For clarify, default0Save1Load2 is renamed to bReusePred
# HG changeset patch
# User Satoshi Nakagawa nakagawa...@oki.com
# Date 1394251730 -32400
# Sat Mar 08 13:08:50 2014 +0900
# Node ID ee48209c7d8bb298adc3a0eebc8f4c664ffe0f4c
# Parent 2bf727dca27d6f69e96d4412850661cbe036cbef
cleanup
ok, rename is more clearly, you can do more optimize in future.
At 2014-03-08 12:21:33,Satoshi Nakagawa nakagawa...@oki.com wrote:
For clarify, default0Save1Load2 is renamed to bReusePred
# HG changeset patch
# User Satoshi Nakagawa nakagawa...@oki.com
# Date 1394251730 -32400
# Sat Mar 08
The pointer recon is unused, your patch is right
But you remove m_sharedPredTransformSkip, it is temporary buffer in tskip
decide path, please verify more with --tskip
At 2014-03-04 18:40:23,Satoshi Nakagawa nakagawa...@oki.com wrote:
# HG changeset patch
# User Satoshi Nakagawa
A little more, I know *pred is readonly, and default0Save1Load2 may 1 or 2 in
same loop, so *pred can safe to reuse, it reduce one memory copy.
But they in different level of function area, so we have to add a comment on
loop (xRecurIntraCodingQT), for warning everyone, when they modify
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
# 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
