ref: f306f96913bb745dc818a77b52e4e7f12067a493
parent: 81aba493930a7849af24217657cf1061608666b2
author: Henrik Gramner <gramner@twoorioles.com>
date: Mon Dec 30 23:14:36 EST 2019
x86: Fix inverse (flip)adst 8x4 clipping
--- a/src/x86/itx.asm
+++ b/src/x86/itx.asm
@@ -1447,7 +1447,7 @@
punpckhwd m2, m0, m1
punpcklwd m0, m1
pxor m3, m3
- psubw m3, m2
+ psubsw m3, m2
punpckhwd m1, m0, m3
punpcklwd m0, m3
jmp tx2q
@@ -1492,7 +1492,7 @@
punpckhwd m1, m3, m2
punpcklwd m3, m2
pxor m0, m0
- psubw m0, m1
+ psubsw m0, m1
punpckhwd m1, m0, m3
punpcklwd m0, m3
jmp tx2q
--- a/src/x86/itx_ssse3.asm
+++ b/src/x86/itx_ssse3.asm
@@ -1008,8 +1008,8 @@
punpckhwd m1, m2, m3
punpcklwd m2, m3
pxor m5, m5
- psubw m3, m5, m1
- psubw m5, m4
+ psubsw m3, m5, m1
+ psubsw m5, m4
punpckhdq m4, m5, m3
punpckldq m5, m3
punpckhdq m3, m0, m2
@@ -1127,8 +1127,8 @@
punpcklwd m1, m0
pxor m0, m0
- psubw m4, m0, m2
- psubw m0, m5
+ psubsw m4, m0, m2
+ psubsw m0, m5
punpckhdq m2, m0, m4
punpckldq m0, m4
punpckhdq m4, m3, m1