| commit | 244929c1fc4f40740356731b7573506872ca7b90 | [log] [tgz] |
|---|---|---|
| author | agl@chromium.org <agl@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | Wed Jun 16 19:52:29 2010 +0000 |
| committer | agl@chromium.org <agl@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | Wed Jun 16 19:52:29 2010 +0000 |
| tree | e991f35c000e407ba757dd807df8e9c62f861e0e | |
| parent | f59799139bacd300bf5251a1ca4e6b2ad3196457 [diff] |
Implementing S32A_Opaque_BlitRow32 using v7 neon instructions. Taking the advantage of 16 channels of each QualWord register. Also using the software pipelining to scatter the loads/stores among vector operations. Got roughly 70% improvements on simulation environments. http://codereview.appspot.com/1148042/show Patch-by: XinQi of codeaurora.org git-svn-id: http://skia.googlecode.com/svn/trunk@578 2bbb7eff-a529-9590-31e7-b0007b416f81