<div dir="ltr">On hold until earlier patch has been fixed.<br><div><div id="__tbSetup"></div></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Nov 13, 2013 at 12:46 PM, <span dir="ltr"><<a href="mailto:murugan@multicorewareinc.com" target="_blank">murugan@multicorewareinc.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"># HG changeset patch<br>
# User Murugan Vairavel <<a href="mailto:murugan@multicorewareinc.com">murugan@multicorewareinc.com</a>><br>
# Date 1384326967 -19800<br>
# Wed Nov 13 12:46:07 2013 +0530<br>
# Node ID 481cdfc251de0f99ef0a3c4fd53c786b79b5f182<br>
# Parent 69d4c1cfc8bed7c63bfdaa1073196e0874d14ebe<br>
pixelsub_ps, Removed unused old code<br>
<br>
diff -r 69d4c1cfc8be -r 481cdfc251de source/common/pixel.cpp<br>
--- a/source/common/pixel.cpp Wed Nov 13 12:31:12 2013 +0530<br>
+++ b/source/common/pixel.cpp Wed Nov 13 12:46:07 2013 +0530<br>
@@ -971,7 +971,6 @@<br>
p.weightpUniPixel = weightUnidirPix;<br>
p.weightpUni = weightUnidir;<br>
<br>
- p.pixelsub_ps = pixelsub_ps_c;<br>
p.pixeladd_ss = pixeladd_ss_c;<br>
<br>
p.scale1D_128to64 = scale1D_128to64;<br>
diff -r 69d4c1cfc8be -r 481cdfc251de source/common/primitives.h<br>
--- a/source/common/primitives.h Wed Nov 13 12:31:12 2013 +0530<br>
+++ b/source/common/primitives.h Wed Nov 13 12:46:07 2013 +0530<br>
@@ -162,7 +162,6 @@<br>
typedef void (*blockcpy_sp_t)(int bx, int by, int16_t *dst, intptr_t dstride, pixel *src, intptr_t sstride); // dst is aligned<br>
typedef void (*blockcpy_ps_t)(int bx, int by, pixel *dst, intptr_t dstride, int16_t *src, intptr_t sstride); // dst is aligned<br>
typedef void (*blockcpy_sc_t)(int bx, int by, int16_t *dst, intptr_t dstride, uint8_t *src, intptr_t sstride); // dst is aligned<br>
-typedef void (*pixelsub_ps_t)(int bx, int by, int16_t *dst, intptr_t dstride, pixel *src0, pixel *src1, intptr_t sstride0, intptr_t sstride1);<br>
typedef void (*pixeladd_ss_t)(int bx, int by, int16_t *dst, intptr_t dstride, int16_t *src0, int16_t *src1, intptr_t sstride0, intptr_t sstride1);<br>
typedef void (*pixelavg_pp_t)(pixel *dst, intptr_t dstride, pixel *src0, intptr_t sstride0, pixel *src1, intptr_t sstride1, int weight);<br>
typedef void (*blockfill_s_t)(int16_t *dst, intptr_t dstride, int16_t val);<br>
@@ -277,7 +276,6 @@<br>
<br>
weightpUni_t weightpUni;<br>
weightpUniPixel_t weightpUniPixel;<br>
- pixelsub_ps_t pixelsub_ps;<br>
pixeladd_ss_t pixeladd_ss;<br>
pixelavg_pp_t pixelavg_pp[NUM_LUMA_PARTITIONS];<br>
<br>
diff -r 69d4c1cfc8be -r 481cdfc251de source/common/vec/blockcopy-sse3.cpp<br>
--- a/source/common/vec/blockcopy-sse3.cpp Wed Nov 13 12:31:12 2013 +0530<br>
+++ b/source/common/vec/blockcopy-sse3.cpp Wed Nov 13 12:46:07 2013 +0530<br>
@@ -170,55 +170,6 @@<br>
}<br>
}<br>
<br>
-void pixelsub_ps(int bx, int by, int16_t *dst, intptr_t dstride, uint8_t *src0, uint8_t *src1, intptr_t sstride0, intptr_t sstride1)<br>
-{<br>
- size_t aligncheck = (size_t)dst | (size_t)src0 | bx | sstride0 | sstride1 | dstride;<br>
-<br>
- if (!(aligncheck & 15))<br>
- {<br>
- // fast path, multiples of 16 pixel wide blocks<br>
- for (int y = 0; y < by; y++)<br>
- {<br>
- for (int x = 0; x < bx; x += 16)<br>
- {<br>
- __m128i word0, word1;<br>
- __m128i word3, word4;<br>
- __m128i mask = _mm_setzero_si128();<br>
-<br>
- word0 = _mm_load_si128((__m128i const*)(src0 + x)); // load 16 bytes from src1<br>
- word1 = _mm_load_si128((__m128i const*)(src1 + x)); // load 16 bytes from src2<br>
-<br>
- word3 = _mm_unpacklo_epi8(word0, mask); // interleave with zero extensions<br>
- word4 = _mm_unpacklo_epi8(word1, mask);<br>
- _mm_store_si128((__m128i*)&dst[x], _mm_subs_epi16(word3, word4)); // store block into dst<br>
-<br>
- word3 = _mm_unpackhi_epi8(word0, mask); // interleave with zero extensions<br>
- word4 = _mm_unpackhi_epi8(word1, mask);<br>
- _mm_store_si128((__m128i*)&dst[x + 8], _mm_subs_epi16(word3, word4)); // store block into dst<br>
- }<br>
-<br>
- src0 += sstride0;<br>
- src1 += sstride1;<br>
- dst += dstride;<br>
- }<br>
- }<br>
- else<br>
- {<br>
- // slow path, irregular memory alignments or sizes<br>
- for (int y = 0; y < by; y++)<br>
- {<br>
- for (int x = 0; x < bx; x++)<br>
- {<br>
- dst[x] = (int16_t)(src0[x] - src1[x]);<br>
- }<br>
-<br>
- src0 += sstride0;<br>
- src1 += sstride1;<br>
- dst += dstride;<br>
- }<br>
- }<br>
-}<br>
-<br>
void pixeladd_ss(int bx, int by, int16_t *dst, intptr_t dstride, int16_t *src0, int16_t *src1, intptr_t sstride0, intptr_t sstride1)<br>
{<br>
size_t aligncheck = (size_t)dst | (size_t)src0 | sstride0 | sstride1 | dstride;<br>
@@ -315,7 +266,6 @@<br>
p.blockcpy_pp = blockcopy_pp;<br>
p.blockcpy_ps = blockcopy_ps;<br>
p.blockcpy_sp = blockcopy_sp;<br>
- p.pixelsub_ps = pixelsub_ps;<br>
p.pixeladd_ss = pixeladd_ss;<br>
#endif // if HIGH_BIT_DEPTH<br>
}<br>
diff -r 69d4c1cfc8be -r 481cdfc251de source/test/pixelharness.cpp<br>
--- a/source/test/pixelharness.cpp Wed Nov 13 12:31:12 2013 +0530<br>
+++ b/source/test/pixelharness.cpp Wed Nov 13 12:46:07 2013 +0530<br>
@@ -358,29 +358,6 @@<br>
return true;<br>
}<br>
<br>
-bool PixelHarness::check_pixelsub_sp(pixelsub_ps_t ref, pixelsub_ps_t opt)<br>
-{<br>
- ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);<br>
- ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);<br>
- int bx = 64;<br>
- int by = 64;<br>
- int j = 0;<br>
- for (int i = 0; i < ITERS; i++)<br>
- {<br>
- opt(bx, by, opt_dest, 64, pbuf2 + j, pbuf1 + j, STRIDE, STRIDE);<br>
- ref(bx, by, ref_dest, 64, pbuf2 + j, pbuf1 + j, STRIDE, STRIDE);<br>
-<br>
- if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))<br>
- return false;<br>
-<br>
- j += INCR;<br>
- bx = 4 * ((rand() & 15) + 1);<br>
- by = 4 * ((rand() & 15) + 1);<br>
- }<br>
-<br>
- return true;<br>
-}<br>
-<br>
bool PixelHarness::check_pixeladd_ss(pixeladd_ss_t ref, pixeladd_ss_t opt)<br>
{<br>
ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);<br>
@@ -886,15 +863,6 @@<br>
}<br>
}<br>
<br>
- if (opt.pixelsub_ps)<br>
- {<br>
- if (!check_pixelsub_sp(ref.pixelsub_ps, opt.pixelsub_ps))<br>
- {<br>
- printf("Luma Substract failed!\n");<br>
- return false;<br>
- }<br>
- }<br>
-<br>
if (opt.pixeladd_ss)<br>
{<br>
if (!check_pixeladd_ss(ref.pixeladd_ss, opt.pixeladd_ss))<br>
@@ -1114,12 +1082,6 @@<br>
REPORT_SPEEDUP(opt.weightpUni, ref.weightpUni, (int16_t*)sbuf1, pbuf1, 64, 64, 32, 32, 128, 1 << 9, 10, 100);<br>
}<br>
<br>
- if (opt.pixelsub_ps)<br>
- {<br>
- printf("Pixel Sub");<br>
- REPORT_SPEEDUP(opt.pixelsub_ps, ref.pixelsub_ps, 64, 64, (int16_t*)pbuf1, FENC_STRIDE, pbuf2, pbuf1, STRIDE, STRIDE);<br>
- }<br>
-<br>
if (opt.pixeladd_ss)<br>
{<br>
printf("pixel_ss add");<br>
diff -r 69d4c1cfc8be -r 481cdfc251de source/test/pixelharness.h<br>
--- a/source/test/pixelharness.h Wed Nov 13 12:31:12 2013 +0530<br>
+++ b/source/test/pixelharness.h Wed Nov 13 12:46:07 2013 +0530<br>
@@ -49,7 +49,6 @@<br>
bool check_calcrecon(calcrecon_t ref, calcrecon_t opt);<br>
bool check_weightpUni(weightpUniPixel_t ref, weightpUniPixel_t opt);<br>
bool check_weightpUni(weightpUni_t ref, weightpUni_t opt);<br>
- bool check_pixelsub_sp(pixelsub_ps_t ref, pixelsub_ps_t opt);<br>
bool check_pixeladd_ss(pixeladd_ss_t ref, pixeladd_ss_t opt);<br>
bool check_downscale_t(downscale_t ref, downscale_t opt);<br>
bool check_cvt32to16_shr_t(cvt32to16_shr_t ref, cvt32to16_shr_t opt);<br>
_______________________________________________<br>
x265-devel mailing list<br>
<a href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a><br>
<a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank">https://mailman.videolan.org/listinfo/x265-devel</a><br>
</blockquote></div><br></div>