From: Yunqing Wang Date: Thu, 22 May 2014 22:42:13 +0000 (-0700) Subject: Fix a decoding mismatch in sub-pixel filters X-Git-Tag: v1.4.0~1506^2~1 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=efcdf946edc7b95eb4811d41ae7d756002fa548c;p=platform%2Fupstream%2Flibvpx.git Fix a decoding mismatch in sub-pixel filters This did the same correction as the one in commit "Correct ssse3 8/16-pixel wide sub-pixel filter calculation" to avoid saturation during filtering. Change-Id: Ife9aa3f62daf9114eb24fe38f7baa3c3f361b2d6 --- diff --git a/vp9/common/x86/vp9_subpixel_8t_ssse3.asm b/vp9/common/x86/vp9_subpixel_8t_ssse3.asm index 634fa77..fd781d4 100644 --- a/vp9/common/x86/vp9_subpixel_8t_ssse3.asm +++ b/vp9/common/x86/vp9_subpixel_8t_ssse3.asm @@ -272,22 +272,23 @@ punpcklbw xmm2, xmm3 ;C D punpcklbw xmm4, xmm5 ;E F - movq xmm6, [rsi + rbx + 8] ;G movq xmm7, [rax + rbx + 8] ;H punpcklbw xmm6, xmm7 ;G H - pmaddubsw xmm0, k0k1 pmaddubsw xmm2, k2k3 pmaddubsw xmm4, k4k5 pmaddubsw xmm6, k6k7 paddsw xmm0, xmm6 - paddsw xmm0, xmm2 + movdqa xmm1, xmm2 + pmaxsw xmm2, xmm4 + pminsw xmm4, xmm1 paddsw xmm0, xmm4 - paddsw xmm0, krd + paddsw xmm0, xmm2 + paddsw xmm0, krd psraw xmm0, 7 packuswb xmm0, xmm0