void
mlib_v_ImageLookUp_U8_U8_124_SrcOff0_D1(
const mlib_u8 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 *table0,
const mlib_u8 *table1,
const mlib_u8 *table2,
const mlib_u8 *table3)
{
/* aligned pointer to source data */
mlib_u32 *sa;
/* pointer to source data */
mlib_u8 *sp;
/* source data */
mlib_u32 s0, s1;
/* pointer to start of destination */
mlib_u8 *dl;
/* pointer to end of destination */
mlib_u8 *dend;
/* aligned pointer to destination */
mlib_d64 *dp;
/* destination data */
mlib_d64 t0, t1, t2;
/* destination data */
mlib_d64 t3, t4, t5;
/* destination data */
mlib_d64 t6, t7, acc0, acc1;
/* edge mask */
mlib_s32 emask;
/* loop variable */
mlib_s32 i, num;
sa = (mlib_u32 *)src;
dl = dst;
dp = (mlib_d64 *)dl;
dend = dl + xsize - 1;
vis_alignaddr((void *)0, 7);
if (xsize >= 8) {
s0 = sa[0];
s1 = sa[1];
sa += 2;
vis_write_bmask(0x012389ab, 0);
#pragma pipeloop(0)
for (i = 0; i <= xsize - 16; i += 8, sa += 2) {
t7 = VIS_LD_U8_I(table3, s1 & 0xFF);
t6 = VIS_LD_U8_I(table2, (s1 >> 8) & 0xFF);
t5 = VIS_LD_U8_I(table1, (s1 >> 16) & 0xFF);
t4 = VIS_LD_U8_I(table0, s1 >> 24);
t3 = VIS_LD_U8_I(table3, s0 & 0xFF);
t2 = VIS_LD_U8_I(table2, (s0 >> 8) & 0xFF);
t1 = VIS_LD_U8_I(table1, (s0 >> 16) & 0xFF);
t0 = VIS_LD_U8_I(table0, s0 >> 24);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
s0 = sa[0];
s1 = sa[1];
(*dp++) = vis_bshuffle(acc0, acc1);
}
t7 = VIS_LD_U8_I(table3, s1 & 0xFF);
t6 = VIS_LD_U8_I(table2, (s1 >> 8) & 0xFF);
t5 = VIS_LD_U8_I(table1, (s1 >> 16) & 0xFF);
t4 = VIS_LD_U8_I(table0, s1 >> 24);
t3 = VIS_LD_U8_I(table3, s0 & 0xFF);
t2 = VIS_LD_U8_I(table2, (s0 >> 8) & 0xFF);
t1 = VIS_LD_U8_I(table1, (s0 >> 16) & 0xFF);
t0 = VIS_LD_U8_I(table0, s0 >> 24);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
(*dp++) = vis_bshuffle(acc0, acc1);
}
void mlib_v_ImageLookUpSI_U16_U8_2_DstA8D1(const mlib_u16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 **table)
{
mlib_u16 *sp; /* pointer to source data */
mlib_s32 s0, s1, s2, s3; /* source data */
mlib_u16 *dl; /* pointer to start of destination */
mlib_u16 *dend; /* pointer to end of destination */
mlib_d64 *dp; /* aligned pointer to destination */
mlib_d64 t0, t1, t2; /* destination data */
mlib_d64 t3, t4, t5; /* destination data */
mlib_d64 t6, t7, acc; /* destination data */
mlib_s32 emask; /* edge mask */
mlib_s32 i, num; /* loop variable */
const mlib_u8 *tab0 = &table[0][0];
const mlib_u8 *tab1 = &table[1][0];
sp = (void *)src;
dl = (mlib_u16 *) dst;
dp = (mlib_d64 *) dl;
dend = dl + xsize - 1;
vis_alignaddr((void *)0, 7);
if (xsize >= 4) {
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
sp += 4;
#pragma pipeloop(0)
for (i = 0; i <= xsize - 8; i += 4, sp += 4) {
t7 = VIS_LD_U8_I(tab1, s3);
t6 = VIS_LD_U8_I(tab0, s3);
t5 = VIS_LD_U8_I(tab1, s2);
t4 = VIS_LD_U8_I(tab0, s2);
t3 = VIS_LD_U8_I(tab1, s1);
t2 = VIS_LD_U8_I(tab0, s1);
t1 = VIS_LD_U8_I(tab1, s0);
t0 = VIS_LD_U8_I(tab0, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
*dp++ = acc;
}
t7 = VIS_LD_U8_I(tab1, s3);
t6 = VIS_LD_U8_I(tab0, s3);
t5 = VIS_LD_U8_I(tab1, s2);
t4 = VIS_LD_U8_I(tab0, s2);
t3 = VIS_LD_U8_I(tab1, s1);
t2 = VIS_LD_U8_I(tab0, s1);
t1 = VIS_LD_U8_I(tab1, s0);
t0 = VIS_LD_U8_I(tab0, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
*dp++ = acc;
}
if ((mlib_addr) dp <= (mlib_addr) dend) {
num = (mlib_u16 *) dend - (mlib_u16 *) dp;
sp += num;
num++;
#pragma pipeloop(0)
for (i = 0; i < num; i++) {
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(tab1, s0);
acc = vis_faligndata(t0, acc);
t0 = VIS_LD_U8_I(tab0, s0);
acc = vis_faligndata(t0, acc);
}
emask = vis_edge16(dp, dend);
vis_pst_16(acc, dp, emask);
}
}
void mlib_v_ImageLookUpSI_U16_U8_4_DstOff3_D1(const mlib_u16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 **table)
{
mlib_u16 *sp; /* pointer to source data */
mlib_s32 s0, s1, s2; /* source data */
mlib_u8 *dl; /* pointer to start of destination */
mlib_d64 *dp; /* aligned pointer to destination */
mlib_d64 t0, t1, t2; /* destination data */
mlib_d64 t3, t4, t5; /* destination data */
mlib_d64 t6, t7, acc; /* destination data */
mlib_s32 i; /* loop variable */
const mlib_u8 *tab0 = &table[0][0];
const mlib_u8 *tab1 = &table[1][0];
const mlib_u8 *tab2 = &table[2][0];
const mlib_u8 *tab3 = &table[3][0];
sp = (void *)src;
dl = dst;
dp = (mlib_d64 *) dl;
vis_alignaddr((void *)0, 7);
s0 = *sp++;
if (xsize >= 2) {
s1 = sp[0];
s2 = sp[1];
sp += 2;
#pragma pipeloop(0)
for (i = 0; i <= xsize - 4; i += 2, sp += 2) {
t7 = VIS_LD_U8_I(tab2, s2);
t6 = VIS_LD_U8_I(tab1, s2);
t5 = VIS_LD_U8_I(tab0, s2);
t4 = VIS_LD_U8_I(tab3, s1);
t3 = VIS_LD_U8_I(tab2, s1);
t2 = VIS_LD_U8_I(tab1, s1);
t1 = VIS_LD_U8_I(tab0, s1);
t0 = VIS_LD_U8_I(tab3, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
s0 = s2;
s1 = sp[0];
s2 = sp[1];
*dp++ = acc;
}
t7 = VIS_LD_U8_I(tab2, s2);
t6 = VIS_LD_U8_I(tab1, s2);
t5 = VIS_LD_U8_I(tab0, s2);
t4 = VIS_LD_U8_I(tab3, s1);
t3 = VIS_LD_U8_I(tab2, s1);
t2 = VIS_LD_U8_I(tab1, s1);
t1 = VIS_LD_U8_I(tab0, s1);
t0 = VIS_LD_U8_I(tab3, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
s0 = s2;
*dp++ = acc;
}
dl = (mlib_u8 *) dp;
if ((xsize & 1) != 0) {
s1 = sp[0];
t7 = VIS_LD_U8_I(tab2, s1);
t6 = VIS_LD_U8_I(tab1, s1);
t5 = VIS_LD_U8_I(tab0, s1);
t4 = VIS_LD_U8_I(tab3, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
*(mlib_f32 *) dl = vis_read_hi(acc);
dl += 4;
s0 = s1;
}
dl[0] = tab3[s0];
}
void mlib_v_ImageLookUpSI_U16_U8_3_D1(const mlib_u16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 **table)
{
mlib_u16 *sp; /* pointer to source data */
mlib_u8 *dl; /* pointer to start of destination */
mlib_d64 *dp; /* aligned pointer to destination */
mlib_d64 t0, t1, t2; /* destination data */
mlib_d64 t3, t4, t5; /* destination data */
mlib_d64 t6, t7; /* destination data */
mlib_d64 acc0, acc1, acc2; /* destination data */
mlib_s32 i; /* loop variable */
const mlib_u8 *tab0 = &table[0][0];
const mlib_u8 *tab1 = &table[1][0];
const mlib_u8 *tab2 = &table[2][0];
mlib_s32 s00, s01, s02, s03;
mlib_s32 s10, s11, s12, s13;
sp = (void *)src;
dl = dst;
dp = (mlib_d64 *) dl;
vis_alignaddr((void *)0, 7);
i = 0;
if (xsize >= 8) {
s00 = sp[0];
s01 = sp[1];
s02 = sp[2];
s03 = sp[3];
s10 = sp[4];
s11 = sp[5];
s12 = sp[6];
s13 = sp[7];
sp += 8;
#pragma pipeloop(0)
for (i = 0; i <= xsize - 16; i += 8, sp += 8) {
t7 = VIS_LD_U8_I(tab1, s02);
t6 = VIS_LD_U8_I(tab0, s02);
t5 = VIS_LD_U8_I(tab2, s01);
t4 = VIS_LD_U8_I(tab1, s01);
t3 = VIS_LD_U8_I(tab0, s01);
t2 = VIS_LD_U8_I(tab2, s00);
t1 = VIS_LD_U8_I(tab1, s00);
t0 = VIS_LD_U8_I(tab0, s00);
acc0 = vis_faligndata(t7, acc0);
acc0 = vis_faligndata(t6, acc0);
acc0 = vis_faligndata(t5, acc0);
acc0 = vis_faligndata(t4, acc0);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
t7 = VIS_LD_U8_I(tab0, s11);
t6 = VIS_LD_U8_I(tab2, s10);
t5 = VIS_LD_U8_I(tab1, s10);
t4 = VIS_LD_U8_I(tab0, s10);
t3 = VIS_LD_U8_I(tab2, s03);
t2 = VIS_LD_U8_I(tab1, s03);
t1 = VIS_LD_U8_I(tab0, s03);
t0 = VIS_LD_U8_I(tab2, s02);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc1 = vis_faligndata(t3, acc1);
acc1 = vis_faligndata(t2, acc1);
acc1 = vis_faligndata(t1, acc1);
acc1 = vis_faligndata(t0, acc1);
t7 = VIS_LD_U8_I(tab2, s13);
t6 = VIS_LD_U8_I(tab1, s13);
t5 = VIS_LD_U8_I(tab0, s13);
t4 = VIS_LD_U8_I(tab2, s12);
t3 = VIS_LD_U8_I(tab1, s12);
t2 = VIS_LD_U8_I(tab0, s12);
t1 = VIS_LD_U8_I(tab2, s11);
t0 = VIS_LD_U8_I(tab1, s11);
acc2 = vis_faligndata(t7, acc2);
acc2 = vis_faligndata(t6, acc2);
acc2 = vis_faligndata(t5, acc2);
acc2 = vis_faligndata(t4, acc2);
acc2 = vis_faligndata(t3, acc2);
acc2 = vis_faligndata(t2, acc2);
acc2 = vis_faligndata(t1, acc2);
acc2 = vis_faligndata(t0, acc2);
s00 = sp[0];
s01 = sp[1];
s02 = sp[2];
s03 = sp[3];
s10 = sp[4];
s11 = sp[5];
s12 = sp[6];
s13 = sp[7];
*dp++ = acc0;
*dp++ = acc1;
*dp++ = acc2;
}
t7 = VIS_LD_U8_I(tab1, s02);
t6 = VIS_LD_U8_I(tab0, s02);
t5 = VIS_LD_U8_I(tab2, s01);
t4 = VIS_LD_U8_I(tab1, s01);
t3 = VIS_LD_U8_I(tab0, s01);
t2 = VIS_LD_U8_I(tab2, s00);
t1 = VIS_LD_U8_I(tab1, s00);
t0 = VIS_LD_U8_I(tab0, s00);
acc0 = vis_faligndata(t7, acc0);
acc0 = vis_faligndata(t6, acc0);
acc0 = vis_faligndata(t5, acc0);
acc0 = vis_faligndata(t4, acc0);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
t7 = VIS_LD_U8_I(tab0, s11);
t6 = VIS_LD_U8_I(tab2, s10);
t5 = VIS_LD_U8_I(tab1, s10);
t4 = VIS_LD_U8_I(tab0, s10);
t3 = VIS_LD_U8_I(tab2, s03);
t2 = VIS_LD_U8_I(tab1, s03);
t1 = VIS_LD_U8_I(tab0, s03);
t0 = VIS_LD_U8_I(tab2, s02);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc1 = vis_faligndata(t3, acc1);
acc1 = vis_faligndata(t2, acc1);
acc1 = vis_faligndata(t1, acc1);
acc1 = vis_faligndata(t0, acc1);
t7 = VIS_LD_U8_I(tab2, s13);
t6 = VIS_LD_U8_I(tab1, s13);
t5 = VIS_LD_U8_I(tab0, s13);
t4 = VIS_LD_U8_I(tab2, s12);
t3 = VIS_LD_U8_I(tab1, s12);
t2 = VIS_LD_U8_I(tab0, s12);
t1 = VIS_LD_U8_I(tab2, s11);
t0 = VIS_LD_U8_I(tab1, s11);
acc2 = vis_faligndata(t7, acc2);
acc2 = vis_faligndata(t6, acc2);
acc2 = vis_faligndata(t5, acc2);
acc2 = vis_faligndata(t4, acc2);
acc2 = vis_faligndata(t3, acc2);
acc2 = vis_faligndata(t2, acc2);
acc2 = vis_faligndata(t1, acc2);
acc2 = vis_faligndata(t0, acc2);
*dp++ = acc0;
*dp++ = acc1;
*dp++ = acc2;
i += 8;
}
dl = (mlib_u8 *) dp;
#pragma pipeloop(0)
for (; i < xsize; i++) {
s00 = sp[0];
dl[0] = tab0[s00];
dl[1] = tab1[s00];
dl[2] = tab2[s00];
dl += 3;
sp++;
}
}
void
mlib_v_ImageLookUp_S16_U8_124_D1(
const mlib_s16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 *table0,
const mlib_u8 *table1,
const mlib_u8 *table2,
const mlib_u8 *table3)
{
/* pointer to source data */
mlib_s16 *sp;
/* source data */
mlib_s32 s0, s1, s2, s3;
/* source data */
mlib_s32 s4, s5, s6, s7;
/* pointer to start of destination */
mlib_u8 *dl;
/* pointer to end of destination */
mlib_u8 *dend;
/* aligned pointer to destination */
mlib_d64 *dp;
/* destination data */
mlib_d64 t0, t1, t2;
/* destination data */
mlib_d64 t3, t4, t5;
/* destination data */
mlib_d64 t6, t7, acc0;
/* edge mask */
mlib_s32 emask;
/* loop variable */
mlib_s32 i, num;
/* destination data */
mlib_d64 acc1;
dl = dst;
dp = (mlib_d64 *)dl;
dend = dl + xsize - 1;
sp = (void *)src;
vis_alignaddr((void *)0, 7);
if (xsize >= 8) {
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
s4 = sp[4];
s5 = sp[5];
s6 = sp[6];
s7 = sp[7];
sp += 8;
vis_write_bmask(0x012389ab, 0);
#pragma pipeloop(0)
for (i = 0; i <= xsize - 16; i += 8, sp += 8) {
t7 = VIS_LD_U8_I(table3, s7);
t6 = VIS_LD_U8_I(table2, s6);
t5 = VIS_LD_U8_I(table1, s5);
t4 = VIS_LD_U8_I(table0, s4);
t3 = VIS_LD_U8_I(table3, s3);
t2 = VIS_LD_U8_I(table2, s2);
t1 = VIS_LD_U8_I(table1, s1);
t0 = VIS_LD_U8_I(table0, s0);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
s4 = sp[4];
s5 = sp[5];
s6 = sp[6];
s7 = sp[7];
(*dp++) = vis_bshuffle(acc0, acc1);
}
t7 = VIS_LD_U8_I(table3, s7);
t6 = VIS_LD_U8_I(table2, s6);
t5 = VIS_LD_U8_I(table1, s5);
t4 = VIS_LD_U8_I(table0, s4);
t3 = VIS_LD_U8_I(table3, s3);
t2 = VIS_LD_U8_I(table2, s2);
t1 = VIS_LD_U8_I(table1, s1);
t0 = VIS_LD_U8_I(table0, s0);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
(*dp++) = vis_bshuffle(acc0, acc1);
}
if ((mlib_addr)dp <= (mlib_addr)dend) {
num = (mlib_addr)dend - (mlib_addr)dp;
sp += num;
num++;
if ((num & 3) == 1) {
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc0 = vis_faligndata(t0, acc0);
num--;
} else if ((num & 3) == 2) {
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table1, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc0 = vis_faligndata(t0, acc0);
num -= 2;
} else if ((num & 3) == 3) {
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table2, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table1, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc0 = vis_faligndata(t0, acc0);
num -= 3;
}
if (num != 0) {
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table3, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table2, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table1, s0);
acc0 = vis_faligndata(t0, acc0);
s0 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc0 = vis_faligndata(t0, acc0);
}
emask = vis_edge8(dp, dend);
vis_pst_8(acc0, dp, emask);
}
}
void mlib_v_ImageLookUp_S16_U8_3_D1(const mlib_s16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 *table0,
const mlib_u8 *table1,
const mlib_u8 *table2)
{
mlib_s16 *sp; /* pointer to source data */
mlib_s32 s0, s1, s2, s3; /* source data */
mlib_s32 s4, s5, s6, s7; /* source data */
mlib_u8 *dl; /* pointer to start of destination */
mlib_u8 *dend; /* pointer to end of destination */
mlib_d64 *dp; /* aligned pointer to destination */
mlib_d64 t0, t1, t2; /* destination data */
mlib_d64 t3, t4, t5; /* destination data */
mlib_d64 t6, t7, acc; /* destination data */
mlib_s32 emask; /* edge mask */
mlib_s32 i, num; /* loop variable */
const mlib_u8 *table;
dl = dst;
sp = (void *)src;
dp = (mlib_d64 *) dl;
dend = dl + xsize - 1;
vis_alignaddr((void *)0, 7);
if (xsize >= 8) {
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
s4 = sp[4];
s5 = sp[5];
s6 = sp[6];
s7 = sp[7];
sp += 8;
#pragma pipeloop(0)
for (i = 0; i <= xsize - 16; i += 8, sp += 8) {
t7 = VIS_LD_U8_I(table1, s7);
t6 = VIS_LD_U8_I(table0, s6);
t5 = VIS_LD_U8_I(table2, s5);
t4 = VIS_LD_U8_I(table1, s4);
t3 = VIS_LD_U8_I(table0, s3);
t2 = VIS_LD_U8_I(table2, s2);
t1 = VIS_LD_U8_I(table1, s1);
t0 = VIS_LD_U8_I(table0, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
table = table0;
table0 = table2;
table2 = table1;
table1 = table;
s0 = sp[0];
s1 = sp[1];
s2 = sp[2];
s3 = sp[3];
s4 = sp[4];
s5 = sp[5];
s6 = sp[6];
s7 = sp[7];
*dp++ = acc;
}
t7 = VIS_LD_U8_I(table1, s7);
t6 = VIS_LD_U8_I(table0, s6);
t5 = VIS_LD_U8_I(table2, s5);
t4 = VIS_LD_U8_I(table1, s4);
t3 = VIS_LD_U8_I(table0, s3);
t2 = VIS_LD_U8_I(table2, s2);
t1 = VIS_LD_U8_I(table1, s1);
t0 = VIS_LD_U8_I(table0, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
table = table0;
table0 = table2;
table2 = table1;
table1 = table;
*dp++ = acc;
}
if ((mlib_addr) dp <= (mlib_addr) dend) {
num = (mlib_addr) dend - (mlib_addr) dp;
sp += num;
num++;
i = num - 3 * (num / 3);
if (i == 2) {
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table1, s0);
acc = vis_faligndata(t0, acc);
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc = vis_faligndata(t0, acc);
num -= 2;
}
else if (i == 1) {
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc = vis_faligndata(t0, acc);
num--;
}
#pragma pipeloop(0)
for (i = 0; i < num; i += 3) {
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table2, s0);
acc = vis_faligndata(t0, acc);
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table1, s0);
acc = vis_faligndata(t0, acc);
s0 = (mlib_s32) * sp;
sp--;
t0 = VIS_LD_U8_I(table0, s0);
acc = vis_faligndata(t0, acc);
}
emask = vis_edge8(dp, dend);
vis_pst_8(acc, dp, emask);
}
}
void
mlib_v_ImageLookUpSI_S16_U8_2_D1(
const mlib_s16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 **table)
{
/* pointer to source data */
mlib_s16 *sp;
/* source data */
mlib_s32 s0, s1, s2, s3, s4;
/* pointer to start of destination */
mlib_u8 *dl;
/* pointer to end of destination */
mlib_u8 *dend;
/* aligned pointer to destination */
mlib_d64 *dp;
/* destination data */
mlib_d64 t0, t1, t2;
/* destination data */
mlib_d64 t3, t4, t5;
/* destination data */
mlib_d64 t6, t7, acc;
/* edge mask */
mlib_s32 emask;
/* loop variable */
mlib_s32 i, num;
const mlib_u8 *tab0 = &table[0][32768];
const mlib_u8 *tab1 = &table[1][32768];
sp = (void *)src;
dl = dst;
dend = dl + 2 * xsize - 1;
vis_alignaddr((void *)0, 7);
s0 = (*sp++);
(*dl++) = tab0[s0];
dp = (mlib_d64 *)dl;
xsize--;
if (xsize >= 4) {
s1 = sp[0];
s2 = sp[1];
s3 = sp[2];
s4 = sp[3];
sp += 4;
#pragma pipeloop(0)
for (i = 0; i <= xsize - 8; i += 4, sp += 4) {
t7 = VIS_LD_U8_I(tab0, s4);
t6 = VIS_LD_U8_I(tab1, s3);
t5 = VIS_LD_U8_I(tab0, s3);
t4 = VIS_LD_U8_I(tab1, s2);
t3 = VIS_LD_U8_I(tab0, s2);
t2 = VIS_LD_U8_I(tab1, s1);
t1 = VIS_LD_U8_I(tab0, s1);
t0 = VIS_LD_U8_I(tab1, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
s0 = s4;
s1 = sp[0];
s2 = sp[1];
s3 = sp[2];
s4 = sp[3];
(*dp++) = acc;
}
t7 = VIS_LD_U8_I(tab0, s4);
t6 = VIS_LD_U8_I(tab1, s3);
t5 = VIS_LD_U8_I(tab0, s3);
t4 = VIS_LD_U8_I(tab1, s2);
t3 = VIS_LD_U8_I(tab0, s2);
t2 = VIS_LD_U8_I(tab1, s1);
t1 = VIS_LD_U8_I(tab0, s1);
t0 = VIS_LD_U8_I(tab1, s0);
acc = vis_faligndata(t7, acc);
acc = vis_faligndata(t6, acc);
acc = vis_faligndata(t5, acc);
acc = vis_faligndata(t4, acc);
acc = vis_faligndata(t3, acc);
acc = vis_faligndata(t2, acc);
acc = vis_faligndata(t1, acc);
acc = vis_faligndata(t0, acc);
s0 = s4;
(*dp++) = acc;
}
num = ((mlib_u8 *)dend - (mlib_u8 *)dp) >> 1;
sp += (num - 1);
for (i = 0; i < num; i++) {
s1 = (mlib_s32)*sp;
sp--;
t0 = VIS_LD_U8_I(tab1, s1);
acc = vis_faligndata(t0, acc);
t0 = VIS_LD_U8_I(tab0, s1);
acc = vis_faligndata(t0, acc);
}
t0 = VIS_LD_U8_I(tab1, s0);
acc = vis_faligndata(t0, acc);
emask = vis_edge8(dp, dend);
vis_pst_8(acc, dp, emask);
}
void
mlib_v_ImageLookUpSI_S16_U8_4_DstOff2_D1(
const mlib_s16 *src,
mlib_u8 *dst,
mlib_s32 xsize,
const mlib_u8 **table)
{
/* pointer to source data */
mlib_s16 *sp;
/* source data */
mlib_s32 s0, s1, s2;
/* pointer to start of destination */
mlib_u8 *dl;
/* aligned pointer to destination */
mlib_d64 *dp;
/* destination data */
mlib_d64 t0, t1, t2;
/* destination data */
mlib_d64 t3, t4, t5;
/* destination data */
mlib_d64 t6, t7, acc0;
/* loop variable */
mlib_s32 i;
const mlib_u8 *tab0 = &table[0][32768];
const mlib_u8 *tab1 = &table[1][32768];
const mlib_u8 *tab2 = &table[2][32768];
const mlib_u8 *tab3 = &table[3][32768];
mlib_d64 acc1;
sp = (void *)src;
dl = dst;
dp = (mlib_d64 *)dl;
vis_alignaddr((void *)0, 7);
s0 = (*sp++);
if (xsize >= 2) {
s1 = sp[0];
s2 = sp[1];
sp += 2;
vis_write_bmask(0x012389ab, 0);
#pragma pipeloop(0)
for (i = 0; i <= xsize - 4; i += 2, sp += 2) {
t7 = VIS_LD_U8_I(tab1, s2);
t6 = VIS_LD_U8_I(tab0, s2);
t5 = VIS_LD_U8_I(tab3, s1);
t4 = VIS_LD_U8_I(tab2, s1);
t3 = VIS_LD_U8_I(tab1, s1);
t2 = VIS_LD_U8_I(tab0, s1);
t1 = VIS_LD_U8_I(tab3, s0);
t0 = VIS_LD_U8_I(tab2, s0);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
s0 = s2;
s1 = sp[0];
s2 = sp[1];
(*dp++) = vis_bshuffle(acc0, acc1);
}
t7 = VIS_LD_U8_I(tab1, s2);
t6 = VIS_LD_U8_I(tab0, s2);
t5 = VIS_LD_U8_I(tab3, s1);
t4 = VIS_LD_U8_I(tab2, s1);
t3 = VIS_LD_U8_I(tab1, s1);
t2 = VIS_LD_U8_I(tab0, s1);
t1 = VIS_LD_U8_I(tab3, s0);
t0 = VIS_LD_U8_I(tab2, s0);
acc1 = vis_faligndata(t7, acc1);
acc1 = vis_faligndata(t6, acc1);
acc1 = vis_faligndata(t5, acc1);
acc1 = vis_faligndata(t4, acc1);
acc0 = vis_faligndata(t3, acc0);
acc0 = vis_faligndata(t2, acc0);
acc0 = vis_faligndata(t1, acc0);
acc0 = vis_faligndata(t0, acc0);
s0 = s2;
(*dp++) = vis_bshuffle(acc0, acc1);
}
dl = (mlib_u8 *)dp;
if ((xsize & 1) != 0) {
s1 = sp[0];
t7 = VIS_LD_U8_I(tab1, s1);
t6 = VIS_LD_U8_I(tab0, s1);
t5 = VIS_LD_U8_I(tab3, s0);
t4 = VIS_LD_U8_I(tab2, s0);
acc0 = vis_faligndata(t7, acc0);
acc0 = vis_faligndata(t6, acc0);
acc0 = vis_faligndata(t5, acc0);
acc0 = vis_faligndata(t4, acc0);
*(mlib_f32 *)dl = vis_read_hi(acc0);
dl += 4;
s0 = s1;
}
dl[0] = tab2[s0];
dl[1] = tab3[s0];
}
