bmahjour added inline comments.
================
Comment at: clang/lib/Headers/altivec.h:17337
static __inline__ vector long long __ATTRS_o_ai
vec_vbpermq(vector unsigned char __a, vector unsigned char __b) {
return __builtin_altivec_vbpermq(__a, __b);
----------------
lei wrote:
> bmahjour wrote:
> > This should be guarded under P8. It would also be good to add a
> > `vec_vbpermd(vector unsigned long long ...)` counter part under
> > `__POWER9_VECTOR__` for consistency.
> I think this is actually guarded under ` __POWER8_VECTOR__`. See line 17237.
> As far as I can see, there is no LLVM intrinsic corresponding to vbpermd.
> I think this is actually guarded under __POWER8_VECTOR__. See line 17237.
You are right. I missed that.
> As far as I can see, there is no LLVM intrinsic corresponding to vbpermd.
`__builtin_altivec_vbpermd` is being called on line 17356...not sure if it
corresponds to an IR intrinsic or not, but regardless we have `vec_vbpermq` in
this header but not `vec_vbpermd`. That seems inconsistent to me.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D107899/new/
https://reviews.llvm.org/D107899
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
