Okay. Will check IACA report and try pxor for m0 and buffer 1023.
On Mon, Jun 22, 2015 at 8:24 PM, chen chenm...@163.com wrote:
right
some comment:
'psignb X, [pb_128]' equal to 'psubb X, 0, X', in AVX2, second type
faster, in SSE4, choice depends on IACA report
in PMINSW, you buffer ZERO
right
some comment:
'psignb X, [pb_128]' equal to 'psubb X, 0, X', in AVX2, second type faster, in
SSE4, choice depends on IACA report
in PMINSW, you buffer ZERO into M0, and use pw_1023 directly, could you try
buffer pw_1023 and use PXOR to get ZERO?
At 2015-06-22
# HG changeset patch
# User Min Chen
# Date 1434999426 18000
# Mon Jun 22 13:57:06 2015 -0500
# Node ID 2cdab8d3b76066b18f32d1aa13c17fe50f9fa289
# Parent 83a7d824442455ba5e0a6b53ea68e6b7043845de
cmake: introduce multilib support for MSVC
Note, the multilib configuration exposes a bug in
# HG changeset patch
# User Min Chen chenm...@163.com
# Date 1435019994 25200
# Node ID e3c31f11936b5e915ec773200e4c1b1d8db2730f
# Parent 965b507acd52ad89160723253b770ef0036c71a5
asm: improve AVX2 sad_x4[32xN] by new faster algorithm
Old:
sad_x4[32x32] 41.91x 2379.45 99724.21
# HG changeset patch
# User Min Chen chenm...@163.com
# Date 1435019988 25200
# Node ID 64167d1ad6d81c6c2d2ab762592d131860825fe9
# Parent b3915bc1febdeb620ecbb0fdef5583fa6c73c44e
reduce shift operator on coeff remain code
---
source/encoder/entropy.cpp |7 +--
1 files changed, 5
On Mon, Jun 22, 2015 at 5:46 AM, Deepthi Nandakumar
deep...@multicorewareinc.com wrote:
I don't think the second error that was reported on XP with no-assembly has
been fixed.
Actually, the second part of issue #146 isn't Windows-specific at all,
XP support enabled or not.
It happens with any
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435013054 18000
# Mon Jun 22 17:44:14 2015 -0500
# Node ID dfdf378a3968a15a1465a3aa3098e507fb4f10e5
# Parent 76015a49e45cf979da0380382a31f11758cae26e
cpu: fix multilib compiling for some rarer build options
diff -r 76015a49e45c
On 06/22, Stephen Hutchinson wrote:
On Mon, Jun 22, 2015 at 5:46 AM, Deepthi Nandakumar
deep...@multicorewareinc.com wrote:
I don't think the second error that was reported on XP with no-assembly has
been fixed.
Actually, the second part of issue #146 isn't Windows-specific at all,
XP
# HG changeset patch
# User Min Chen chenm...@163.com
# Date 1435019991 25200
# Node ID 965b507acd52ad89160723253b770ef0036c71a5
# Parent 64167d1ad6d81c6c2d2ab762592d131860825fe9
asm: AVX2 of SAD_x4[32xN]
AVX:
sad_x4[32x32] 36.69x 2843.87 104330.24
sad_x4[32x16] 35.67x 1547.93
# HG changeset patch
# User Min Chen chenm...@163.com
# Date 1435019985 25200
# Node ID b3915bc1febdeb620ecbb0fdef5583fa6c73c44e
# Parent 83a7d824442455ba5e0a6b53ea68e6b7043845de
faster algorithm to calculate ctxSet in codeCoeffNxN()
---
source/encoder/entropy.cpp |5 ++---
1 files changed,
Am 22.06.2015, 11:46 Uhr, schrieb Deepthi Nandakumar
deep...@multicorewareinc.com:
Ah, this was a typo.
OK, with the fix patch (name space identifier all upper case), compiling
passes.
--
Fun and success!
Mario *LigH* Rohkrämer
mailto:cont...@ligh.de
Ah, this was a typo.
We dont have a Windows XP machine in our test farm, so we're hitting blind.
I don't think the second error that was reported on XP with no-assembly has
been fixed.
On Mon, Jun 22, 2015 at 3:12 PM, Mario *LigH* Rohkrämer cont...@ligh.de
wrote:
A lot of warnings in XP
These are compilation errors. I don't have a machine with XP running
either; I just compile on Windows 7 SP1 x64 in XhmikosR's older Win32 MSYS
environment with the following calls:
cmake -G MSYS Makefiles -DWINXP_SUPPORT:BOOL=TRUE ../../source
cmake -G MSYS Makefiles
A lot of warnings in XP compatible Win32 builds with GCC 4.8.2 (here the
HIGH_BIT_DEPTH setup, the 8-bit fails too):
+
Scanning dependencies of target clean-generated
Built target clean-generated
-- cmake version 3.3.0-rc1
-- Detected x86 target processor
-- Found Yasm 1.3.0 to build
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435001649 18000
# Mon Jun 22 14:34:09 2015 -0500
# Node ID 537c5fd8f5d59f5a5f8b3c7a5616e856906eeb49
# Parent af4ab43b2cf0d240a76a100d5174c792ac50a393
cmake: make it more clear that the extra libs are only linked to the CLI
this
On Mon, Jun 22, 2015 at 6:44 PM, Steve Borho st...@borho.org wrote:
ninja? I didn't know that cmake generator worked on Windows
It does, either provided through MSys2 or compiled on its own. It's
not as pretty to look at/compact when executed under MSys2's shell,
though.
Although I was
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435028287 18000
# Mon Jun 22 21:58:07 2015 -0500
# Node ID 98d42fd1c07c746ed4c4257b36ef5d05fb0dc8f3
# Parent 25390edc448aef5ff1732bb077d080f2d254e0d2
doc: fix dither docs on api page (not applicable if internal bit depth != 8)
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435028295 18000
# Mon Jun 22 21:58:15 2015 -0500
# Node ID 05165a2be4f86a5454aa9da4ab7b8f86ad8a7764
# Parent 98d42fd1c07c746ed4c4257b36ef5d05fb0dc8f3
doc: nit
diff -r 98d42fd1c07c -r 05165a2be4f8 doc/reST/api.rst
---
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435022545 18000
# Mon Jun 22 20:22:25 2015 -0500
# Node ID 78c562d38eea28d7b99cbb8ff4a228440c50ca2c
# Parent d5376bf40aeab3eece8504045cbc603a145788b5
clarify 8bit/10bit in code comments
diff -r d5376bf40aea -r 78c562d38eea
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435028560 18000
# Mon Jun 22 22:02:40 2015 -0500
# Node ID 21183e6f9444662cccd84b4309cb9bc4889a65dc
# Parent 05165a2be4f86a5454aa9da4ab7b8f86ad8a7764
doc: attempt to document statically linked multi-library implications
This
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435022674 18000
# Mon Jun 22 20:24:34 2015 -0500
# Node ID 901fc6837a25aca531c9056ed59a6e75126c4a53
# Parent 78c562d38eea28d7b99cbb8ff4a228440c50ca2c
param: clarify 8bit/10bit in logs
removes the internal bit-depth logging for
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435021884 18000
# Mon Jun 22 20:11:24 2015 -0500
# Node ID d5376bf40aeab3eece8504045cbc603a145788b5
# Parent dfdf378a3968a15a1465a3aa3098e507fb4f10e5
doc: replace 'bpp' in docs with 'bit' (do not imply pixels)
diff -r
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435023159 18000
# Mon Jun 22 20:32:39 2015 -0500
# Node ID 0b36c393bb3a2ddaacad0c05dafd78aa3ca44682
# Parent 901fc6837a25aca531c9056ed59a6e75126c4a53
cmake: change msvc multilib build folders to 8bit/10bit
diff -r 901fc6837a25
# HG changeset patch
# User Steve Borho st...@borho.org
# Date 1435023306 18000
# Mon Jun 22 20:35:06 2015 -0500
# Node ID 25390edc448aef5ff1732bb077d080f2d254e0d2
# Parent 0b36c393bb3a2ddaacad0c05dafd78aa3ca44682
cmake: change multilib namespaces to x265_8bit/x265_10bit
diff -r
# HG changeset patch
# User Dnyaneshwar G dnyanesh...@multicorewareinc.com
# Date 1434712676 -19800
# Fri Jun 19 16:47:56 2015 +0530
# Node ID a94e9a1f0fde08e060a9b52e3353ce2f242d9257
# Parent 83a7d824442455ba5e0a6b53ea68e6b7043845de
asm: 10bpp sse4 code for saoCuOrgE0, improved 8740c-974c,
# HG changeset patch
# User Dnyaneshwar G dnyanesh...@multicorewareinc.com
# Date 1434963191 -19800
# Mon Jun 22 14:23:11 2015 +0530
# Node ID f85c15cc0e1d70e63182b03e294c2778f598143d
# Parent 558ffdc4e832061d99f1ec688fe1ae64db48642f
asm: 10bpp sse4 code for saoCuOrgE2
Performance
# HG changeset patch
# User Dnyaneshwar G dnyanesh...@multicorewareinc.com
# Date 1434973065 -19800
# Mon Jun 22 17:07:45 2015 +0530
# Node ID f282fc4d8915ec712f3915f387fc6018481fd467
# Parent a946a0178f57e65c02e6be2d9c6485c58658fe20
asm: 10bpp sse4 code for saoCuOrgB0, improved
# HG changeset patch
# User Dnyaneshwar G dnyanesh...@multicorewareinc.com
# Date 1434969412 -19800
# Mon Jun 22 16:06:52 2015 +0530
# Node ID a946a0178f57e65c02e6be2d9c6485c58658fe20
# Parent f85c15cc0e1d70e63182b03e294c2778f598143d
asm: 10bpp sse4 code for saoCuOrgE3
Performance
SAO_EO_08.97x974.03 8740.81
SAO_EO_110.18x 492.67 5017.42
SAO_EO_1_2Rows 11.21x 900.82 10095.86
SAO_EO_2[0] 6.27x207.22 1298.92
SAO_EO_2[1] 8.92x555.20 4949.69
SAO_EO_3[0] 4.97x236.72 1177.29
# HG changeset patch
# User Dnyaneshwar G dnyanesh...@multicorewareinc.com
# Date 1434948494 -19800
# Mon Jun 22 10:18:14 2015 +0530
# Node ID 1c02df66f093b5b1bacc8a1bbf9be2ef81591ad5
# Parent a94e9a1f0fde08e060a9b52e3353ce2f242d9257
asm: 10bpp sse4 code for saoCuOrgE1, improved 5017c-470c,
# HG changeset patch
# User Rajesh Paulrajraj...@multicorewareinc.com
# Date 1434981872 -19800
# Mon Jun 22 19:34:32 2015 +0530
# Node ID d4c7638a0d5b842ca2657969b0f1a2bcd8a82d0b
# Parent 83a7d824442455ba5e0a6b53ea68e6b7043845de
asm: pixelavg_pp[16xN] avx2 code for 10bpp
avx2:
avg_pp[ 16x4]
31 matches
Mail list logo