[x265] [PATCH] asm: 8bpp and 10bpp code for idct8x8
# HG changeset patch
# User Murugan Vairavel muru...@multicorewareinc.com
# Date 1394014787 -19800
# Wed Mar 05 15:49:47 2014 +0530
# Node ID b190855462b4d2a721f52d7a5ae093c4742e3182
# Parent 5cad3652bee8130e8861f5b7b2f6ff1c8695cc1a
asm: 8bpp and 10bpp code for idct8x8
diff -r 5cad3652bee8 -r b190855462b4 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp Tue Mar 04 20:53:17 2014 -0600
+++ b/source/common/x86/asm-primitives.cpp Wed Mar 05 15:49:47 2014 +0530
@@ -909,6 +909,7 @@
SETUP_INTRA_ANG32(34, 2, ssse3);
p.dct[DST_4x4] = x265_dst4_ssse3;
+p.idct[IDCT_8x8] = x265_idct8_ssse3;
}
if (cpuMask X265_CPU_SSE4)
{
@@ -1234,6 +1235,7 @@
p.chroma_p2s[X265_CSP_I444] = x265_luma_p2s_ssse3; // for i444 ,
chroma_p2s can be replaced by luma_p2s
p.dct[DST_4x4] = x265_dst4_ssse3;
+p.idct[IDCT_8x8] = x265_idct8_ssse3;
}
if (cpuMask X265_CPU_SSE4)
{
diff -r 5cad3652bee8 -r b190855462b4 source/common/x86/dct8.asm
--- a/source/common/x86/dct8.asmTue Mar 04 20:53:17 2014 -0600
+++ b/source/common/x86/dct8.asmWed Mar 05 15:49:47 2014 +0530
@@ -61,8 +61,26 @@
times 1 dd 50, -89, 18, 75
times 1 dd 18, -50, 75, -89
+tab_idct8_3:times 4 dw 89, 75
+times 4 dw 50, 18
+times 4 dw 75, -18
+times 4 dw -89, -50
+times 4 dw 50, -89
+times 4 dw 18, 75
+times 4 dw 18, -50
+times 4 dw 75, -89
+
pb_unpackhlw1: db 0,1,8,9,2,3,10,11,4,5,12,13,6,7,14,15
+pb_idct8even: db 0, 1, 8, 9, 4, 5, 12, 13, 0, 1, 8, 9, 4, 5, 12, 13
+
+tab_idct8_1:times 1 dw 64, -64, 36, -83, 64, 64, 83, 36
+
+tab_idct8_2:times 1 dw 89, 75, 50, 18, 75, -18, -89, -50
+times 1 dw 50, -89, 18, 75, 18, -50, 75, -89
+
+pb_idct8odd:db 2, 3, 6, 7, 10, 11, 14, 15, 2, 3, 6, 7, 10, 11, 14, 15
+
SECTION .text
cextern pd_1
cextern pd_2
@@ -665,3 +683,187 @@
dec r2
jnz.pass2
RET
+
+;---
+; void idct8(int32_t *src, int16_t *dst, intptr_t stride)
+;---
+INIT_XMM ssse3
+
+cglobal patial_butterfly_inverse_internal_pass1
+movum0, [r0]
+movum1, [r0 + 4 * 32]
+movum2, [r0 + 2 * 32]
+movum3, [r0 + 6 * 32]
+packssdwm0, m2
+packssdwm1, m3
+punpckhwd m2, m0, m1 ; [2 6]
+punpcklwd m0, m1 ; [0 4]
+pmaddwd m1, m0, [r6]; EE[0]
+pmaddwd m0, [r6 + 32] ; EE[1]
+pmaddwd m3, m2, [r6 + 16] ; EO[0]
+pmaddwd m2, [r6 + 48] ; EO[1]
+
+paddd m4, m1, m3 ; E[0]
+psubd m1, m3 ; E[3]
+paddd m3, m0, m2 ; E[1]
+psubd m0, m2 ; E[2]
+
+;E[K] = E[k] + add
+movam5, [pd_64]
+paddd m0, m5
+paddd m1, m5
+paddd m3, m5
+paddd m4, m5
+
+movum2, [r0 + 32]
+movum5, [r0 + 5 * 32]
+packssdwm2, m5
+movum5, [r0 + 3 * 32]
+movum6, [r0 + 7 * 32]
+packssdwm5, m6
+punpcklwd m6, m2, m5 ;[1 3]
+punpckhwd m2, m5 ;[5 7]
+
+pmaddwd m5, m6, [r4]
+pmaddwd m7, m2, [r4 + 16]
+paddd m5, m7 ; O[0]
+
+paddd m7, m4, m5
+psrad m7, 7
+
+psubd m4, m5
+psrad m4, 7
+
+packssdwm7, m4
+movh[r5 + 0 * 16], m7
+movhps [r5 + 7 * 16], m7
+
+pmaddwd m5, m6, [r4 + 32]
+pmaddwd m4, m2, [r4 + 48]
+paddd m5, m4 ; O[1]
+
+paddd m4, m3, m5
+psrad m4, 7
+
+psubd m3, m5
+psrad m3, 7
+
+packssdwm4, m3
+movh[r5 + 1 * 16], m4
+movhps [r5 + 6 * 16], m4
+
+pmaddwd m5, m6, [r4 + 64]
+pmaddwd m4, m2, [r4 + 80]
+paddd m5, m4 ; O[2]
+
+paddd m4, m0, m5
+psrad m4, 7
+
+psubd m0, m5
+psrad m0, 7
+
+packssdwm4, m0
+movh[r5 + 2 * 16], m4
+movhps [r5 + 5 * 16], m4
+
+pmaddwd m5, m6, [r4 + 96]
+pmaddwd m4, m2, [r4 + 112]
+paddd m5, m4 ; O[3]
+
+paddd m4, m1, m5
+psrad m4, 7
+
+psubd m1, m5
+psrad m1, 7
+
+packssdwm4, m1
+movh[r5 + 3 * 16], m4
+movhps [r5 + 4 * 16], m4
+
+ret
+
+%macro PARTIAL_BUTTERFLY_PROCESS_ROW 1
+%if BIT_DEPTH == 10
+%define IDCT_SHIFT 10
+%elif BIT_DEPTH == 8
+%define IDCT_SHIFT 12
+%else
+%error Unsupported BIT_DEPTH!
+%endif
+pshufb
[x265] [PATCH 1/2] weight prediction: use sqrtf
--- source/encoder/weightPrediction.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/encoder/weightPrediction.cpp b/source/encoder/weightPrediction.cpp index c04361a..c0710fd 100644 --- a/source/encoder/weightPrediction.cpp +++ b/source/encoder/weightPrediction.cpp @@ -203,7 +203,7 @@ bool WeightPrediction::checkDenom(int denom) uint64_t fencVar = fenc-wp_ssd[yuv] + !ref-wp_ssd[yuv]; uint64_t refVar = ref-wp_ssd[yuv] + !ref-wp_ssd[yuv]; if (fencVar refVar) -guessScale[yuv] = Clip3(-2.f, 1.8f, std::sqrt((float)fencVar / refVar)); +guessScale[yuv] = Clip3(-2.f, 1.8f, sqrtf((float)fencVar / refVar)); else guessScale[yuv] = 1.8f; fencMean[yuv] = (float)fenc-wp_sum[yuv] / (height[yuv] * width[yuv]) / (1 (X265_DEPTH - 8)); -- 1.9.0 ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
[x265] [PATCH 2/2] Fix x86_64 build
---
source/CMakeLists.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/source/CMakeLists.txt b/source/CMakeLists.txt
index 2febfaa..d2fea1a 100644
--- a/source/CMakeLists.txt
+++ b/source/CMakeLists.txt
@@ -21,7 +21,7 @@ configure_file(${PROJECT_SOURCE_DIR}/x265_config.h.in
SET(CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake ${CMAKE_MODULE_PATH})
-if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86)
+if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86 OR ${CMAKE_SYSTEM_PROCESSOR} STREQUAL x86_64)
set(X86 1)
add_definitions(-DX265_ARCH_X86=1)
if(${CMAKE_SIZEOF_VOID_P} MATCHES 8)
--
1.9.0
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
[x265] [PATCH] primitives: cleaned up asm_primitives of intra_pred_ang
# HG changeset patch # User Murugan Vairavel muru...@multicorewareinc.com # Date 1394017591 -19800 # Wed Mar 05 16:36:31 2014 +0530 # Node ID 724fb2004447992ea22be5edcf3c8a908118d831 # Parent b190855462b4d2a721f52d7a5ae093c4742e3182 primitives: cleaned up asm_primitives of intra_pred_ang diff -r b190855462b4 -r 724fb2004447 source/common/x86/asm-primitives.cpp --- a/source/common/x86/asm-primitives.cpp Wed Mar 05 15:49:47 2014 +0530 +++ b/source/common/x86/asm-primitives.cpp Wed Mar 05 16:36:31 2014 +0530 @@ -710,14 +710,113 @@ SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 24, cpu); \ SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 32, cpu); \ +#define SETUP_INTRA_ANG_COMMON(mode, fno, cpu) \ +p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu; + +#define SETUP_INTRA_ANG(mode, fno, cpu) \ +p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu; + #define SETUP_INTRA_ANG4(mode, fno, cpu) \ p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; -#define SETUP_INTRA_ANG8(mode, fno, cpu) \ + +#define SETUP_INTRA_ANG16_32(mode, fno, cpu) \ +p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \ +p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu; + +#define SETUP_INTRA_ANG4_8(mode, fno, cpu) \ +p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \ p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; -#define SETUP_INTRA_ANG16(mode, fno, cpu) \ -p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; -#define SETUP_INTRA_ANG32(mode, fno, cpu) \ -p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu; + +#define INTRA_ANG_SSSE3(cpu) \ +SETUP_INTRA_ANG_COMMON(2, 2, cpu); \ +SETUP_INTRA_ANG_COMMON(34, 2, cpu); + +#define INTRA_ANG_SSE4_COMMON(cpu) \ +SETUP_INTRA_ANG_COMMON(3, 3, cpu); \ +SETUP_INTRA_ANG_COMMON(4, 4, cpu); \ +SETUP_INTRA_ANG_COMMON(5, 5, cpu); \ +SETUP_INTRA_ANG_COMMON(6, 6, cpu); \ +SETUP_INTRA_ANG_COMMON(7, 7, cpu); \ +SETUP_INTRA_ANG_COMMON(8, 8, cpu); \ +SETUP_INTRA_ANG_COMMON(9, 9, cpu); \ +SETUP_INTRA_ANG_COMMON(10, 10, cpu); \ +SETUP_INTRA_ANG_COMMON(11, 11, cpu); \ +SETUP_INTRA_ANG_COMMON(12, 12, cpu); \ +SETUP_INTRA_ANG_COMMON(13, 13, cpu); \ +SETUP_INTRA_ANG_COMMON(14, 14, cpu); \ +SETUP_INTRA_ANG_COMMON(15, 15, cpu); \ +SETUP_INTRA_ANG_COMMON(16, 16, cpu); \ +SETUP_INTRA_ANG_COMMON(17, 17, cpu); \ +SETUP_INTRA_ANG_COMMON(18, 18, cpu); + +#define INTRA_ANG_SSE4_HIGH(cpu) \ +SETUP_INTRA_ANG(19, 19, cpu); \ +SETUP_INTRA_ANG(20, 20, cpu); \ +SETUP_INTRA_ANG(21, 21, cpu); \ +SETUP_INTRA_ANG(22, 22, cpu); \ +SETUP_INTRA_ANG(23, 23, cpu); \ +SETUP_INTRA_ANG(24, 24, cpu); \ +SETUP_INTRA_ANG(25, 25, cpu); \ +SETUP_INTRA_ANG(26, 26, cpu); \ +SETUP_INTRA_ANG(27, 27, cpu); \ +SETUP_INTRA_ANG(28, 28, cpu); \ +SETUP_INTRA_ANG(29, 29, cpu); \ +SETUP_INTRA_ANG(30, 30, cpu); \ +SETUP_INTRA_ANG(31, 31, cpu); \ +SETUP_INTRA_ANG(32, 32, cpu); \ +SETUP_INTRA_ANG(33, 33, cpu); \ +SETUP_INTRA_ANG4(19, 17, cpu); \ +SETUP_INTRA_ANG4(20, 16, cpu); \ +SETUP_INTRA_ANG4(21, 15, cpu); \ +SETUP_INTRA_ANG4(22, 14, cpu);\ +SETUP_INTRA_ANG4(23, 13, cpu); \ +SETUP_INTRA_ANG4(24, 12, cpu); \ +SETUP_INTRA_ANG4(25, 11, cpu); \ +SETUP_INTRA_ANG4(26, 26, cpu); \ +SETUP_INTRA_ANG4(27, 9, cpu); \ +SETUP_INTRA_ANG4(28, 8, cpu); \ +SETUP_INTRA_ANG4(29, 7, cpu); \ +SETUP_INTRA_ANG4(30, 6, cpu); \ +SETUP_INTRA_ANG4(31, 5, cpu); \ +SETUP_INTRA_ANG4(32, 4, cpu); \ +SETUP_INTRA_ANG4(33, 3, cpu); + +#define INTRA_ANG_SSE4(cpu) \ +SETUP_INTRA_ANG4_8(19, 17, cpu); \ +SETUP_INTRA_ANG4_8(20, 16, cpu); \ +SETUP_INTRA_ANG4_8(21, 15, cpu); \ +SETUP_INTRA_ANG4_8(22, 14, cpu);\ +SETUP_INTRA_ANG4_8(23, 13, cpu); \ +SETUP_INTRA_ANG4_8(24, 12, cpu); \ +SETUP_INTRA_ANG4_8(25, 11, cpu); \ +SETUP_INTRA_ANG4_8(26, 26, cpu); \ +SETUP_INTRA_ANG4_8(27, 9, cpu); \ +SETUP_INTRA_ANG4_8(28, 8, cpu); \ +SETUP_INTRA_ANG4_8(29, 7, cpu); \ +SETUP_INTRA_ANG4_8(30, 6, cpu); \ +SETUP_INTRA_ANG4_8(31, 5, cpu); \ +SETUP_INTRA_ANG4_8(32, 4, cpu); \ +SETUP_INTRA_ANG4_8(33, 3, cpu); \ +SETUP_INTRA_ANG16_32(19, 19, cpu); \ +SETUP_INTRA_ANG16_32(20, 20, cpu); \ +SETUP_INTRA_ANG16_32(21, 21, cpu); \ +SETUP_INTRA_ANG16_32(22, 22, cpu); \ +
Re: [x265] [PATCH 2/2] Fix x86_64 build
Can you convert this to an hg patch? Git patches don't apply cleanly
On Wed, Mar 5, 2014 at 4:08 PM, Rafaël Carré fun...@videolan.org wrote:
---
source/CMakeLists.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/source/CMakeLists.txt b/source/CMakeLists.txt
index 2febfaa..d2fea1a 100644
--- a/source/CMakeLists.txt
+++ b/source/CMakeLists.txt
@@ -21,7 +21,7 @@ configure_file(${PROJECT_SOURCE_DIR}/x265_config.h.in
SET(CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake
${CMAKE_MODULE_PATH})
-if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86)
+if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86 OR ${CMAKE_SYSTEM_PROCESSOR} STREQUAL x86_64)
set(X86 1)
add_definitions(-DX265_ARCH_X86=1)
if(${CMAKE_SIZEOF_VOID_P} MATCHES 8)
--
1.9.0
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] [PATCH 2/2] Fix x86_64 build
I know, I sent patches to mercurial to properly apply git patches with utf8,
but never finished them :/
Can you e.g. apply the diff manually and fix my name after?
I'm using git hg clone for all my development so making a hg patch is
not easy for me.
If you have a suggestion I'm all ears though
On 03/05/14 12:15, Deepthi Nandakumar wrote:
Can you convert this to an hg patch? Git patches don't apply cleanly
On Wed, Mar 5, 2014 at 4:08 PM, Rafaël Carré fun...@videolan.org wrote:
---
source/CMakeLists.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/source/CMakeLists.txt b/source/CMakeLists.txt
index 2febfaa..d2fea1a 100644
--- a/source/CMakeLists.txt
+++ b/source/CMakeLists.txt
@@ -21,7 +21,7 @@ configure_file(${PROJECT_SOURCE_DIR}/x265_config.h.in
SET(CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake
${CMAKE_MODULE_PATH})
-if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86)
+if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86 OR ${CMAKE_SYSTEM_PROCESSOR} STREQUAL x86_64)
set(X86 1)
add_definitions(-DX265_ARCH_X86=1)
if(${CMAKE_SIZEOF_VOID_P} MATCHES 8)
--
1.9.0
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
[x265] Issue #35: Crash when encoding yuv420p10le/yuv420p10be video using ffmpeg (multicoreware/x265)
New issue 35: Crash when encoding yuv420p10le/yuv420p10be video using ffmpeg
https://bitbucket.org/multicoreware/x265/issue/35/crash-when-encoding-yuv420p10le
sfan5:
Segfaults immediately(before ffmpeg outputs the frame=... status text) when
using yuv420p10le.
Segfaults after encoding 25 frames when using yuv420p10be.
I'm not sure whether this is an ffmpeg bug or libx265 bug.
libx265 version: 31731a78d994
ffmpeg version: b2880a3de9e3
Backtrace:
```
#!none
Program received signal SIGSEGV, Segmentation fault.
0x723b63b8 in x265::TComPicYuv::copyFromPicture (this=0xa285220,
pic=..., pad=0x66c870) at
/tmp/build/src/libx265/source/Lib/TLibCommon/TComPicYuv.cpp:291
291 V[c] = (pixel)((v[c] shift) mask);
(gdb) bt full
#0 0x723b62b5 in x265::TComPicYuv::copyFromPicture (this=0xa278c20,
pic=..., pad=0x66c850)
at /tmp/build/src/libx265/source/Lib/TLibCommon/TComPicYuv.cpp:279
c = 832
r = 982
Y = 0x7fffdc518400
U = 0x130800e0
y = 0x7fffd904e980
u = 0x7fffd8d20a70
mask = 1023
V = 0x131df120
v = 0x7fffd8e2c090
shift = 0
pady = 9
padx = 1
width = 1920
height = 1080
rem = 8 '\b'
#1 0x723046e4 in x265::Encoder::encode (this=0x66bf30, flush=false,
pic_in=0x7fffd470, pic_out=0x7fffd4c0, nalunits=0x7fffd410)
at /tmp/build/src/libx265/source/encoder/encoder.cpp:323
pic = 0x117ad820
curEncoder = 0x66f500
ret = 0
out = 0x74dd74d0 __GI___libc_malloc+96
#2 0x7230aaaf in x265_encoder_encode (enc=0x66bf30,
pp_nal=0x7fffd468, pi_nal=0x7fffd464, pic_in=0x7fffd470,
pic_out=0x7fffd4c0)
at /tmp/build/src/libx265/source/encoder/api.cpp:105
encoder = 0x66bf30
nalunits = {0x0, 0x0, 0x0, 0x0, 0x0}
numEncoded = 119
#3 0x767a558f in ?? ()
from /home/stefan/Dokumente/ffmpeg-git/lib/libavcodec.so.55
No symbol table info available.
#4 0x76906672 in avcodec_encode_video2 ()
from /home/stefan/Dokumente/ffmpeg-git/lib/libavcodec.so.55
No symbol table info available.
#5 0x0041e95e in ?? ()
No symbol table info available.
#6 0x00408117 in ?? ()
No symbol table info available.
#7 0x74d75de5 in __libc_start_main (main=0x406ec0, argc=12,
ubp_av=0x7fffdea8, init=optimized out, fini=optimized out,
rtld_fini=optimized out, stack_end=0x7fffde98) at libc-start.c:260
result = optimized out
unwind_buf = {cancel_jmp_buf = {{jmp_buf = {0, -7112016030735285240,
4234400, 140737488346784, 0, 0, 7112016031736120328,
7111996038656410632}, mask_was_saved = 0}}, priv = {pad = {
0x0, 0x0, 0x4244c0, 0x7fffdea8}, data = {prev = 0x0,
cleanup = 0x0, canceltype = 4342976}}}
not_first_call = optimized out
#8 0x00409cc9 in ?? ()
No symbol table info available.
```
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
[x265] Issue #34: Cannot encode a raw YUV video (multicoreware/x265)
New issue 34: Cannot encode a raw YUV video https://bitbucket.org/multicoreware/x265/issue/34/cannot-encode-a-raw-yuv-video Jacques Boscq: Hello, I tried to encode a raw YUV file into a HEVC raw file but it crashed: ``` #!c++ yuv [info]: 1920x1080 fps 23980/1000 i420 frames 0 - 130425 of 130426 [New Thread 0x75f1a700 (LWP 31281)] x265 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX x265 [info]: HEVC encoder version 0.7+375-6662df480e39 x265 [info]: build info [Linux][GCC 4.8.2][64 bit] 8bpp x265 [info]: Main profile, Level-4 (Main tier) [New Thread 0x75719700 (LWP 31282)] [New Thread 0x74f18700 (LWP 31283)] [New Thread 0x74717700 (LWP 31284)] [New Thread 0x73f16700 (LWP 31285)] x265 [info]: WPP streams / pool / frames : 17 / 4 / 2 x265 [info]: CU size : 64 x265 [info]: Max RQT depth inter / intra : 1 / 1 x265 [info]: ME / range / subpel / merge : hex / 57 / 2 / 2 x265 [info]: Keyframe min / max / scenecut : 23 / 250 / 40 x265 [info]: Lookahead / bframes / badapt: 20 / 4 / 2 x265 [info]: b-pyramid / weightp / refs : 1 / 1 / 3 x265 [info]: Rate Control / AQ-Strength / CUTree : CRF-28.0 / 1.0 / 1 x265 [info]: tools: rect amp rd=3 lft sao-lcu sign-hide [New Thread 0x72099700 (LWP 31286)] [New Thread 0x700f2700 (LWP 31287)] [4.5%] 5912/130426 frames, 2.14 fps, 849.07 kb/s, eta 16:09:46 Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x74717700 (LWP 31284)] 0x00622e3e in x265::BitCost::bitcost (this=0x264f7e8, mv=...) at /home/da/x265/source/encoder/bitcost.h:52 52s_bitsizes[(abs(mv.y - m_mvp.y) 1) + !!(mv.y m_mvp.y)] + 0.5f); (gdb) bt #0 0x00622e3e in x265::BitCost::bitcost (this=0x264f7e8, mv=...) at /home/da/x265/source/encoder/bitcost.h:52 #1 0x006196b7 in x265::TEncSearch::predInterSearch (this=0x264f658, cu=0x3fe9eb0, predYuv=0x3970d10, bUseMRG=false, bLuma=true, bChroma=false) at /home/da/x265/source/Lib/TLibEncoder/TEncSearch.cpp:2707 #2 0x0062f6f2 in x265::TEncCu::xComputeCostInter (this=0x264fb58, outTempCU=0x3fe9eb0, outPredYuv=0x3970d10, partSize=x265::SIZE_Nx2N, bUseMRG=false) at /home/da/x265/source/encoder/compress.cpp:205 #3 0x00630ff8 in x265::TEncCu::xCompressInterCU (this=0x264fb58, outBestCU=@0x74716cb8: 0x3fd41f0, outTempCU=@0x1dbb190: 0x3fbe530, cu=@0x74716ca0: 0x2924aba0, depth=0, PartitionIndex=0, minDepth=0 '\000') at /home/da/x265/source/encoder/compress.cpp:447 #4 0x005f29a1 in x265::TEncCu::compressCU (this=0x264fb58, cu=0x2924aba0) at /home/da/x265/source/Lib/TLibEncoder/TEncCu.cpp:392 #5 0x0051e0e2 in x265::CTURow::processCU (this=0x264f178, cu=0x2924aba0, slice=0x27f94c90, bufferSbac=0x0, bSaveSBac=false) at /home/da/x265/source/encoder/cturow.cpp:82 #6 0x00519121 in x265::FrameEncoder::processRowEncoder ( this=0x8cc788, row=14) at /home/da/x265/source/encoder/frameencoder.cpp:1086 #7 0x0051b46b in x265::FrameEncoder::processRow (this=0x8cc788, row=28) at /home/da/x265/source/encoder/frameencoder.h:101 #8 0x005edad0 in x265::WaveFront::findJob (this=0x8cc788) at /home/da/x265/source/common/wavefront.cpp:126 #9 0x0050ac2c in x265::PoolThread::threadMain (this=0x8c8600) at /home/da/x265/source/common/threadpool.cpp:159 #10 0x00496003 in x265::ThreadShim (opaque=0x8c8600) at /home/da/x265/source/common/threading.cpp:73 #11 0x77bc6e0e in start_thread (arg=0x74717700) at pthread_create.c:311 #12 0x76edc0fd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:113 (gdb) p (abs(mv.y - m_mvp.y) 1) + !!(mv.y m_mvp.y) $1 = 41217 (gdb) p s_bitsizes[41217] Cannot access memory at address 0x77fff414 (gdb) p s_bitsizes[3] $2 = 31.4634457 (gdb) p s_bitsizes[4] $3 = -1.02578287e+34 (gdb) p (abs(mv.x - m_mvp.x) 1) $4 = 5376 ``` Regards ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] [PATCH] asm: added 10bpp primitives for chroma_vpp and chroma_vps filters
At 2014-03-05 21:20:49,muru...@multicorewareinc.com wrote: # HG changeset patch # User Murugan Vairavel muru...@multicorewareinc.com # Date 1394025626 -19800 # Wed Mar 05 18:50:26 2014 +0530 # Node ID 820704b966e9c0fa3537a5789c92d897435ae304 # Parent 724fb2004447992ea22be5edcf3c8a908118d831 asm: added 10bpp primitives for chroma_vpp and chroma_vps filters no asm code in patch, so comment use 'Enable' is better. ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] [PATCH 2/2] Fix x86_64 build
On 05 Mar 2014, at 11:38, Rafaël Carré fun...@videolan.org wrote: --- source/CMakeLists.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) What was the issue fixed by this (build failure on x86_64, but caused by what)? ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] [PATCH 2/2] Fix x86_64 build
On Wed, Mar 5, 2014 at 4:38 AM, Rafaël Carré fun...@videolan.org wrote:
---
source/CMakeLists.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/source/CMakeLists.txt b/source/CMakeLists.txt
index 2febfaa..d2fea1a 100644
--- a/source/CMakeLists.txt
+++ b/source/CMakeLists.txt
@@ -21,7 +21,7 @@ configure_file(${PROJECT_SOURCE_DIR}/x265_config.h.in
SET(CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake ${CMAKE_MODULE_PATH})
-if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86)
+if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL i386 OR ${CMAKE_SYSTEM_PROCESSOR}
STREQUAL x86 OR ${CMAKE_SYSTEM_PROCESSOR} STREQUAL x86_64)
set(X86 1)
add_definitions(-DX265_ARCH_X86=1)
if(${CMAKE_SIZEOF_VOID_P} MATCHES 8)
I suspect there is something wrong with your git clone. This line now
looks like:
string(TOLOWER ${CMAKE_SYSTEM_PROCESSOR} SYSPROC)
if(${SYSPROC} STREQUAL i386 OR ${SYSPROC} STREQUAL amd64 OR
${SYSPROC} STREQUAL x86_64 OR ${SYSPROC} STREQUAL x86 OR
${SYSPROC} STREQUAL )
message(STATUS Detected x86 system processor)
The line you have is over a month old.
--
Steve Borho
___
x265-devel mailing list
x265-devel@videolan.org
https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] A new screenshots comparator and loads of new x264 x265 vp8 vp9 tests
On Tue, Mar 4, 2014 at 8:04 PM, Niccolò Belli darkba...@linuxsystems.it wrote: http://www.linuxsystems.it/2014/03/new-screenshots-comparator-loads-new-x264-x265-vp8-vp9-tests/ Congratulations to the x265 team: at low bitrates x265 completely CRUSH x264 if you use a very high quality source like the Blu-ray of The Hobbit: An Unexpected Journey. Hopefully in the future it will improve at high bitrates too. hi10p is just too bugged to be useful right now. Hi Niccolò, I'd like to understand the steps you took to encode 10bit video. I think it should show much better, particularly with the 0.8 tag or later. The two basic preconditions x265 needs is to build with HIGH_BIT_DEPTH enabled and then provide it input pixels that are at least 10bits of resolution and in YUV 4:2:0 or 4:4:4 format (4:4:4 is less well tested, so I suggest 4:2:0 for eval). A 10bit build of x265 will happily encode an 8bit input video, it just shifts up all the pixels by two bits. Not necessarily a good demonstration of 10bit encode, but it is useful for comparison purposes. Similarly, an 8bit build of x265 will happily encode a 10bit video if you ask it to (by downshifting). Not necessarily useful except for comparison. -- Steve Borho ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
[x265] [PATCH] api: change meaning of pic.stride to be in bytes rather than pixels (fixes #35)
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1394045238 21600
# Wed Mar 05 12:47:18 2014 -0600
# Node ID 59f0664b3d9060a56cd64187cb5e9d1d6b173f09
# Parent 6d55869ed5e29c97977b89aa0218a5c8510c671a
api: change meaning of pic.stride to be in bytes rather than pixels (fixes #35)
x264's pic.plane pointer is a uint8_t* so their input strides are byte based,
ffmpeg is currently assuming our input strides are byte based. This commit
will make that assumption correct.
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/Lib/TLibCommon/TComPicYuv.cpp
--- a/source/Lib/TLibCommon/TComPicYuv.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/Lib/TLibCommon/TComPicYuv.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -205,7 +205,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(pixel);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -218,8 +218,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
else if (pic.bitDepth == 8)
@@ -240,7 +240,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(*y);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -253,8 +253,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
else /* pic.bitDepth 8 */
@@ -280,7 +280,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(*y);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -293,8 +293,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/encoder/encoder.cpp
--- a/source/encoder/encoder.cppWed Mar 05 11:48:14 2014 -0600
+++ b/source/encoder/encoder.cppWed Mar 05 12:47:18 2014 -0600
@@ -390,11 +390,11 @@
}
pic_out-planes[0] = recpic-getLumaAddr();
-pic_out-stride[0] = recpic-getStride();
+pic_out-stride[0] = recpic-getStride() * sizeof(pixel);
pic_out-planes[1] = recpic-getCbAddr();
-pic_out-stride[1] = recpic-getCStride();
+pic_out-stride[1] = recpic-getCStride() * sizeof(pixel);
pic_out-planes[2] = recpic-getCrAddr();
-pic_out-stride[2] = recpic-getCStride();
+pic_out-stride[2] = recpic-getCStride() * sizeof(pixel);
}
if (out-getSlice()-m_numWPRefs 0)
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/input/yuv.cpp
--- a/source/input/yuv.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/input/yuv.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -219,11 +219,11 @@
pic.colorSpace = colorSpace;
pic.bitDepth = depth;
+pic.stride[0] = width * pixelbytes;
+pic.stride[1] = pic.stride[2] = pic.stride[0] 1;
pic.planes[0] = buf[head];
-pic.planes[1] = (char*)(pic.planes[0]) + width * height * pixelbytes;
-pic.planes[2] = (char*)(pic.planes[1]) + ((width * height * pixelbytes)
2);
-pic.stride[0] = width;
-pic.stride[1] = pic.stride[2] = pic.stride[0] 1;
+pic.planes[1] = (char*)(pic.planes[0]) + height * pic.stride[0];
+pic.planes[2] = (char*)(pic.planes[1]) + ((height * pic.stride[1]) 2);
head = (head + 1) % QUEUE_SIZE;
notFull.trigger();
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/output/y4m.cpp
--- a/source/output/y4m.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/output/y4m.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -93,18 +93,19 @@
}
ofs.write(buf, width x265_cli_csps[colorSpace].width[i]);
-src += pic.stride[i];
+src += pic.stride[i] / sizeof(*src);
}
}
#else // if HIGH_BIT_DEPTH
+
for (int i = 0; i x265_cli_csps[colorSpace].planes; i++)
{
char *src = (char*)pic.planes[i];
for (int h = 0; h height x265_cli_csps[colorSpace].height[i]; h++)
{
ofs.write(src, width x265_cli_csps[colorSpace].width[i]);
-src += pic.stride[i];
+src += pic.stride[i] / sizeof(*src);
}
}
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/output/yuv.cpp
--- a/source/output/yuv.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/output/yuv.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -73,7 +73,7 @@
}
ofs.write(buf, width
Re: [x265] [PATCH] api: change meaning of pic.stride to be in bytes rather than pixels (fixes #35)
On Wed, Mar 5, 2014 at 12:47 PM, Steve Borho st...@borho.org wrote:
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1394045238 21600
# Wed Mar 05 12:47:18 2014 -0600
# Node ID 59f0664b3d9060a56cd64187cb5e9d1d6b173f09
# Parent 6d55869ed5e29c97977b89aa0218a5c8510c671a
api: change meaning of pic.stride to be in bytes rather than pixels (fixes
#35)
x264's pic.plane pointer is a uint8_t* so their input strides are byte based,
ffmpeg is currently assuming our input strides are byte based. This commit
will make that assumption correct.
If no one objects, I'll push this with a bump to X265_BUILD
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/Lib/TLibCommon/TComPicYuv.cpp
--- a/source/Lib/TLibCommon/TComPicYuv.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/Lib/TLibCommon/TComPicYuv.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -205,7 +205,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(pixel);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -218,8 +218,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
else if (pic.bitDepth == 8)
@@ -240,7 +240,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(*y);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -253,8 +253,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
else /* pic.bitDepth 8 */
@@ -280,7 +280,7 @@
}
Y += getStride();
-y += pic.stride[0];
+y += pic.stride[0] / sizeof(*y);
}
for (int r = 0; r height m_vChromaShift; r++)
@@ -293,8 +293,8 @@
U += getCStride();
V += getCStride();
-u += pic.stride[1];
-v += pic.stride[2];
+u += pic.stride[1] / sizeof(*u);
+v += pic.stride[2] / sizeof(*v);
}
}
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/encoder/encoder.cpp
--- a/source/encoder/encoder.cppWed Mar 05 11:48:14 2014 -0600
+++ b/source/encoder/encoder.cppWed Mar 05 12:47:18 2014 -0600
@@ -390,11 +390,11 @@
}
pic_out-planes[0] = recpic-getLumaAddr();
-pic_out-stride[0] = recpic-getStride();
+pic_out-stride[0] = recpic-getStride() * sizeof(pixel);
pic_out-planes[1] = recpic-getCbAddr();
-pic_out-stride[1] = recpic-getCStride();
+pic_out-stride[1] = recpic-getCStride() * sizeof(pixel);
pic_out-planes[2] = recpic-getCrAddr();
-pic_out-stride[2] = recpic-getCStride();
+pic_out-stride[2] = recpic-getCStride() * sizeof(pixel);
}
if (out-getSlice()-m_numWPRefs 0)
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/input/yuv.cpp
--- a/source/input/yuv.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/input/yuv.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -219,11 +219,11 @@
pic.colorSpace = colorSpace;
pic.bitDepth = depth;
+pic.stride[0] = width * pixelbytes;
+pic.stride[1] = pic.stride[2] = pic.stride[0] 1;
pic.planes[0] = buf[head];
-pic.planes[1] = (char*)(pic.planes[0]) + width * height * pixelbytes;
-pic.planes[2] = (char*)(pic.planes[1]) + ((width * height * pixelbytes)
2);
-pic.stride[0] = width;
-pic.stride[1] = pic.stride[2] = pic.stride[0] 1;
+pic.planes[1] = (char*)(pic.planes[0]) + height * pic.stride[0];
+pic.planes[2] = (char*)(pic.planes[1]) + ((height * pic.stride[1]) 2);
head = (head + 1) % QUEUE_SIZE;
notFull.trigger();
diff -r 6d55869ed5e2 -r 59f0664b3d90 source/output/y4m.cpp
--- a/source/output/y4m.cpp Wed Mar 05 11:48:14 2014 -0600
+++ b/source/output/y4m.cpp Wed Mar 05 12:47:18 2014 -0600
@@ -93,18 +93,19 @@
}
ofs.write(buf, width x265_cli_csps[colorSpace].width[i]);
-src += pic.stride[i];
+src += pic.stride[i] / sizeof(*src);
}
}
#else // if HIGH_BIT_DEPTH
+
for (int i = 0; i x265_cli_csps[colorSpace].planes; i++)
{
char *src = (char*)pic.planes[i];
for (int h = 0; h height x265_cli_csps[colorSpace].height[i];
h++)
{
ofs.write(src, width x265_cli_csps[colorSpace].width[i]);
-src += pic.stride[i];
+src += pic.stride[i] / sizeof(*src);
}
}
diff -r 6d55869ed5e2 -r 59f0664b3d90
Re: [x265] new link
On Wed, Mar 5, 2014 at 12:32 PM, Roger Pack rogerdpa...@gmail.com wrote: Hello. As a note, on this page: http://x265.org/developers.html possibly you could/should add a link to http://www.videolan.org/developers/x265.html Oh right, our videolan page is finally live. We should do that. On a side note, the Encoder features on the videolan.org page appears to be a direct copy of the x264 list, which is, uhm, wishful thinking in some spots and just wrong in others. I suggest this list: Full prediction and transform quad-tree recursion supported Adaptive B-frame placement B-frames as references / arbitrary frame order CABAC entropy coding Intra: all block types (32x32, 16x16, 8x8, 4x4, and PCM with all predictions) Inter P: all partitions (from 64x64 down to 8x4) Inter B: partitions from 64x64 down to 8x4 (including all merge modes and biprediction) Weighted prediction for P slices Multiple reference frames Ratecontrol: constant quantizer, constant quality, single pass ABR, optional VBV Scenecut detection Parallel encoding on multiple CPUs, both frame-level and wavefront parallelism -- Steve Borho ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] A new screenshots comparator and loads of new x264 x265 vp8 vp9 tests
In data mercoledì 5 marzo 2014 11:38:41, Steve Borho ha scritto: I'd like to understand the steps you took to encode 10bit video I simply misunderstood the 16bpp flag: it doesn't enable high bit depth but only 16 bit variables. Niccolò -- www.linuxsystems.it ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] A new screenshots comparator and loads of new x264 x265 vp8 vp9 tests
On Wed, Mar 5, 2014 at 2:51 PM, Niccolò Belli darkba...@linuxsystems.it wrote: In data mercoledì 5 marzo 2014 11:38:41, Steve Borho ha scritto: I'd like to understand the steps you took to encode 10bit video I simply misunderstood the 16bpp flag: it doesn't enable high bit depth but only 16 bit variables. When you compile with HIGH_BIT_DEPTH, the encoder will encode 10bit streams, just as x264 does. There was an impedance mismatch between x264 and ffmpeg which was preventing 10bit inputs from working properly. I sent a patch to the ML for it this morning. I will push it later today if no one objects. -- Steve Borho ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
Re: [x265] cleanup m_sharedPredTransformSkip[]
On Tue, Mar 4, 2014 at 4:40 AM, Satoshi Nakagawa nakagawa...@oki.com wrote:
# HG changeset patch
# User Satoshi Nakagawa nakagawa...@oki.com
# Date 1393929339 -32400
# Tue Mar 04 19:35:39 2014 +0900
# Node ID 7a61566806f691ddff84cbbc42801f6c2d46df88
# Parent 3cbde0b893e34e5770cc311d3f4b6fe064c27774
cleanup m_sharedPredTransformSkip[]
NEW_CALCRECON macro is TODO mark for asm experts, to optimize register
assignment.
Sorry I haven't responded to this yet; I would like Min to review it
before I push it.
diff -r 3cbde0b893e3 -r 7a61566806f6 source/Lib/TLibEncoder/TEncSearch.cpp
--- a/source/Lib/TLibEncoder/TEncSearch.cpp Mon Mar 03 13:37:35 2014 -0600
+++ b/source/Lib/TLibEncoder/TEncSearch.cpp Tue Mar 04 19:35:39 2014 +0900
@@ -63,7 +63,6 @@
m_qtTempTUCoeffCr = NULL;
for (int i = 0; i 3; i++)
{
-m_sharedPredTransformSkip[i] = NULL;
m_qtTempTransformSkipFlag[i] = NULL;
m_qtTempCbf[i] = NULL;
}
@@ -96,7 +95,6 @@
for (uint32_t i = 0; i 3; ++i)
{
X265_FREE(m_qtTempCbf[i]);
-X265_FREE(m_sharedPredTransformSkip[i]);
X265_FREE(m_qtTempTransformSkipFlag[i]);
}
@@ -153,9 +151,6 @@
CHECKED_MALLOC(m_qtTempTransformSkipFlag[1], uint8_t, numPartitions);
CHECKED_MALLOC(m_qtTempTransformSkipFlag[2], uint8_t, numPartitions);
-CHECKED_MALLOC(m_sharedPredTransformSkip[0], pixel, MAX_TS_WIDTH *
MAX_TS_HEIGHT);
-CHECKED_MALLOC(m_sharedPredTransformSkip[1], pixel, MAX_TS_WIDTH *
MAX_TS_HEIGHT);
-CHECKED_MALLOC(m_sharedPredTransformSkip[2], pixel, MAX_TS_WIDTH *
MAX_TS_HEIGHT);
CHECKED_MALLOC(m_qtTempTUCoeffY, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
CHECKED_MALLOC(m_qtTempTUCoeffCb, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
CHECKED_MALLOC(m_qtTempTUCoeffCr, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
@@ -414,7 +409,6 @@
Pel* fenc = fencYuv-getLumaAddr(absPartIdx);
Pel* pred = predYuv-getLumaAddr(absPartIdx);
int16_t* residual = resiYuv-getLumaAddr(absPartIdx);
-Pel* recon= predYuv-getLumaAddr(absPartIdx);
int chFmt= cu-getChromaFormat();
int part = partitionFromSizes(width, height);
@@ -439,15 +433,6 @@
cu-getPattern()-initAdiPattern(cu, absPartIdx, trDepth, m_predBuf,
m_predBufStride, m_predBufHeight, m_refAbove, m_refLeft, m_refAboveFlt,
m_refLeftFlt);
//= get prediction signal =
predIntraLumaAng(lumaPredMode, pred, stride, width);
-// save prediction
-if (default0Save1Load2 == 1)
-{
-primitives.luma_copy_pp[part](m_sharedPredTransformSkip[0],
width, pred, stride);
-}
-}
-else
-{
-primitives.luma_copy_pp[part](pred, stride,
m_sharedPredTransformSkip[0], width);
}
//= get residual signal =
@@ -491,12 +476,19 @@
primitives.blockfill_s[size](resiTmp, stride, 0);
}
+assert(width = 32);
+#if NEW_CALCRECON
//= reconstruction =
-assert(width = 32);
+primitives.calcrecon[size](pred, residual, 0, reconQt, reconIPred,
stride, MAX_CU_SIZE, reconIPredStride);
+//= update distortion =
+outDist += primitives.sse_sp[part](reconQt, MAX_CU_SIZE, fenc, stride);
+#else
+ALIGN_VAR_32(pixel, recon[MAX_CU_SIZE * MAX_CU_SIZE]);
+//= reconstruction =
primitives.calcrecon[size](pred, residual, recon, reconQt, reconIPred,
stride, MAX_CU_SIZE, reconIPredStride);
-
//= update distortion =
outDist += primitives.sse_pp[part](fenc, stride, recon, stride);
+#endif
}
void TEncSearch::xIntraCodingChromaBlk(TComDataCU* cu,
@@ -534,7 +526,6 @@
Pel* fenc = (chromaId 0 ? fencYuv-getCrAddr(absPartIdx)
: fencYuv-getCbAddr(absPartIdx));
Pel* pred = (chromaId 0 ? predYuv-getCrAddr(absPartIdx)
: predYuv-getCbAddr(absPartIdx));
int16_t* residual = (chromaId 0 ? resiYuv-getCrAddr(absPartIdx)
: resiYuv-getCbAddr(absPartIdx));
-Pel* recon = (chromaId 0 ? predYuv-getCrAddr(absPartIdx)
: predYuv-getCbAddr(absPartIdx));
uint32_t qtlayer=
cu-getSlice()-getSPS()-getQuadtreeTULog2MaxSize() - trSizeLog2;
uint32_t numCoeffPerInc = (cu-getSlice()-getSPS()-getMaxCUWidth() *
cu-getSlice()-getSPS()-getMaxCUHeight()
(cu-getSlice()-getSPS()-getMaxCUDepth() 1)) (m_hChromaShift +
m_vChromaShift);
@@ -561,19 +552,6 @@
//= get prediction signal =
predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, width,
height, chFmt);
-
-// save prediction
-if (default0Save1Load2 == 1)
-{
-Pel* predbuf = m_sharedPredTransformSkip[1 + chromaId];
-primitives.luma_copy_pp[part](predbuf, width, pred, stride);
-}
-}
-else
-{
-// load
[x265] Unfixed typo in patch 6408 (889edfd2c4c3): VBV occupancy
Just caught this while reading over latest changes: Missing u was not recognized during clarify attempt in source/common/param.cpp -- __ Fun and success! Mario *LigH* Rohkrämer mailto:cont...@ligh.de ___ x265-devel mailing list x265-devel@videolan.org https://mailman.videolan.org/listinfo/x265-devel
