static void
avx_test ()
{
int i;
union256d source1, source2;
double s1[4] = {0, -5463, 86456, 0};
double s2[4] = {0, -1223, 0, 1};
int d[1];
int e[1];
source1.x = _mm256_loadu_pd(s1);
source2.x = _mm256_loadu_pd(s2);
d[0] = _mm256_testz_pd(source1.x, source2.x);
e[0] = 1;
for (i = 0; i < 4; i++) {
union ieee754_double u1, u2;
u1.d = s1[i];
u2.d = s2[i];
if (u1.bits.sign && u2.bits.sign) {
e[0] = 0;
}
}
if (checkVi(d, e, 1))
abort ();
}
static void
avx2_test (void)
{
int i;
int ck[8];
int r[8];
unsigned int imm;
int fail = 0;
union256i_q s1, s2, d;
for (i = 0; i < 256; i += 16)
for (imm = 0; imm < 100; imm++)
{
/* Recompute the results for 256-bits */
compute_correct_result_256 (&vals[i + 0], &vals[i + 8], imm, ck);
s1.x = _mm256_loadu_si256 ((__m256i *) & vals[i + 0]);
s2.x = _mm256_loadu_si256 ((__m256i *) & vals[i + 8]);
/* Run the 256-bit tests */
avx2_test_palignr256 (s1.x, s2.x, imm, &d.x);
_mm256_storeu_si256 ((__m256i *) r, d.x);
fail += checkVi (r, ck, 8);
}
if (fail != 0)
abort ();
}
static void
avx_test ()
{
int i;
union128 source1, source2;
float s1[4] = {0, -5463, 86456, 0};
float s2[4] = {0, -1223, 0, 0};
int d[1];
int e[1];
int c=1;
int z=1;
source1.x = _mm_loadu_ps(s1);
source2.x = _mm_loadu_ps(s2);
d[0] = _mm_testnzc_ps(source1.x, source2.x);
e[0] = 1;
for (i = 0; i < 4; i++) {
union ieee754_float u1, u2;
u1.d = s1[i];
u2.d = s2[i];
if (!u1.bits.sign && u2.bits.sign)
c = 0;
if (u1.bits.sign && u2.bits.sign)
z = 0;
}
e[0] = (c == 0 && z == 0) ? 1:0;
if (checkVi (d, e, 1))
abort ();
}
static void
TEST (void)
{
union128 s1, s2;
int d[1];
int e[1];
s1.x = _mm_set_ps (24.43, 68.346, 43.35, 546.46);
s2.x = _mm_set_ps (1.17, 2.16, 3.15, 4.14);
d[0] = test (s1.x, s2.x);
e[0] = s1.a[0] > s2.a[0];
if (checkVi (d, e, 1))
abort ();
}
static void
TEST (void)
{
union128d s1, s2;
int d[1];
int e[1];
s1.x = _mm_set_pd (2134.3343,2344.2354);
s2.x = _mm_set_pd (41124.234,2344.2354);
d[0] = test (s1.x, s2.x);
e[0] = s1.a[0] == s2.a[0];
if (checkVi (d, e, 1))
abort ();
}
void static
avx_test (void)
{
int i;
int d;
union256d s1;
double source[4] = {-45, -3, -34.56, 35};
int e = 0;
s1.x = _mm256_loadu_pd (source);
d = _mm256_movemask_pd (s1.x);
for (i = 0; i < 4; i++)
if (source[i] < 0)
e |= (1 << i);
if (checkVi (&d, &e, 1))
abort ();
}
static void
TEST (void)
{
union128 u;
float s[4] = {-2134.3343, 1234.635654, 1.2234, -876.8976};
int d;
int e = 0;
int i;
u.x = _mm_loadu_ps (s);
d = test (u.x);
for (i = 0; i < 4; i++)
if (s[i] < 0)
e |= (1 << i);
if (checkVi (&d, &e, 1))
abort ();
}
void
TEST (void)
{
int i, sign;
UNION_TYPE (AVX512F_LEN_HALF, i_d) res1, res2, res3;
int res4[SIZE_HALF];
UNION_TYPE (AVX512F_LEN, i_q) src;
MASK_TYPE mask = MASK_VALUE;
int res_ref[SIZE_HALF];
sign = -1;
for (i = 0; i < SIZE; i++)
{
src.a[i] = 1 + 34 * i * sign;
sign = sign * -1;
res2.a[i] = DEFAULT_VALUE;
res4[i] = DEFAULT_VALUE;
}
res1.x = INTRINSIC (_cvtepi64_epi32) (src.x);
res2.x = INTRINSIC (_mask_cvtepi64_epi32) (res2.x, mask, src.x);
res3.x = INTRINSIC (_maskz_cvtepi64_epi32) (mask, src.x);
INTRINSIC (_mask_cvtepi64_storeu_epi32) (res4, mask, src.x);
CALC (res_ref, src.a);
if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res1, res_ref))
abort ();
MASK_MERGE (i_d) (res_ref, mask, SIZE);
if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res2, res_ref))
abort ();
if (checkVi (res4, res_ref, SIZE_HALF))
abort ();
MASK_ZERO (i_d) (res_ref, mask, SIZE);
if (UNION_CHECK (AVX512F_LEN_HALF, i_d) (res3, res_ref))
abort ();
}
void static
avx_test (void)
{
int i;
union256 u, s1, s2;
int source1[8]={34545, 95567, 23443, 5675, 2323, 67, 2345, 45667};
int source2[8]={674, 57897, 93459, 45624, 54674, 1237, 67436, 79608};
int d[8];
int e[8];
s1.x = _mm256_loadu_ps ((float *)source1);
s2.x = _mm256_loadu_ps ((float *)source2);
u.x = _mm256_andnot_ps (s1.x, s2.x);
_mm256_storeu_ps ((float *)d, u.x);
for (i = 0; i < 8; i++)
e[i] = (~source1[i]) & source2[i];
if (checkVi (d, e, 8))
abort ();
}
void check(unsigned imm, char *id)
{
if(checkVi(d.i, e, 4)){
printf("mm_cmp_ps(0x%x, %s) FAILED\n", imm, id);
}
}