mlib_status
__mlib_VideoInterpAveX_U8_U8_8x16(
mlib_u8 *curr_block,
const mlib_u8 *ref_block,
mlib_s32 frame_stride,
mlib_s32 field_stride)
{
mlib_s32 y;
mlib_d64 *dd, ss0[16], *sp1, *sp2, s1hi, s1lo, s2hi, s2lo, s2;
mlib_d64 mthree = vis_fone();
mlib_f32 fzero = vis_fzeros();
mlib_f32 fexpd2 = vis_to_float(0x1000200);
mthree = vis_fpadd16(mthree, vis_fpadd16(mthree, mthree));
dd = (mlib_d64 *)curr_block;
sp1 = (mlib_d64 *)vis_alignaddr((void *)ref_block, 0);
#pragma pipeloop(0)
MLIB_V_VIDEOCOPY8(16);
vis_write_gsr((5 << 3) + ((mlib_s32)(ref_block + 1) & 7));
sp2 = (mlib_d64 *)((mlib_addr)(ref_block + 1) & ~7);
#pragma pipeloop(0)
MLIB_V_VIDEOINTERPAVG8(16);
return (MLIB_SUCCESS);
}
mlib_status
__mlib_VideoInterpX_U8_U8(
mlib_u8 *curr_block,
const mlib_u8 *ref_block,
mlib_s32 width,
mlib_s32 height,
mlib_s32 frame_stride,
mlib_s32 field_stride)
{
mlib_d64 s0, s1, s2, s3, s4, s5, s6, s7, d0, d1, d2, d3, *sd, *dd;
mlib_d64 dzero = vis_fzero();
const mlib_f32 fm1 = vis_to_float(0x100);
mlib_f32 fzero = vis_read_hi(dzero);
mlib_d64 rounder = vis_fone();
mlib_s32 y;
vis_write_gsr((6 << 3) + ((mlib_u32)ref_block & 7));
dd = (mlib_d64 *)curr_block;
sd = (mlib_d64 *)((mlib_addr)ref_block & ~7);
if (width == 8) {
y = height >> 2;
if (((mlib_s32)(ref_block + 1) & 7)) {
do {
s0 = sd[0];
s1 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d0 = vis_faligndata(s0, s1);
s2 = sd[0];
s3 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d1 = vis_faligndata(s2, s3);
s4 = sd[0];
s5 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d2 = vis_faligndata(s4, s5);
s6 = sd[0];
s7 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d3 = vis_faligndata(s6, s7);
vis_alignaddr((void *)(ref_block + 1), 0);
s0 = vis_faligndata(s0, s1);
s1 = vis_faligndata(s2, s3);
s2 = vis_faligndata(s4, s5);
s3 = vis_faligndata(s6, s7);
MLIB_V_VIDEOINTERP(d0, d0, s0);
MLIB_V_VIDEOINTERP(d1, d1, s1);
MLIB_V_VIDEOINTERP(d2, d2, s2);
MLIB_V_VIDEOINTERP4(d3, d3, s3);
*dd = d0;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d1;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d2;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d3;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
vis_alignaddr((void *)ref_block, 0);
} while (--y);
} else {
do {
s0 = sd[0];
s1 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d0 = vis_faligndata(s0, s1);
s2 = sd[0];
s3 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d1 = vis_faligndata(s2, s3);
s4 = sd[0];
s5 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d2 = vis_faligndata(s4, s5);
s6 = sd[0];
s7 = sd[1];
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d3 = vis_faligndata(s6, s7);
MLIB_V_VIDEOINTERP4(d0, d0, s1);
MLIB_V_VIDEOINTERP4(d1, d1, s3);
MLIB_V_VIDEOINTERP4(d2, d2, s5);
MLIB_V_VIDEOINTERP4(d3, d3, s7);
*dd = d0;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d1;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d2;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
*dd = d3;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
} while (--y);
}
} else {
mlib_u8 *dst,
const mlib_u8 *src,
mlib_s32 width,
mlib_s32 height,
mlib_s32 dst_stride,
mlib_s32 src_stride)
{
mlib_s32 x, y, x4 = width >> 2;
mlib_d64 *sl1, *sl2, s1hi, s1lo, s2hi, s2lo, s1, s2;
mlib_d64 done = vis_to_double_dup(0x1000100);
mlib_d64 dmask;
mlib_f32 *dp;
mlib_f32 frnd = vis_to_float(0x40404040);
mlib_s32 src_stride2 = 2 * src_stride;
dmask = vis_fpadd16(done, vis_fone());
vis_write_gsr(7 << 3);
sl1 = (mlib_d64 *)src;
sl2 = (mlib_d64 *)(src + src_stride);
dp = (mlib_f32 *)dst;
for (y = 0; y < height; y++) {
#pragma pipeloop(0)
for (x = 0; x < x4; x++) {
s1 = sl1[x];
s2 = sl2[x];
s1lo = vis_fand(s1, dmask);
s1hi = vis_fmul8sux16(s1, done);
s2lo = vis_fand(s2, dmask);
s2hi = vis_fmul8sux16(s2, done);
s1lo = vis_fpadd16(s1lo, s2lo);
void ADD_SUFF(ThreeByteBgrToIntArgbConvert)(BLIT_PARAMS)
{
mlib_s32 dstScan = pDstInfo->scanStride;
mlib_s32 srcScan = pSrcInfo->scanStride;
mlib_d64 *sp;
mlib_d64 s_0;
mlib_d64 s0, s1, s2, s3, sd0, sd1, sd2, dd0, dd1, dd2, dd3;
mlib_s32 i, i0, j;
if (width < 16) {
for (j = 0; j < height; j++) {
mlib_u8 *src = srcBase;
mlib_s32 *dst = dstBase;
for (i = 0; i < width; i++) {
dst[i] = GBR_PIXEL(i);
}
PTR_ADD(dstBase, dstScan);
PTR_ADD(srcBase, srcScan);
}
return;
}
if (srcScan == 3*width && dstScan == 4*width) {
width *= height;
height = 1;
}
s_0 = vis_fone();
for (j = 0; j < height; j++) {
mlib_u8 *src = srcBase;
mlib_f32 *dst = dstBase;
i = i0 = 0;
if ((mlib_s32)dst & 7) {
((mlib_s32*)dst)[i] = GBR_PIXEL(i);
i0 = 1;
}
sp = vis_alignaddr(src, 3*i0);
s3 = *sp++;
#pragma pipeloop(0)
for (i = i0; i <= (mlib_s32)width - 8; i += 8) {
s0 = s3;
s1 = *sp++;
s2 = *sp++;
s3 = *sp++;
sd0 = vis_faligndata(s0, s1);
sd1 = vis_faligndata(s1, s2);
sd2 = vis_faligndata(s2, s3);
BGR_TO_ARGB
*(mlib_d64*)(dst + i ) = dd0;
*(mlib_d64*)(dst + i + 2) = dd1;
*(mlib_d64*)(dst + i + 4) = dd2;
*(mlib_d64*)(dst + i + 6) = dd3;
}
for (; i < width; i++) {
((mlib_s32*)dst)[i] = GBR_PIXEL(i);
}
PTR_ADD(dstBase, dstScan);
PTR_ADD(srcBase, srcScan);
}
}
void ADD_SUFF(ThreeByteBgrToIntArgbScaleConvert)(SCALE_PARAMS)
{
mlib_s32 dstScan = pDstInfo->scanStride;
mlib_s32 srcScan = pSrcInfo->scanStride;
mlib_d64 dd, maskFF;
mlib_s32 i, i0, i1, j;
if (width < 16) {
for (j = 0; j < height; j++) {
mlib_u8 *src = srcBase;
mlib_s32 *dst = dstBase;
mlib_s32 *dst_end = dst + width;
mlib_s32 tmpsxloc = sxloc;
PTR_ADD(src, (syloc >> shift) * srcScan);
for (; dst < dst_end; dst++) {
i = tmpsxloc >> shift;
tmpsxloc += sxinc;
*(mlib_s32*)dst = GBR_PIXEL(i);
}
PTR_ADD(dstBase, dstScan);
syloc += syinc;
}
return;
}
maskFF = vis_fone();
vis_alignaddr(NULL, 7);
for (j = 0; j < height; j++) {
mlib_u8 *src = srcBase;
mlib_f32 *dst = dstBase;
mlib_f32 *dst_end = dst + width;
mlib_s32 tmpsxloc = sxloc;
PTR_ADD(src, (syloc >> shift) * srcScan);
if ((mlib_s32)dst & 7) {
i = tmpsxloc >> shift;
tmpsxloc += sxinc;
*(mlib_s32*)dst = GBR_PIXEL(i);
dst++;
}
#pragma pipeloop(0)
for (; dst <= dst_end - 2; dst += 2) {
i0 = tmpsxloc >> shift;
i1 = (tmpsxloc + sxinc) >> shift;
tmpsxloc += 2*sxinc;
dd = vis_faligndata(vis_ld_u8(src + 3*i1 ), dd);
dd = vis_faligndata(vis_ld_u8(src + 3*i1 + 1), dd);
dd = vis_faligndata(vis_ld_u8(src + 3*i1 + 2), dd);
dd = vis_faligndata(maskFF, dd);
dd = vis_faligndata(vis_ld_u8(src + 3*i0 ), dd);
dd = vis_faligndata(vis_ld_u8(src + 3*i0 + 1), dd);
dd = vis_faligndata(vis_ld_u8(src + 3*i0 + 2), dd);
dd = vis_faligndata(maskFF, dd);
*(mlib_d64*)dst = dd;
}
for (; dst < dst_end; dst++) {
i = tmpsxloc >> shift;
tmpsxloc += sxinc;
*(mlib_s32*)dst = GBR_PIXEL(i);
}
PTR_ADD(dstBase, dstScan);
syloc += syinc;
}
mlib_status
__mlib_VideoInterpAveX_U8_U8_16x16(
mlib_u8 *curr_block,
const mlib_u8 *ref_block,
mlib_s32 frame_stride,
mlib_s32 field_stride)
{
mlib_d64 s0, s1, s2, s3, s4, s5, s6;
mlib_d64 sd0, sd1, sd2, sd3, d0, d1, d2, d3;
mlib_d64 *sd, *dd;
mlib_d64 dzero = vis_fzero();
const mlib_f32 fm2 = vis_to_float(0x1000200);
mlib_f32 fzero = vis_read_hi(dzero);
mlib_d64 rounder = vis_fpsub16(dzero, vis_fone());
mlib_s32 y;
rounder = vis_fpadd16(vis_fpadd16(rounder, rounder), rounder);
vis_write_gsr((5 << 3) + ((mlib_u32)ref_block & 7));
dd = (mlib_d64 *)curr_block;
sd = (mlib_d64 *)((mlib_addr)ref_block & ~7);
y = 8;
if (((mlib_s32)(ref_block + 1) & 7)) {
do {
s0 = sd[0];
s1 = sd[1];
s2 = sd[2];
sd0 = vis_faligndata(s0, s1);
sd1 = vis_faligndata(s1, s2);
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
s4 = sd[0];
s5 = sd[1];
s6 = sd[2];
sd2 = vis_faligndata(s4, s5);
sd3 = vis_faligndata(s5, s6);
vis_alignaddr((void *)(ref_block + 1), 0);
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d0 = dd[0];
d1 = dd[1];
d2 = ((mlib_d64 *)((mlib_u8 *)dd + field_stride))[0];
d3 = ((mlib_d64 *)((mlib_u8 *)dd + field_stride))[1];
s0 = vis_faligndata(s0, s1);
s1 = vis_faligndata(s1, s2);
s2 = vis_faligndata(s4, s5);
s3 = vis_faligndata(s5, s6);
MLIB_V_VIDEOINTERPAVG(d0, sd0, s0);
MLIB_V_VIDEOINTERPAVG(d1, sd1, s1);
MLIB_V_VIDEOINTERPAVG(d2, sd2, s2);
MLIB_V_VIDEOINTERPAVG(d3, sd3, s3);
dd[0] = d0;
dd[1] = d1;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
dd[0] = d2;
dd[1] = d3;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
vis_alignaddr((void *)ref_block, 0);
} while (--y);
} else {
do {
s0 = sd[0];
s1 = sd[1];
s2 = sd[2];
sd0 = vis_faligndata(s0, s1);
sd1 = vis_faligndata(s1, s2);
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
s4 = sd[0];
s5 = sd[1];
s6 = sd[2];
sd2 = vis_faligndata(s4, s5);
sd3 = vis_faligndata(s5, s6);
sd = (mlib_d64 *)((mlib_u8 *)sd + field_stride);
d0 = dd[0];
d1 = dd[1];
d2 = ((mlib_d64 *)((mlib_u8 *)dd + field_stride))[0];
d3 = ((mlib_d64 *)((mlib_u8 *)dd + field_stride))[1];
MLIB_V_VIDEOINTERPAVG0(d0, sd0, s1);
MLIB_V_VIDEOINTERPAVG(d1, sd1, s2);
MLIB_V_VIDEOINTERPAVG(d2, sd2, s5);
MLIB_V_VIDEOINTERPAVG(d3, sd3, s6);
dd[0] = d0;
dd[1] = d1;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
dd[0] = d2;
dd[1] = d3;
dd = (mlib_d64 *)((mlib_u8 *)dd + field_stride);
} while (--y);
}
return (MLIB_SUCCESS);
}