| Message ID | 20190122163158.GU30353@tucnak |
|---|---|
| State | New |
| Headers | show |
| Series | Disable *vsx_reduc_<VEC_reduc_name>_v2df_scalar on little endian (PR target/87064) | expand |
Hi Jakub, On Tue, Jan 22, 2019 at 05:31:58PM +0100, Jakub Jelinek wrote: > The following patch disables *vsx_reduc_<VEC_reduc_name>_v2df_scalar > define_insn_and_split for little endian, because it both generates > wrong code there (fixable), but also even if fixed, emit much worse code > than when it is not present at all. > The comment on the pattern says: > ;; Combiner patterns with the vector reduction patterns that knows we can get > ;; to the top element of the V2DF array without doing an extract. > but for little endian that is not the case, the end result is that a vector > is unnecessarily forced into memory. > > Fixed thusly, bootstrapped/regtested on powerpc64le-linux, fixes > FAIL: libgomp.oacc-fortran/reduction-3.f90 -DACC_DEVICE_TYPE_host=1 -DACC_MEM_SHARED=1 -O1 execution test > , ok for trunk? Okay. Thanks! Also okay for backports. Segher > 2019-01-22 Jakub Jelinek <jakub@redhat.com> > > PR target/87064 > * config/rs6000/vsx.md (*vsx_reduc_<VEC_reduc_name>_v2df_scalar): > Disable for little endian.
--- gcc/config/rs6000/vsx.md.jj 2019-01-19 09:39:13.224924063 +0100 +++ gcc/config/rs6000/vsx.md 2019-01-21 23:39:27.527774247 +0100 @@ -4351,7 +4351,7 @@ (define_insn_and_split "*vsx_reduc_<VEC_ (match_dup 1)) (parallel [(const_int 1)]))) (clobber (match_scratch:DF 2 "=0,0,&wd,&wa"))] - "VECTOR_UNIT_VSX_P (V2DFmode)" + "BYTES_BIG_ENDIAN && VECTOR_UNIT_VSX_P (V2DFmode)" "#" "" [(const_int 0)]