[Bug target/70321] [6 Regression] STV generates less optimized code

ienkovich at gcc dot gnu.org gcc-bugzilla@gcc.gnu.org
Tue Mar 22 14:07:00 GMT 2016


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70321

--- Comment #7 from Ilya Enkovich <ienkovich at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #6)
> Why couldn't STV just "vectorize" AND and NOT patterns and let the combiner
> combine that in the vectorized code?

I think the only thing we miss for that is corresponding NOT pattern in DI
mode.

For comparison we have

{r94:SI=r89:DI#4|r89:DI#0;clobber flags:CC;}
flags:CCZ=cmp(r94:SI,0)

combined into

{flags:CCZ=cmp(r89:DI#4|r89:DI#0,0);clobber scratch;}

which is converted into PTEST by STV.


More information about the Gcc-bugs mailing list