0 = _mm_cvttpd_epi32(rtab);
eps = _mm_sub_sd(rtab,_mm_cvtepi32_pd(n0));
eps2 = _mm_mul_sd(eps,eps);
nnn = _mm_slli_epi32(n0,3);
/* Dispersion */
Y = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0)));
F = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+2);
H = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_sd(G,eps);
H = _mm_mul_sd(H,eps2);
Fp = _mm_add_sd(F,G);
Fp = _mm_add_sd(Fp,H);
VV = _mm_mul_sd(Fp,eps);
VV = _mm_add_sd(Y,VV);
xmm1 = _mm_mul_sd(two,H);
FF = _mm_add_sd(Fp,G);
FF = _mm_add_sd(FF,xmm1);
vvdw6 = _mm_mul_sd(c6,VV);
fijD = _mm_mul_sd(c6,FF);
/* Dispersion */
Y = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+4);
F = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(Y,F);
G = _mm_load_pd(VFtab+(gmx_mm_extract_epi32(nnn,0))+6);
H = _mm_setzero_pd();
GMX_MM_TRANSPOSE2_PD(G,H);
G = _mm_mul_sd(G,eps);
H = _mm_mul_sd(H,eps2);
Fp = _mm_add_sd(F,G);
Fp = _mm_add_sd(Fp,H);
VV = _mm_mul_sd(Fp,eps);
VV = _mm_add_sd(Y,VV);
xmm1 = _mm_mul_sd(two,H);
FF = _mm_add_sd(Fp,G);
FF = _mm_add_sd(FF,xmm1);
vvdw12 = _mm_mul_sd(c12,VV);
fijR = _mm_mul_sd(c12,FF);
vvdwtmp = _mm_add_sd(vvdw12,vvdw6);
vvdwtot = _mm_add_pd(vvdwtot,vvdwtmp); /* (**) */
xmm1 = _mm_add_sd(fijD,fijR);
xmm1 = _mm_mul_sd(xmm1,tabscale);
xmm1 = _mm_add_sd(xmm1,fijGB);
xmm1 = _mm_sub_sd(xmm1,fscal);
fscal = _mm_mul_sd(xmm1,neg);
fscal = _mm_mul_sd(fscal,rinv);
/***********************************/
/* INTERACTION SECTION ENDS HERE */
/***********************************/
/* Calculate temporary vectorial force */
tx = _mm_mul_sd(fscal,dx);
ty = _mm_mul_sd(fscal,dy);
tz = _mm_mul_sd(fscal,dz);
/* Increment i atom force */
fix = _mm_add_sd(fix,tx);
fiy = _mm_add_sd(fiy,ty);
fiz = _mm_add_sd(fiz,tz);
/* Store j forces back */
GMX_MM_DECREMENT_1RVEC_1POINTER_PD(faction+j3A,tx,ty,tz);
}
dvdasum = _mm_mul_pd(dvdasum, _mm_mul_pd(isai,isai));
gmx_mm_update_iforce_1atom_pd(&fix,&fiy,&fiz,faction+ii3,fshift+is3);
ggid = gid[n];
gmx_mm_update_1pot_pd(vctot,vc+ggid);
gmx_mm_update_1pot_pd(vgbtot,gpol+ggid);
gmx_mm_update_1pot_pd(dvdasum,dvda+ii);
gmx_mm_update_1pot_pd(vvdwtot,vvdw+ggid);
}
*outeriter = nri;
*inneriter = nj1;
}
void exchlaplacecoeff_gmrfData_5(unsigned int slot) {
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((!neighbor_isValid[0][0])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1306, S1309, S1300, S1308, S1302, S1305, S1310, S1304, S1307, S1301, S1303 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6302)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6338)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6302)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3782)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3818)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3782)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7562)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7598)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7562)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+38)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2522)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2558)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2522)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posBegin[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<31); i1 += 4) {
/* xPos = posBegin[0]; */
__m128d vec0 = _mm_load1_pd((&posBegin[0]));
__m128d vec0_2 = _mm_load1_pd((&posBegin[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<34); i1 += 1) {
xPos = posBegin[0];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8822)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8858)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8822)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1262)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1298)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1262)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(3.200000e+01);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<31); i1 += 4) {
/* yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<34); i1 += 1) {
yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5042)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5078)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5042)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10082)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10118)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10082)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[0][1])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1312, S1320, S1314, S1317, S1311, S1319, S1313, S1316, S1321, S1315, S1318 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1294)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1330)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+1294)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5074)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5110)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+5074)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8854)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8890)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+8854)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2554)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2590)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+2554)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(3.200000e+01);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<31); i1 += 4) {
/* yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<34); i1 += 1) {
yPos = ((((i1-1)/3.200000e+01)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+34)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+70)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+34)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3814)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3850)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+3814)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7594)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7630)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+7594)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6334)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6370)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+6334)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posEnd[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<31); i1 += 4) {
/* xPos = posEnd[0]; */
__m128d vec0 = _mm_load1_pd((&posEnd[0]));
__m128d vec0_2 = _mm_load1_pd((&posEnd[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<34); i1 += 1) {
xPos = posEnd[0];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i1 = 1;
for (; (i1<=32); i1 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10114)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10150)] = 0.000000e+00;
}
for (; (i1<=33); i1 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i1*36)+10114)] = 0.000000e+00;
}
}
}
}
}
if ((!neighbor_isValid[0][2])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1327, S1332, S1326, S1329, S1323, S1322, S1331, S1325, S1328, S1330, S1324 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+5076)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+5077)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+5076)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8856)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8857)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8856)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6336)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6337)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6336)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7596)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7597)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7596)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=33); i2 += 2) {
xPos = ((((i2-2)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=34); i2 += 1) {
xPos = ((((i2-2)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+36)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+37)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+36)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2556)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2557)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2556)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10116)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10117)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10116)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3816)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3817)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3816)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1296)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1297)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1296)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=33); i2 += 2) {
yPos = posBegin[1];
yPos = posBegin[1];
}
for (; (i2<=34); i2 += 1) {
yPos = posBegin[1];
}
}
}
}
}
if ((!neighbor_isValid[0][3])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S1338, S1341, S1335, S1340, S1343, S1337, S1334, S1333, S1342, S1336, S1339 */
{
{
{
{
{
{
{
{
{
{
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7488)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7489)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+7488)] = 0.000000e+00;
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2448)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2449)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+2448)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+11268)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+11269)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+11268)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=33); i2 += 2) {
yPos = posEnd[1];
yPos = posEnd[1];
}
for (; (i2<=34); i2 += 1) {
yPos = posEnd[1];
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10008)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10009)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+10008)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6228)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6229)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+6228)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8748)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8749)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+8748)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=33); i2 += 2) {
xPos = ((((i2-2)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=34); i2 += 1) {
xPos = ((((i2-2)/3.200000e+01)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3708)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3709)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+3708)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1188)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1189)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+1188)] = 0.000000e+00;
}
}
}
{
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][0]);
int i2 = 2;
for (; (i2<=33); i2 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+4968)] = 0.000000e+00;
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+4969)] = 0.000000e+00;
}
for (; (i2<=34); i2 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[(i2+4968)] = 0.000000e+00;
}
}
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1344 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Send_1_p1 = (&buffer_Send[1][(i3*33)]);
int i4 = 1;
for (; (i4<=32); i4 += 2) {
buffer_Send_1_p1[(i4-1)] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+34)];
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+70)];
}
for (; (i4<=33); i4 += 1) {
buffer_Send_1_p1[(i4-1)] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+34)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(buffer_Send[1], 297, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(buffer_Recv[0], 297, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1345 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i3*33)]);
int i4 = 3;
for (; (i4<=34); i4 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-70)] = buffer_Recv_0_p1[(i4-3)];
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-34)] = buffer_Recv_0_p1[(i4-2)];
}
for (; (i4<=35); i4 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-70)] = buffer_Recv_0_p1[(i4-3)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[5][1190], 1, mpiDatatype_9_33_1260, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[5][38], 1, mpiDatatype_9_33_1260, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1346 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Send_0_p1 = (&buffer_Send[0][(i3*35)]);
int i4 = 0;
for (; (i4<=33); i4 += 2) {
buffer_Send_0_p1[i4] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+3)];
buffer_Send_0_p1[(i4+1)] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+39)];
}
for (; (i4<=34); i4 += 1) {
buffer_Send_0_p1[i4] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+3)];
}
}
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1347 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Send_1_p1 = (&buffer_Send[1][(i3*35)]);
int i4 = 0;
for (; (i4<=33); i4 += 2) {
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+33)];
buffer_Send_1_p1[(i4+1)] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+69)];
}
for (; (i4<=34); i4 += 1) {
buffer_Send_1_p1[i4] = fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+33)];
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Isend(buffer_Send[0], 315, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][0]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[0]);
reqOutstanding_Send[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(buffer_Send[1], 315, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(buffer_Recv[0], 315, MPI_DOUBLE, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Irecv(buffer_Recv[1], 315, MPI_DOUBLE, neighbor_remoteRank[0][1], ((unsigned int)(neighbor_fragCommId[0][1]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
if (reqOutstanding_Recv[1]) {
waitForMPIReq(&mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
/* Statements in this Scop: S1348 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Recv_0_p1 = (&buffer_Recv[0][(i3*35)]);
int i4 = 1;
for (; (i4<=34); i4 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-35)] = buffer_Recv_0_p1[(i4-1)];
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)+1)] = buffer_Recv_0_p1[i4];
}
for (; (i4<=35); i4 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-35)] = buffer_Recv_0_p1[(i4-1)];
}
}
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
/* Statements in this Scop: S1349 */
for (int i3 = 0; (i3<=8); i3 += 1) {
double* fieldData_LaplaceCoeff_GMRF_5_p1 = (&fieldData_LaplaceCoeff_GMRF[5][(i3*1260)]);
double* buffer_Recv_1_p1 = (&buffer_Recv[1][(i3*35)]);
int i4 = 35;
for (; (i4<=68); i4 += 2) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-1225)] = buffer_Recv_1_p1[(i4-35)];
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-1189)] = buffer_Recv_1_p1[(i4-34)];
}
for (; (i4<=69); i4 += 1) {
fieldData_LaplaceCoeff_GMRF_5_p1[((i4*36)-1225)] = buffer_Recv_1_p1[(i4-35)];
}
}
}
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[0]) {
waitForMPIReq(&mpiRequest_Send[0]);
reqOutstanding_Send[0] = false;
}
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[5][73], 1, mpiDatatype_9_35_1260, neighbor_remoteRank[0][2], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][2]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[2]);
reqOutstanding_Send[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_LaplaceCoeff_GMRF[5][1153], 1, mpiDatatype_9_35_1260, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[5][1], 1, mpiDatatype_9_35_1260, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Irecv(&fieldData_LaplaceCoeff_GMRF[5][1225], 1, mpiDatatype_9_35_1260, neighbor_remoteRank[0][3], ((unsigned int)(neighbor_fragCommId[0][3]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
if (reqOutstanding_Recv[3]) {
waitForMPIReq(&mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[2]) {
waitForMPIReq(&mpiRequest_Send[2]);
reqOutstanding_Send[2] = false;
}
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
}
// rotate a matrix U = su2_i*U where su2_i is an su2 matrix embedded in suN
void
su2_rotate( GLU_complex U[ NCNC ] ,
const GLU_complex s0 ,
const GLU_complex s1 ,
const size_t su2_index )
{
#if NC == 3
__m128d *u = (__m128d*)U ;
register const __m128d sm0 = _mm_setr_pd( creal( s0 ) , cimag( s0 ) ) ;
register const __m128d sm1 = _mm_setr_pd( creal( s1 ) , cimag( s1 ) ) ;
register __m128d tmp0 , tmp1 , a , b ;
switch( su2_index%3 ) { // again I don't like this
case 0 :
// first one
a = *( u + 0 ) ; b = *( u + 3 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 0 ) = tmp0 ;
*( u + 3 ) = tmp1 ;
// second one
a = *( u + 1 ) ; b = *( u + 4 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 1 ) = tmp0 ;
*( u + 4 ) = tmp1 ;
// third
a = *( u + 2 ) ; b = *( u + 5 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 2 ) = tmp0 ;
*( u + 5 ) = tmp1 ;
break ;
case 1 :
// first one
a = *( u + 3 ) ; b = *( u + 6 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 3 ) = tmp0 ;
*( u + 6 ) = tmp1 ;
// second one
a = *( u + 4 ) ; b = *( u + 7 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 4 ) = tmp0 ;
*( u + 7 ) = tmp1 ;
// third
a = *( u + 5 ) ; b = *( u + 8 ) ;
tmp0 = _mm_add_pd( SSE2_MUL( sm0 , a ) ,
SSE2_MUL( sm1 , b ) ) ;
tmp1 = _mm_sub_pd( SSE2_MULCONJ( sm0 , b ) ,
SSE2_MULCONJ( sm1 , a ) ) ;
*( u + 5 ) = tmp0 ;
*( u + 8 ) = tmp1 ;
break ;
case 2 :
// first one
a = *( u + 0 ) ; b = *( u + 6 ) ;
tmp0 = _mm_sub_pd( SSE2_MULCONJ( sm0 , a ) ,
SSE2_MULCONJ( sm1 , b ) ) ;
tmp1 = _mm_add_pd( SSE2_MUL( sm0 , b ) ,
SSE2_MUL( sm1 , a ) ) ;
*( u + 0 ) = tmp0 ;
*( u + 6 ) = tmp1 ;
// second
a = *( u + 1 ) ; b = *( u + 7 ) ;
tmp0 = _mm_sub_pd( SSE2_MULCONJ( sm0 , a ) ,
SSE2_MULCONJ( sm1 , b ) ) ;
tmp1 = _mm_add_pd( SSE2_MUL( sm0 , b ) ,
SSE2_MUL( sm1 , a ) ) ;
*( u + 1 ) = tmp0 ;
*( u + 7 ) = tmp1 ;
// third
a = *( u + 2 ) ; b = *( u + 8 ) ;
tmp0 = _mm_sub_pd( SSE2_MULCONJ( sm0 , a ) ,
SSE2_MULCONJ( sm1 , b ) ) ;
tmp1 = _mm_add_pd( SSE2_MUL( sm0 , b ) ,
SSE2_MUL( sm1 , a ) ) ;
*( u + 2 ) = tmp0 ;
*( u + 8 ) = tmp1 ;
break ;
}
#elif NC == 2
__m128d *u = (__m128d*)U ;
register const __m128d sm0 = _mm_setr_pd( creal( s0 ) , cimag( s0 ) ) ;
register const __m128d sm1 = _mm_setr_pd( creal( s1 ) , cimag( s1 ) ) ;
*( u + 0 ) = _mm_add_pd( SSE2_MUL( sm0 , *( u + 0 ) ) ,
SSE2_MUL( sm1 , *( u + 2 ) ) ) ;
*( u + 1 ) = _mm_add_pd( SSE2_MUL( sm0 , *( u + 1 ) ) ,
SSE2_MUL( sm1 , *( u + 3 ) ) ) ;
*( u + 2 ) = SSE_FLIP( SSE2_CONJ( *( u + 1 ) ) ) ;
*( u + 3 ) = SSE2_CONJ( *( u + 0 ) ) ;
#else
// just a call to su2 multiply
shortened_su2_multiply( U , s0 , s1 , su2_index ) ;
#endif
return ;
}
DBL AVXFMA4Noise(const Vector3d& EPoint, int noise_generator)
{
DBL x, y, z;
DBL *mp;
int ix, iy, iz;
int ixiy_hash, ixjy_hash, jxiy_hash, jxjy_hash;
DBL sum;
// TODO FIXME - global statistics reference
// Stats[Calls_To_Noise]++;
if (noise_generator==kNoiseGen_Perlin)
{
// The 1.59 and 0.985 are to correct for some biasing problems with
// the random # generator used to create the noise tables. Final
// range of values is about 5.0e-4 below 0.0 and above 1.0. Mean
// value is 0.49 (ideally it would be 0.5).
sum = 0.5 * (1.59 * SolidNoise(EPoint) + 0.985);
// Clamp final value to 0-1 range
if (sum < 0.0) sum = 0.0;
if (sum > 1.0) sum = 1.0;
return sum;
}
x = EPoint[X];
y = EPoint[Y];
z = EPoint[Z];
/* its equivalent integer lattice point. */
/* ix = (int)x; iy = (int)y; iz = (long)z; */
/* JB fix for the range problem */
__m128d xy = _mm_setr_pd(x, y);
__m128d zn = _mm_set_sd(z);
__m128d epsy = _mm_set1_pd(1.0 - EPSILON);
__m128d xy_e = _mm_sub_pd(xy, epsy);
__m128d zn_e = _mm_sub_sd(zn, epsy);
__m128i tmp_xy = _mm_cvttpd_epi32(_mm_blendv_pd(xy, xy_e, xy));
__m128i tmp_zn = _mm_cvttpd_epi32(_mm_blendv_pd(zn, zn_e, zn));
__m128i noise_min_xy = _mm_setr_epi32(NOISE_MINX, NOISE_MINY, 0, 0);
__m128i noise_min_zn = _mm_set1_epi32(NOISE_MINZ);
__m128d xy_ixy = _mm_sub_pd(xy, _mm_cvtepi32_pd(tmp_xy));
__m128d zn_izn = _mm_sub_sd(zn, _mm_cvtepi32_pd(tmp_zn));
const __m128i fff = _mm_set1_epi32(0xfff);
__m128i i_xy = _mm_and_si128(_mm_sub_epi32(tmp_xy, noise_min_xy), fff);
__m128i i_zn = _mm_and_si128(_mm_sub_epi32(tmp_zn, noise_min_zn), fff);
ix = _mm_extract_epi32(i_xy, 0);
iy = _mm_extract_epi32(i_xy, 1);
iz = _mm_extract_epi32(i_zn, 0);
ixiy_hash = Hash2d(ix, iy);
jxiy_hash = Hash2d(ix + 1, iy);
ixjy_hash = Hash2d(ix, iy + 1);
jxjy_hash = Hash2d(ix + 1, iy + 1);
mp = &RTable[Hash1dRTableIndex(ixiy_hash, iz)];
DBL *mp2 = &RTable[Hash1dRTableIndex(ixjy_hash, iz)];
DBL *mp3 = &RTable[Hash1dRTableIndex(ixiy_hash, iz + 1)];
DBL *mp4 = &RTable[Hash1dRTableIndex(ixjy_hash, iz + 1)];
DBL *mp5 = &RTable[Hash1dRTableIndex(jxiy_hash, iz)];
DBL *mp6 = &RTable[Hash1dRTableIndex(jxjy_hash, iz)];
DBL *mp7 = &RTable[Hash1dRTableIndex(jxiy_hash, iz + 1)];
DBL *mp8 = &RTable[Hash1dRTableIndex(jxjy_hash, iz + 1)];
const __m128d three = _mm_set1_pd(3.0);
const __m128d two = _mm_set1_pd(2.0);
const __m128d one = _mm_set1_pd(1.0);
__m128d ix_mm = _mm_unpacklo_pd(xy_ixy, xy_ixy);
__m128d iy_mm = _mm_unpackhi_pd(xy_ixy, xy_ixy);
__m128d iz_mm = _mm_unpacklo_pd(zn_izn, zn_izn);
__m128d jx_mm = _mm_sub_pd(ix_mm, one);
__m128d jy_mm = _mm_sub_pd(iy_mm, one);
__m128d jz_mm = _mm_sub_pd(iz_mm, one);
__m128d mm_sxy = _mm_mul_pd(_mm_mul_pd(xy_ixy, xy_ixy), _mm_nmacc_pd(two, xy_ixy, three));
__m128d mm_sz = _mm_mul_pd(_mm_mul_pd(iz_mm, iz_mm), _mm_nmacc_pd(two, iz_mm, three));
__m128d mm_tz = _mm_sub_pd(one, mm_sz);
__m128d mm_txy = _mm_sub_pd(one, mm_sxy);
__m128d mm_tysy = _mm_unpackhi_pd(mm_txy, mm_sxy);
__m128d mm_txty_txsy = _mm_mul_pd(_mm_unpacklo_pd(mm_txy, mm_txy), mm_tysy);
__m128d mm_sxty_sxsy = _mm_mul_pd(_mm_unpacklo_pd(mm_sxy, mm_sxy), mm_tysy);
__m128d y_mm = _mm_unpacklo_pd(iy_mm, jy_mm);
__m128d mp_t1, mp_t2, mp1_mm, mp2_mm, mp4_mm, mp6_mm, sum_p, s_mm;
__m128d int_sum1 = _mm_setzero_pd();
s_mm = _mm_mul_pd(mm_txty_txsy, mm_tz);
INCRSUMP2(mp, mp2, s_mm, ix_mm, y_mm, iz_mm, int_sum1);
s_mm = _mm_mul_pd(mm_txty_txsy, mm_sz);
INCRSUMP2(mp3, mp4, s_mm, ix_mm, y_mm, jz_mm, int_sum1);
s_mm = _mm_mul_pd(mm_sxty_sxsy, mm_tz);
INCRSUMP2(mp5, mp6, s_mm, jx_mm, y_mm, iz_mm, int_sum1);
s_mm = _mm_mul_pd(mm_sxty_sxsy, mm_sz);
INCRSUMP2(mp7, mp8, s_mm, jx_mm, y_mm, jz_mm, int_sum1);
int_sum1 = _mm_hadd_pd(int_sum1, int_sum1);
if(noise_generator==kNoiseGen_RangeCorrected)
{
/* details of range here:
Min, max: -1.05242, 0.988997
Mean: -0.0191481, Median: -0.535493, Std Dev: 0.256828
We want to change it to as close to [0,1] as possible.
*/
const __m128d r2 = _mm_set_sd(0.48985582);
const __m128d r1r2 = _mm_set_sd(1.05242*0.48985582);
int_sum1 = _mm_macc_sd(int_sum1, r2, r1r2);
}
else
{
int_sum1 = _mm_add_sd(int_sum1, _mm_set_sd(0.5));
}
int_sum1 = _mm_min_sd(one, int_sum1);
int_sum1 = _mm_max_sd(_mm_setzero_pd(), int_sum1);
_mm_store_sd(&sum, int_sum1);
return (sum);
}
void multipath_channel(channel_desc_t *desc,
double **tx_sig_re,
double **tx_sig_im,
double **rx_sig_re,
double **rx_sig_im,
uint32_t length,
uint8_t keep_channel)
{
int i,ii,j,l;
int length1, length2, tail;
__m128d rx_tmp128_re_f,rx_tmp128_im_f,rx_tmp128_re,rx_tmp128_im, rx_tmp128_1,rx_tmp128_2,rx_tmp128_3,rx_tmp128_4,tx128_re,tx128_im,ch128_x,ch128_y,pathloss128;
double path_loss = pow(10,desc->path_loss_dB/20);
int dd = abs(desc->channel_offset);
pathloss128 = _mm_set1_pd(path_loss);
#ifdef DEBUG_CH
printf("[CHANNEL] keep = %d : path_loss = %g (%f), nb_rx %d, nb_tx %d, dd %d, len %d \n",keep_channel,path_loss,desc->path_loss_dB,desc->nb_rx,desc->nb_tx,dd,desc->channel_length);
#endif
if (keep_channel) {
// do nothing - keep channel
} else {
random_channel(desc,0);
}
start_meas(&desc->convolution);
#ifdef DEBUG_CH
for (l = 0; l<(int)desc->channel_length; l++) {
printf("%p (%f,%f) ",desc->ch[0],desc->ch[0][l].x,desc->ch[0][l].y);
}
printf("\n");
#endif
tail = ((int)length-dd)%2;
if(tail)
length1 = ((int)length-dd)-1;
else
length1 = ((int)length-dd);
length2 = length1/2;
for (i=0; i<length2; i++) { //
for (ii=0; ii<desc->nb_rx; ii++) {
// rx_tmp.x = 0;
// rx_tmp.y = 0;
rx_tmp128_re_f = _mm_setzero_pd();
rx_tmp128_im_f = _mm_setzero_pd();
for (j=0; j<desc->nb_tx; j++) {
for (l = 0; l<(int)desc->channel_length; l++) {
if ((i>=0) && (i-l)>=0) { //SIMD correct only if length1 > 2*channel_length...which is almost always satisfied
// tx.x = tx_sig_re[j][i-l];
// tx.y = tx_sig_im[j][i-l];
tx128_re = _mm_loadu_pd(&tx_sig_re[j][2*i-l]); // tx_sig_re[j][i-l+1], tx_sig_re[j][i-l]
tx128_im = _mm_loadu_pd(&tx_sig_im[j][2*i-l]);
} else {
//tx.x =0;
//tx.y =0;
tx128_re = _mm_setzero_pd();
tx128_im = _mm_setzero_pd();
}
ch128_x = _mm_set1_pd(desc->ch[ii+(j*desc->nb_rx)][l].x);
ch128_y = _mm_set1_pd(desc->ch[ii+(j*desc->nb_rx)][l].y);
// rx_tmp.x += (tx.x * desc->ch[ii+(j*desc->nb_rx)][l].x) - (tx.y * desc->ch[ii+(j*desc->nb_rx)][l].y);
// rx_tmp.y += (tx.y * desc->ch[ii+(j*desc->nb_rx)][l].x) + (tx.x * desc->ch[ii+(j*desc->nb_rx)][l].y);
rx_tmp128_1 = _mm_mul_pd(tx128_re,ch128_x);
rx_tmp128_2 = _mm_mul_pd(tx128_re,ch128_y);
rx_tmp128_3 = _mm_mul_pd(tx128_im,ch128_x);
rx_tmp128_4 = _mm_mul_pd(tx128_im,ch128_y);
rx_tmp128_re = _mm_sub_pd(rx_tmp128_1,rx_tmp128_4);
rx_tmp128_im = _mm_add_pd(rx_tmp128_2,rx_tmp128_3);
rx_tmp128_re_f = _mm_add_pd(rx_tmp128_re_f,rx_tmp128_re);
rx_tmp128_im_f = _mm_add_pd(rx_tmp128_im_f,rx_tmp128_im);
} //l
} // j
//rx_sig_re[ii][i+dd] = rx_tmp.x*path_loss;
//rx_sig_im[ii][i+dd] = rx_tmp.y*path_loss;
rx_tmp128_re_f = _mm_mul_pd(rx_tmp128_re_f,pathloss128);
rx_tmp128_im_f = _mm_mul_pd(rx_tmp128_im_f,pathloss128);
_mm_storeu_pd(&rx_sig_re[ii][2*i+dd],rx_tmp128_re_f); // max index: length-dd -1 + dd = length -1
_mm_storeu_pd(&rx_sig_im[ii][2*i+dd],rx_tmp128_im_f);
/*
if ((ii==0)&&((i%32)==0)) {
printf("%p %p %f,%f => %e,%e\n",rx_sig_re[ii],rx_sig_im[ii],rx_tmp.x,rx_tmp.y,rx_sig_re[ii][i-dd],rx_sig_im[ii][i-dd]);
}
*/
//rx_sig_re[ii][i] = sqrt(.5)*(tx_sig_re[0][i] + tx_sig_re[1][i]);
//rx_sig_im[ii][i] = sqrt(.5)*(tx_sig_im[0][i] + tx_sig_im[1][i]);
} // ii
} // i
stop_meas(&desc->convolution);
}
void AVXFMA4DNoise(Vector3d& result, const Vector3d& EPoint)
{
DBL x, y, z;
int ix, iy, iz;
int ixiy_hash, ixjy_hash, jxiy_hash, jxjy_hash;
// TODO FIXME - global statistics reference
// Stats[Calls_To_DNoise]++;
x = EPoint[X];
y = EPoint[Y];
z = EPoint[Z];
/* its equivalent integer lattice point. */
/*ix = (int)x; iy = (int)y; iz = (int)z;
x_ix = x - ix; y_iy = y - iy; z_iz = z - iz;*/
/* JB fix for the range problem */
__m128d xy = _mm_setr_pd(x, y);
__m128d zn = _mm_set_sd(z);
__m128d epsy = _mm_set1_pd(1.0 - EPSILON);
__m128d xy_e = _mm_sub_pd(xy, epsy);
__m128d zn_e = _mm_sub_sd(zn, epsy);
__m128i tmp_xy = _mm_cvttpd_epi32(_mm_blendv_pd(xy, xy_e, xy));
__m128i tmp_zn = _mm_cvttpd_epi32(_mm_blendv_pd(zn, zn_e, zn));
__m128i noise_min_xy = _mm_setr_epi32(NOISE_MINX, NOISE_MINY, 0, 0);
__m128i noise_min_zn = _mm_set1_epi32(NOISE_MINZ);
__m128d xy_ixy = _mm_sub_pd(xy, _mm_cvtepi32_pd(tmp_xy));
__m128d zn_izn = _mm_sub_sd(zn, _mm_cvtepi32_pd(tmp_zn));
const __m128i fff = _mm_set1_epi32(0xfff);
__m128i i_xy = _mm_and_si128(_mm_sub_epi32(tmp_xy, noise_min_xy), fff);
__m128i i_zn = _mm_and_si128(_mm_sub_epi32(tmp_zn, noise_min_zn), fff);
ix = _mm_extract_epi32(i_xy, 0);
iy = _mm_extract_epi32(i_xy, 1);
iz = _mm_extract_epi32(i_zn, 0);
ixiy_hash = Hash2d(ix, iy);
jxiy_hash = Hash2d(ix + 1, iy);
ixjy_hash = Hash2d(ix, iy + 1);
jxjy_hash = Hash2d(ix + 1, iy + 1);
DBL* mp1 = &RTable[Hash1dRTableIndex(ixiy_hash, iz)];
DBL* mp2 = &RTable[Hash1dRTableIndex(jxiy_hash, iz)];
DBL* mp3 = &RTable[Hash1dRTableIndex(jxjy_hash, iz)];
DBL* mp4 = &RTable[Hash1dRTableIndex(ixjy_hash, iz)];
DBL* mp5 = &RTable[Hash1dRTableIndex(ixjy_hash, iz + 1)];
DBL* mp6 = &RTable[Hash1dRTableIndex(jxjy_hash, iz + 1)];
DBL* mp7 = &RTable[Hash1dRTableIndex(jxiy_hash, iz + 1)];
DBL* mp8 = &RTable[Hash1dRTableIndex(ixiy_hash, iz + 1)];
const __m128d three = _mm_set1_pd(3.0);
const __m128d two = _mm_set1_pd(2.0);
const __m128d one = _mm_set1_pd(1.0);
__m128d ix_mm = _mm_unpacklo_pd(xy_ixy, xy_ixy);
__m128d iy_mm = _mm_unpackhi_pd(xy_ixy, xy_ixy);
__m128d iz_mm = _mm_unpacklo_pd(zn_izn, zn_izn);
__m128d jx_mm = _mm_sub_pd(ix_mm, one);
__m128d jy_mm = _mm_sub_pd(iy_mm, one);
__m128d jz_mm = _mm_sub_pd(iz_mm, one);
__m128d mm_sz = _mm_mul_pd(_mm_mul_pd(iz_mm, iz_mm), _mm_nmacc_pd(two, iz_mm, three));
__m128d mm_tz = _mm_sub_pd(one, mm_sz);
__m128d mm_sxy = _mm_mul_pd(_mm_mul_pd(xy_ixy, xy_ixy), _mm_nmacc_pd(two, xy_ixy, three));
__m128d mm_txy = _mm_sub_pd(one, mm_sxy);
__m128d mm_tysy = _mm_unpackhi_pd(mm_txy, mm_sxy);
__m128d mm_txty_txsy = _mm_mul_pd(_mm_unpacklo_pd(mm_txy, mm_txy), mm_tysy);
__m128d mm_sxty_sxsy = _mm_mul_pd(_mm_unpacklo_pd(mm_sxy, mm_sxy), mm_tysy);
__m128d mm_txty_txsy_tz = _mm_mul_pd(mm_txty_txsy, mm_tz);
__m128d mm_txty_txsy_sz = _mm_mul_pd(mm_txty_txsy, mm_sz);
__m128d mm_sxty_sxsy_tz = _mm_mul_pd(mm_sxty_sxsy, mm_tz);
__m128d mm_sxty_sxsy_sz = _mm_mul_pd(mm_sxty_sxsy, mm_sz);
__m128d mp_t1, mp_t2, mp1_mm, mp2_mm, mp4_mm, mp6_mm, sum_p;
__m128d sum_X_Y = _mm_setzero_pd();
__m128d sum__Z = _mm_setzero_pd();
__m128d mm_s1 = _mm_unpacklo_pd(mm_txty_txsy_tz, mm_txty_txsy_tz);
INCRSUMP2(mp1, mp1 + 8, mm_s1, ix_mm, iy_mm, iz_mm, sum_X_Y);
__m128d mm_s2 = _mm_unpacklo_pd(mm_sxty_sxsy_tz, mm_sxty_sxsy_tz);
INCRSUMP2(mp2, mp2 + 8, mm_s2, jx_mm, iy_mm, iz_mm, sum_X_Y);
__m128d mm_s3 = _mm_unpackhi_pd(mm_sxty_sxsy_tz, mm_sxty_sxsy_tz);
INCRSUMP2(mp3, mp3 + 8, mm_s3, jx_mm, jy_mm, iz_mm, sum_X_Y);
__m128d mm_s4 = _mm_unpackhi_pd(mm_txty_txsy_tz, mm_txty_txsy_tz);
INCRSUMP2(mp4, mp4 + 8, mm_s4, ix_mm, jy_mm, iz_mm, sum_X_Y);
__m128d mm_s5 = _mm_unpackhi_pd(mm_txty_txsy_sz, mm_txty_txsy_sz);
INCRSUMP2(mp5, mp5 + 8, mm_s5, ix_mm, jy_mm, jz_mm, sum_X_Y);
__m128d mm_s6 = _mm_unpackhi_pd(mm_sxty_sxsy_sz, mm_sxty_sxsy_sz);
INCRSUMP2(mp6, mp6 + 8, mm_s6, jx_mm, jy_mm, jz_mm, sum_X_Y);
__m128d mm_s7 = _mm_unpacklo_pd(mm_sxty_sxsy_sz, mm_sxty_sxsy_sz);
INCRSUMP2(mp7, mp7 + 8, mm_s7, jx_mm, iy_mm, jz_mm, sum_X_Y);
__m128d mm_s8 = _mm_unpacklo_pd(mm_txty_txsy_sz, mm_txty_txsy_sz);
INCRSUMP2(mp8, mp8 + 8, mm_s8, ix_mm, iy_mm, jz_mm, sum_X_Y);
__m128d iy_jy = _mm_unpacklo_pd(iy_mm, jy_mm);
INCRSUMP2(mp1 + 16, mp4 + 16, mm_txty_txsy_tz, ix_mm, iy_jy, iz_mm, sum__Z);
INCRSUMP2(mp8 + 16, mp5 + 16, mm_txty_txsy_sz, ix_mm, iy_jy, jz_mm, sum__Z);
INCRSUMP2(mp2 + 16, mp3 + 16, mm_sxty_sxsy_tz, jx_mm, iy_jy, iz_mm, sum__Z);
INCRSUMP2(mp7 + 16, mp6 + 16, mm_sxty_sxsy_sz, jx_mm, iy_jy, jz_mm, sum__Z);
sum__Z = _mm_hadd_pd(sum__Z, sum__Z);
_mm_storeu_pd(*result, sum_X_Y);
_mm_store_sd(&result[Z], sum__Z);
}
void exchsolution_gmrfData_1(unsigned int slot) {
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((!neighbor_isValid[0][0])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S493, S492, S494 */
{
{
{
double* fieldData_Solution_GMRF_1_p1 = (&fieldData_Solution_GMRF[1][0]);
int i1 = 1;
for (; (i1<=2); i1 += 2) {
fieldData_Solution_GMRF_1_p1[((i1*6)+2)] = 0.000000e+00;
fieldData_Solution_GMRF_1_p1[((i1*6)+8)] = 0.000000e+00;
}
for (; (i1<=3); i1 += 1) {
fieldData_Solution_GMRF_1_p1[((i1*6)+2)] = 0.000000e+00;
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posBegin[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<1); i1 += 4) {
/* xPos = posBegin[0]; */
__m128d vec0 = _mm_load1_pd((&posBegin[0]));
__m128d vec0_2 = _mm_load1_pd((&posBegin[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<4); i1 += 1) {
xPos = posBegin[0];
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(2.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<1); i1 += 4) {
/* yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<4); i1 += 1) {
yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
}
}
if ((!neighbor_isValid[0][1])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S496, S495, S497 */
{
{
{
double* fieldData_Solution_GMRF_1_p1 = (&fieldData_Solution_GMRF[1][0]);
int i1 = 1;
for (; (i1<=2); i1 += 2) {
fieldData_Solution_GMRF_1_p1[((i1*6)+4)] = 0.000000e+00;
fieldData_Solution_GMRF_1_p1[((i1*6)+10)] = 0.000000e+00;
}
for (; (i1<=3); i1 += 1) {
fieldData_Solution_GMRF_1_p1[((i1*6)+4)] = 0.000000e+00;
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posEnd[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<1); i1 += 4) {
/* xPos = posEnd[0]; */
__m128d vec0 = _mm_load1_pd((&posEnd[0]));
__m128d vec0_2 = _mm_load1_pd((&posEnd[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<4); i1 += 1) {
xPos = posEnd[0];
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(2.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<1); i1 += 4) {
/* yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<4); i1 += 1) {
yPos = ((((i1-1)/2.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
}
}
if ((!neighbor_isValid[0][2])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S500, S499, S498 */
{
{
{
int i2 = 2;
for (; (i2<=3); i2 += 2) {
xPos = ((((i2-2)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=4); i2 += 1) {
xPos = ((((i2-2)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
{
double* fieldData_Solution_GMRF_1_p1 = (&fieldData_Solution_GMRF[1][0]);
int i2 = 2;
for (; (i2<=3); i2 += 2) {
fieldData_Solution_GMRF_1_p1[(i2+6)] = 0.000000e+00;
fieldData_Solution_GMRF_1_p1[(i2+7)] = 0.000000e+00;
}
for (; (i2<=4); i2 += 1) {
fieldData_Solution_GMRF_1_p1[(i2+6)] = 0.000000e+00;
}
}
}
{
int i2 = 2;
for (; (i2<=3); i2 += 2) {
yPos = posBegin[1];
yPos = posBegin[1];
}
for (; (i2<=4); i2 += 1) {
yPos = posBegin[1];
}
}
}
}
}
if ((!neighbor_isValid[0][3])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S503, S502, S501 */
{
{
{
int i2 = 2;
for (; (i2<=3); i2 += 2) {
xPos = ((((i2-2)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=4); i2 += 1) {
xPos = ((((i2-2)/2.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
{
int i2 = 2;
for (; (i2<=3); i2 += 2) {
yPos = posEnd[1];
yPos = posEnd[1];
}
for (; (i2<=4); i2 += 1) {
yPos = posEnd[1];
}
}
}
{
double* fieldData_Solution_GMRF_1_p1 = (&fieldData_Solution_GMRF[1][0]);
int i2 = 2;
for (; (i2<=3); i2 += 2) {
fieldData_Solution_GMRF_1_p1[(i2+18)] = 0.000000e+00;
fieldData_Solution_GMRF_1_p1[(i2+19)] = 0.000000e+00;
}
for (; (i2<=4); i2 += 1) {
fieldData_Solution_GMRF_1_p1[(i2+18)] = 0.000000e+00;
}
}
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(&fieldData_Solution_GMRF[1][10], 1, mpiDatatype_3_1_6, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][8], 1, mpiDatatype_3_1_6, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_Solution_GMRF[1][20], 1, mpiDatatype_1_3_6, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][8], 1, mpiDatatype_1_3_6, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Isend(&fieldData_Solution_GMRF[1][3], 1, mpiDatatype_5_1_6, neighbor_remoteRank[0][0], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][0]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[0]);
reqOutstanding_Send[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Isend(&fieldData_Solution_GMRF[1][3], 1, mpiDatatype_5_1_6, neighbor_remoteRank[0][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][0]&&neighbor_isRemote[0][0])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][1], 1, mpiDatatype_5_1_6, neighbor_remoteRank[0][0], ((unsigned int)(neighbor_fragCommId[0][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
if ((neighbor_isValid[0][1]&&neighbor_isRemote[0][1])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][5], 1, mpiDatatype_5_1_6, neighbor_remoteRank[0][1], ((unsigned int)(neighbor_fragCommId[0][1]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
if (reqOutstanding_Recv[1]) {
waitForMPIReq(&mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[0]) {
waitForMPIReq(&mpiRequest_Send[0]);
reqOutstanding_Send[0] = false;
}
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Isend(&fieldData_Solution_GMRF[1][13], 1, mpiDatatype_1_5_6, neighbor_remoteRank[0][2], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][2]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[2]);
reqOutstanding_Send[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Isend(&fieldData_Solution_GMRF[1][13], 1, mpiDatatype_1_5_6, neighbor_remoteRank[0][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[0][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if ((neighbor_isValid[0][2]&&neighbor_isRemote[0][2])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][1], 1, mpiDatatype_1_5_6, neighbor_remoteRank[0][2], ((unsigned int)(neighbor_fragCommId[0][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
if ((neighbor_isValid[0][3]&&neighbor_isRemote[0][3])) {
MPI_Irecv(&fieldData_Solution_GMRF[1][25], 1, mpiDatatype_1_5_6, neighbor_remoteRank[0][3], ((unsigned int)(neighbor_fragCommId[0][3]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
if (reqOutstanding_Recv[3]) {
waitForMPIReq(&mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[0]) {
if (reqOutstanding_Send[2]) {
waitForMPIReq(&mpiRequest_Send[2]);
reqOutstanding_Send[2] = false;
}
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
}
inline const vector2d operator-(const vector2d& lhs, const vector2d& rhs) {
return _mm_sub_pd(lhs, rhs);
}
static forcedinline ParallelType sub (ParallelType a, ParallelType b) noexcept { return _mm_sub_pd (a, b); }
// it moves horizontally inside a block
void kernel_dtrmv_u_n_2_lib4(int kmax, double *A, double *x, double *y, int alg)
{
if(kmax<=0)
return;
const int lda = 4;
int k;
__m128d
ax_temp,
a_00_10, a_01_11, a_02_12, a_03_13,
x_0, x_1, x_2, x_3,
y_0_1, y_0_1_b, y_0_1_c, y_0_1_d, z_0_1;
/* y_0_1 = _mm_setzero_pd(); */
// second col (avoid zero y_0_1)
x_0 = _mm_loaddup_pd( &x[1] );
a_00_10 = _mm_load_pd( &A[0+lda*1] );
y_0_1 = _mm_mul_pd( a_00_10, x_0 );
// first col
x_0 = _mm_load_sd( &x[0] );
a_00_10 = _mm_load_sd( &A[0+lda*0] );
ax_temp = _mm_mul_sd( a_00_10, x_0 );
y_0_1 = _mm_add_sd( y_0_1, ax_temp );
A += 2*lda;
x += 2;
k=2;
for(; k<kmax-1; k+=2)
{
x_0 = _mm_loaddup_pd( &x[0] );
x_1 = _mm_loaddup_pd( &x[1] );
a_00_10 = _mm_load_pd( &A[0+lda*0] );
a_01_11 = _mm_load_pd( &A[0+lda*1] );
ax_temp = _mm_mul_pd( a_00_10, x_0 );
y_0_1 = _mm_add_pd( y_0_1, ax_temp );
ax_temp = _mm_mul_pd( a_01_11, x_1 );
y_0_1 = _mm_add_pd( y_0_1, ax_temp );
A += 2*lda;
x += 2;
}
if(kmax%2==1)
{
x_0 = _mm_loaddup_pd( &x[0] );
a_00_10 = _mm_load_pd( &A[0+lda*0] );
ax_temp = _mm_mul_pd( a_00_10, x_0 );
y_0_1 = _mm_add_pd( y_0_1, ax_temp );
}
if(alg==0)
{
_mm_storeu_pd(&y[0], y_0_1);
}
else if(alg==1)
{
z_0_1 = _mm_loadu_pd( &y[0] );
z_0_1 = _mm_add_pd( z_0_1, y_0_1 );
_mm_storeu_pd(&y[0], z_0_1);
}
else // alg==-1
{
z_0_1 = _mm_loadu_pd( &y[0] );
z_0_1 = _mm_sub_pd( z_0_1, y_0_1 );
_mm_storeu_pd(&y[0], z_0_1);
}
}
static void
filterYule(const Float_t* input, Float_t* output, size_t nSamples, const Float_t* kernel)
{
#ifdef HAVE_SSE2
__m128d __kernel, __result, __temp;
__declspec(align(16)) Float_t __temp2[2];
while (nSamples--) {
__kernel = _mm_loadr_pd(&kernel[0]);
__temp = _mm_loadu_pd(&input[-1]);
__result = _mm_mul_pd(__temp, __kernel);
__kernel = _mm_loadr_pd(&kernel[12]);
__temp = _mm_loadu_pd(&output[-2]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_sub_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[2]);
__temp = _mm_loadu_pd(&input[-3]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_add_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[14]);
__temp = _mm_loadu_pd(&output[-4]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_sub_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[4]);
__temp = _mm_loadu_pd(&input[-5]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_add_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[16]);
__temp = _mm_loadu_pd(&output[-6]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_sub_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[6]);
__temp = _mm_loadu_pd(&input[-7]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_add_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[18]);
__temp = _mm_loadu_pd(&output[-8]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_sub_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[8]);
__temp = _mm_loadu_pd(&input[-9]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_add_pd(__result, __temp);
__kernel = _mm_loadr_pd(&kernel[20]);
__temp = _mm_loadu_pd(&output[-10]);
__temp = _mm_mul_pd(__kernel, __temp);
__result = _mm_sub_pd(__result, __temp);
_mm_store_pd(__temp2, __result);
*output = 1e-10 /* 1e-10 is a hack to avoid slowdown because of denormals */
+ __temp2[0]
+ __temp2[1]
+ input [-10] * kernel[11];
;
++output;
++input;
}
#else
while (nSamples--) {
*output = 1e-10 /* 1e-10 is a hack to avoid slowdown because of denormals */
+ input [0] * kernel[0] - output[-1] * kernel[1]
+ input [-1] * kernel[2] - output[-2] * kernel[3]
+ input [-2] * kernel[4] - output[-3] * kernel[5]
+ input [-3] * kernel[6] - output[-4] * kernel[7]
+ input [-4] * kernel[8] - output[-5] * kernel[9]
+ input [-5] * kernel[10] - output[-6] * kernel[11]
+ input [-6] * kernel[12] - output[-7] * kernel[13]
+ input [-7] * kernel[14] - output[-8] * kernel[15]
+ input [-8] * kernel[16] - output[-9] * kernel[17]
+ input [-9] * kernel[18] - output[-10]* kernel[19]
+ input [-10]* kernel[20];
++output;
++input;
}
#endif
}
// it moves vertically across blocks
void kernel_dtrmv_u_t_2_lib4(int kmax, double *A, int sda, double *x, double *y, int alg)
{
/* if(kmax<=0) */
/* return;*/
const int lda = 4;
double *tA, *tx;
int k;
__m256d
tmp0, tmp1,
a_00_10_20_30, a_01_11_21_31,
x_0_1_2_3,
y_00, y_11;
y_00 = _mm256_setzero_pd();
y_11 = _mm256_setzero_pd();
k=0;
for(; k<kmax-7; k+=8)
{
x_0_1_2_3 = _mm256_loadu_pd( &x[0] );
a_00_10_20_30 = _mm256_load_pd( &A[0+lda*0] );
a_01_11_21_31 = _mm256_load_pd( &A[0+lda*1] );
tmp0 = _mm256_mul_pd( a_00_10_20_30, x_0_1_2_3 );
tmp1 = _mm256_mul_pd( a_01_11_21_31, x_0_1_2_3 );
y_00 = _mm256_add_pd( y_00, tmp0 );
y_11 = _mm256_add_pd( y_11, tmp1 );
A += 4 + (sda-1)*lda;
x += 4;
x_0_1_2_3 = _mm256_loadu_pd( &x[0] );
a_00_10_20_30 = _mm256_load_pd( &A[0+lda*0] );
a_01_11_21_31 = _mm256_load_pd( &A[0+lda*1] );
tmp0 = _mm256_mul_pd( a_00_10_20_30, x_0_1_2_3 );
tmp1 = _mm256_mul_pd( a_01_11_21_31, x_0_1_2_3 );
y_00 = _mm256_add_pd( y_00, tmp0 );
y_11 = _mm256_add_pd( y_11, tmp1 );
A += 4 + (sda-1)*lda;
x += 4;
}
for(; k<kmax-3; k+=4)
{
x_0_1_2_3 = _mm256_loadu_pd( &x[0] );
a_00_10_20_30 = _mm256_load_pd( &A[0+lda*0] );
a_01_11_21_31 = _mm256_load_pd( &A[0+lda*1] );
tmp0 = _mm256_mul_pd( a_00_10_20_30, x_0_1_2_3 );
tmp1 = _mm256_mul_pd( a_01_11_21_31, x_0_1_2_3 );
y_00 = _mm256_add_pd( y_00, tmp0 );
y_11 = _mm256_add_pd( y_11, tmp1 );
A += 4 + (sda-1)*lda;
x += 4;
}
__m128d
tm0, tm1,
a_00_10, a_01_11,
x_0_1,
y_0, y_1, y_0_1;
tm0 = _mm256_extractf128_pd( y_00, 0x1 );
tm1 = _mm256_extractf128_pd( y_11, 0x1 );
y_0 = _mm256_castpd256_pd128( y_00 );
y_1 = _mm256_castpd256_pd128( y_11 );
y_0 = _mm_add_pd( y_0, tm0 );
y_1 = _mm_add_pd( y_1, tm1 );
x_0_1 = _mm_loadu_pd( &x[0] );
a_00_10 = _mm_load_sd( &A[0+lda*0] );
a_01_11 = _mm_load_pd( &A[0+lda*1] );
tm0 = _mm_mul_sd( a_00_10, x_0_1 );
tm1 = _mm_mul_pd( a_01_11, x_0_1 );
y_0 = _mm_add_sd( y_0, tm0 );
y_1 = _mm_add_pd( y_1, tm1 );
y_0 = _mm_hadd_pd( y_0, y_1 );
if(alg==0)
{
_mm_storeu_pd(&y[0], y_0);
}
else if(alg==1)
{
y_0_1 = _mm_loadu_pd( &y[0] );
y_0_1 = _mm_add_pd( y_0_1, y_0 );
_mm_storeu_pd(&y[0], y_0_1);
}
else // alg==-1
{
y_0_1 = _mm_loadu_pd( &y[0] );
y_0_1 = _mm_sub_pd( y_0_1, y_0 );
_mm_storeu_pd(&y[0], y_0_1);
}
}
void exchsolutionData_2(unsigned int slot) {
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((!neighbor_isValid[1][0])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S397, S396, S398 */
{
{
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(4.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<3); i1 += 4) {
/* yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<6); i1 += 1) {
yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
{
double* fieldData_Solution_2_p1 = (&fieldData_Solution[2][0]);
int i1 = 1;
for (; (i1<=4); i1 += 2) {
fieldData_Solution_2_p1[((i1*8)+2)] = 0.000000e+00;
fieldData_Solution_2_p1[((i1*8)+10)] = 0.000000e+00;
}
for (; (i1<=5); i1 += 1) {
fieldData_Solution_2_p1[((i1*8)+2)] = 0.000000e+00;
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posBegin[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<3); i1 += 4) {
/* xPos = posBegin[0]; */
__m128d vec0 = _mm_load1_pd((&posBegin[0]));
__m128d vec0_2 = _mm_load1_pd((&posBegin[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<6); i1 += 1) {
xPos = posBegin[0];
}
}
}
}
}
if ((!neighbor_isValid[1][1])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S401, S400, S399 */
{
{
{
double* fieldData_Solution_2_p1 = (&fieldData_Solution[2][0]);
int i1 = 1;
for (; (i1<=4); i1 += 2) {
fieldData_Solution_2_p1[((i1*8)+6)] = 0.000000e+00;
fieldData_Solution_2_p1[((i1*8)+14)] = 0.000000e+00;
}
for (; (i1<=5); i1 += 1) {
fieldData_Solution_2_p1[((i1*8)+6)] = 0.000000e+00;
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
xPos = posEnd[0];
}
__m128d vec1 = _mm_set1_pd(xPos);
for (; (i1<3); i1 += 4) {
/* xPos = posEnd[0]; */
__m128d vec0 = _mm_load1_pd((&posEnd[0]));
__m128d vec0_2 = _mm_load1_pd((&posEnd[0]));
vec1 = vec0;
vec1 = vec0_2;
}
for (; (i1<6); i1 += 1) {
xPos = posEnd[0];
}
}
}
{
int i1 = 1;
for (; (i1<(2&(~1))); i1 += 1) {
yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
__m128d vec1 = _mm_set1_pd(1.000000e+00);
__m128d vec2 = _mm_set1_pd(4.000000e+00);
__m128d vec5 = _mm_set1_pd(yPos);
for (; (i1<3); i1 += 4) {
/* yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]); */
__m128d vec0 = _mm_set_pd(i1+1,i1);
__m128d vec0_2 = _mm_set_pd(i1+1,i1);
__m128d vec3 = _mm_load1_pd((&posEnd[1]));
__m128d vec3_2 = _mm_load1_pd((&posEnd[1]));
__m128d vec4 = _mm_load1_pd((&posBegin[1]));
__m128d vec4_2 = _mm_load1_pd((&posBegin[1]));
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0, vec1), vec2), _mm_sub_pd(vec3, vec4)), vec4);
vec5 = _mm_add_pd(_mm_mul_pd(_mm_div_pd(_mm_sub_pd(vec0_2, vec1), vec2), _mm_sub_pd(vec3_2, vec4_2)), vec4_2);
}
for (; (i1<6); i1 += 1) {
yPos = ((((i1-1)/4.000000e+00)*(posEnd[1]-posBegin[1]))+posBegin[1]);
}
}
}
}
}
if ((!neighbor_isValid[1][2])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S404, S403, S402 */
{
{
{
double* fieldData_Solution_2_p1 = (&fieldData_Solution[2][0]);
int i2 = 2;
for (; (i2<=5); i2 += 2) {
fieldData_Solution_2_p1[(i2+8)] = 0.000000e+00;
fieldData_Solution_2_p1[(i2+9)] = 0.000000e+00;
}
for (; (i2<=6); i2 += 1) {
fieldData_Solution_2_p1[(i2+8)] = 0.000000e+00;
}
}
{
int i2 = 2;
for (; (i2<=5); i2 += 2) {
xPos = ((((i2-2)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=6); i2 += 1) {
xPos = ((((i2-2)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
int i2 = 2;
for (; (i2<=5); i2 += 2) {
yPos = posBegin[1];
yPos = posBegin[1];
}
for (; (i2<=6); i2 += 1) {
yPos = posBegin[1];
}
}
}
}
}
if ((!neighbor_isValid[1][3])) {
{
double xPos;
double yPos;
/* Statements in this Scop: S407, S406, S405 */
{
{
{
double* fieldData_Solution_2_p1 = (&fieldData_Solution[2][0]);
int i2 = 2;
for (; (i2<=5); i2 += 2) {
fieldData_Solution_2_p1[(i2+40)] = 0.000000e+00;
fieldData_Solution_2_p1[(i2+41)] = 0.000000e+00;
}
for (; (i2<=6); i2 += 1) {
fieldData_Solution_2_p1[(i2+40)] = 0.000000e+00;
}
}
{
int i2 = 2;
for (; (i2<=5); i2 += 2) {
xPos = ((((i2-2)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
xPos = ((((i2-1)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
for (; (i2<=6); i2 += 1) {
xPos = ((((i2-2)/4.000000e+00)*(posEnd[0]-posBegin[0]))+posBegin[0]);
}
}
}
{
int i2 = 2;
for (; (i2<=5); i2 += 2) {
yPos = posEnd[1];
yPos = posEnd[1];
}
for (; (i2<=6); i2 += 1) {
yPos = posEnd[1];
}
}
}
}
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Isend(&fieldData_Solution[2][14], 1, mpiDatatype_5_1_8, neighbor_remoteRank[1][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Irecv(&fieldData_Solution[2][10], 1, mpiDatatype_5_1_8, neighbor_remoteRank[1][0], ((unsigned int)(neighbor_fragCommId[1][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Isend(&fieldData_Solution[2][42], 1, mpiDatatype_1_5_8, neighbor_remoteRank[1][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Irecv(&fieldData_Solution[2][10], 1, mpiDatatype_1_5_8, neighbor_remoteRank[1][2], ((unsigned int)(neighbor_fragCommId[1][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Isend(&fieldData_Solution[2][3], 1, mpiDatatype_7_1_8, neighbor_remoteRank[1][0], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][0]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[0]);
reqOutstanding_Send[0] = true;
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Isend(&fieldData_Solution[2][5], 1, mpiDatatype_7_1_8, neighbor_remoteRank[1][1], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][1]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[1]);
reqOutstanding_Send[1] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][0]&&neighbor_isRemote[1][0])) {
MPI_Irecv(&fieldData_Solution[2][1], 1, mpiDatatype_7_1_8, neighbor_remoteRank[1][0], ((unsigned int)(neighbor_fragCommId[1][0]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = true;
}
if ((neighbor_isValid[1][1]&&neighbor_isRemote[1][1])) {
MPI_Irecv(&fieldData_Solution[2][7], 1, mpiDatatype_7_1_8, neighbor_remoteRank[1][1], ((unsigned int)(neighbor_fragCommId[1][1]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[0]) {
waitForMPIReq(&mpiRequest_Recv[0]);
reqOutstanding_Recv[0] = false;
}
if (reqOutstanding_Recv[1]) {
waitForMPIReq(&mpiRequest_Recv[1]);
reqOutstanding_Recv[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[0]) {
waitForMPIReq(&mpiRequest_Send[0]);
reqOutstanding_Send[0] = false;
}
if (reqOutstanding_Send[1]) {
waitForMPIReq(&mpiRequest_Send[1]);
reqOutstanding_Send[1] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Isend(&fieldData_Solution[2][17], 1, mpiDatatype_1_7_8, neighbor_remoteRank[1][2], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][2]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[2]);
reqOutstanding_Send[2] = true;
}
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Isend(&fieldData_Solution[2][33], 1, mpiDatatype_1_7_8, neighbor_remoteRank[1][3], ((unsigned int)commId << 16) + ((unsigned int)(neighbor_fragCommId[1][3]) & 0x0000ffff), mpiCommunicator, &mpiRequest_Send[3]);
reqOutstanding_Send[3] = true;
}
}
}
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if ((neighbor_isValid[1][2]&&neighbor_isRemote[1][2])) {
MPI_Irecv(&fieldData_Solution[2][1], 1, mpiDatatype_1_7_8, neighbor_remoteRank[1][2], ((unsigned int)(neighbor_fragCommId[1][2]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = true;
}
if ((neighbor_isValid[1][3]&&neighbor_isRemote[1][3])) {
MPI_Irecv(&fieldData_Solution[2][49], 1, mpiDatatype_1_7_8, neighbor_remoteRank[1][3], ((unsigned int)(neighbor_fragCommId[1][3]) << 16) + ((unsigned int)commId & 0x0000ffff), mpiCommunicator, &mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = true;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Recv[2]) {
waitForMPIReq(&mpiRequest_Recv[2]);
reqOutstanding_Recv[2] = false;
}
if (reqOutstanding_Recv[3]) {
waitForMPIReq(&mpiRequest_Recv[3]);
reqOutstanding_Recv[3] = false;
}
}
}
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
;
;
}
}
;
;
for (int fragmentIdx = 0; fragmentIdx < 1; ++fragmentIdx) {
if (isValidForSubdomain[1]) {
if (reqOutstanding_Send[2]) {
waitForMPIReq(&mpiRequest_Send[2]);
reqOutstanding_Send[2] = false;
}
if (reqOutstanding_Send[3]) {
waitForMPIReq(&mpiRequest_Send[3]);
reqOutstanding_Send[3] = false;
}
}
}
}
static inline Simd sub(const Simd& lhs, const Simd& rhs) {
Simd res;
res.reg[0] = _mm_sub_pd(lhs.reg[0], rhs.reg[0]);
res.reg[1] = _mm_sub_pd(lhs.reg[1], rhs.reg[1]);
return res;
}