Message ID | 20220223092126.78821-1-hongtao.liu@intel.com |
---|---|
State | New |
Headers | show |
Series | [i386] Fix typo in <code>v1ti3. | expand |
On Wed, Feb 23, 2022 at 05:21:26PM +0800, liuhongt via Gcc-patches wrote: > For evex encoding vp{xor,or,and}, suffix is needed. > > Or there would be an error for > vpxor %ymm0, %ymm31, %ymm1 The insn is about V1TImode, so the error would be on vpxor %xmm0, %xmm31, %xmm1 > > Error: unsupported instruction `vpxor' > > Bootstrapped and regtested x86_64-pc-linux-gnu{-m32,}. > Pushed to trunk. > > gcc/ChangeLog: > > * config/i386/sse.md (<code>v1ti3): Add suffix and replace > isa attr of alternative 2 from avx to avx512vl. The patch looks good, but I think it would be nice to have a dg-do assemble testcase for it. Something like untested: /* { dg-do assemble { target { int128 && avx512vl } } } */ /* { dg-options "-O2 -mavx512vl" } */ typedef __int128 V __attribute__((vector_size (16))); void foo (V *x, V *y, V *z) { register V a __asm ("xmm31") = *z; __asm ("" : "+v" (a)); x[0] = y[0] & a; x[1] = y[1] | a; x[2] = y[2] ^ a; } Jakub
On Wed, Feb 23, 2022 at 5:48 PM Jakub Jelinek via Gcc-patches <gcc-patches@gcc.gnu.org> wrote: > > On Wed, Feb 23, 2022 at 05:21:26PM +0800, liuhongt via Gcc-patches wrote: > > For evex encoding vp{xor,or,and}, suffix is needed. > > > > Or there would be an error for > > vpxor %ymm0, %ymm31, %ymm1 > > The insn is about V1TImode, so the error would be on > vpxor %xmm0, %xmm31, %xmm1 > > > > > Error: unsupported instruction `vpxor' > > > > Bootstrapped and regtested x86_64-pc-linux-gnu{-m32,}. > > Pushed to trunk. > > > > gcc/ChangeLog: > > > > * config/i386/sse.md (<code>v1ti3): Add suffix and replace > > isa attr of alternative 2 from avx to avx512vl. > > The patch looks good, but I think it would be nice to have a dg-do assemble > testcase for it. Yes will add. > Something like untested: > /* { dg-do assemble { target { int128 && avx512vl } } } */ > /* { dg-options "-O2 -mavx512vl" } */ > > typedef __int128 V __attribute__((vector_size (16))); > > void > foo (V *x, V *y, V *z) > { > register V a __asm ("xmm31") = *z; > __asm ("" : "+v" (a)); > x[0] = y[0] & a; > x[1] = y[1] | a; > x[2] = y[2] ^ a; > } > > Jakub >
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md index b2f56345c65..3066ea3734a 100644 --- a/gcc/config/i386/sse.md +++ b/gcc/config/i386/sse.md @@ -17025,8 +17025,8 @@ (define_insn "<code>v1ti3" "@ p<logic>\t{%2, %0|%0, %2} vp<logic>\t{%2, %1, %0|%0, %1, %2} - vp<logic>\t{%2, %1, %0|%0, %1, %2}" - [(set_attr "isa" "noavx,avx,avx") + vp<logic>d\t{%2, %1, %0|%0, %1, %2}" + [(set_attr "isa" "noavx,avx,avx512vl") (set_attr "prefix" "orig,vex,evex") (set_attr "prefix_data16" "1,*,*") (set_attr "type" "sselog")