On Tue, Jan 30, 2018 at 2:38 AM, Jan Vesely <jan.ves...@rutgers.edu> wrote: > On Mon, 2018-01-29 at 22:15 +0300, Roman Lebedev via cfe-commits wrote: >> On Mon, Jan 29, 2018 at 10:05 PM, Jan Vesely via cfe-commits >> <cfe-commits@lists.llvm.org> wrote: >> > Author: jvesely >> > Date: Mon Jan 29 11:05:08 2018 >> > New Revision: 323677 >> > >> > URL: http://llvm.org/viewvc/llvm-project?rev=323677&view=rev >> > Log: >> > math.h: Set HAVE_HW_FMA32 based on compiler provided macro >> > >> > Fixes sin/cos piglits on non-FMA capable asics. >> > Bugzilla: https://bugs.llvm.org/show_bug.cgi?id=35983 >> > >> > Reviewer: Tom Stellard >> > Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu> >> > >> > Modified: >> > libclc/trunk/generic/lib/math/math.h >> > >> > Modified: libclc/trunk/generic/lib/math/math.h >> > URL: >> > http://llvm.org/viewvc/llvm-project/libclc/trunk/generic/lib/math/math.h?rev=323677&r1=323676&r2=323677&view=diff >> > ============================================================================== >> > --- libclc/trunk/generic/lib/math/math.h (original) >> > +++ libclc/trunk/generic/lib/math/math.h Mon Jan 29 11:05:08 2018 >> > @@ -31,7 +31,12 @@ >> > #define PNOR 0x100 >> > #define PINF 0x200 >> > >> > +#if (defined __AMDGCN__ | defined __R600__) & !defined __HAS_FMAF__ >> >> Shouldn't that be || and && ? > > Yes, that would be more readable.
> It should work the same though, since the values are 0 and 1. Yes. >Do you want to send a patch? No, feel free to just commit the fix :) > Jan > >> >> > +#define HAVE_HW_FMA32() (0) >> > +#else >> > #define HAVE_HW_FMA32() (1) >> > +#endif >> > + >> > #define HAVE_BITALIGN() (0) >> > #define HAVE_FAST_FMA32() (0) >> > >> > >> > >> > _______________________________________________ >> > cfe-commits mailing list >> > cfe-commits@lists.llvm.org >> > http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits >> >> _______________________________________________ >> cfe-commits mailing list >> cfe-commits@lists.llvm.org >> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits