shithub: dav1d

Info  •  Files  •  Log  •  Branches

x86: don't use vptest in SSSE3 version

This is the VEX (AVX) encoded variant for the SSE4 instruction ptest,
so emulate it using pmovmskb in the SSSE3 version.

--- a/src/x86/mc_ssse3.asm

+++ b/src/x86/mc_ssse3.asm

@@ -5212,8 +5212,16 @@

     ; if no emulation is required, we don't need to shuffle or emulate edges

     ; this also saves 2 quasi-vpgatherdqs

-    vptest               m3, m3

-    jz .filter

+    pxor                 m6, m6

+    pcmpeqb              m6, m3

+%if ARCH_X86_64

+    pmovmskb            r8d, m6

+    cmp                 r8d, 0xffff

+%else

+    pmovmskb            r3d, m6

+    cmp                 r3d, 0xffff

+%endif

+    je .filter

 %if ARCH_X86_64

     movd                r8d, xm3