https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105966
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Blocks| |88670
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Is not having AVX512VL relevant in the real world? Some operations (division)
require different handling than zero-extending, masking might be a way out
but that might turn out to be (way?) more expensive.
I agree that it might be interesting to support SLP with not power-of-two
or generally not fully populated lanes. The load/store side requires
masking support for this (and the missed optimization would be that we do
not define the contents of the masked elements for loads).
Likewise vector lowering could avoid splitting vector ops into scalars when
there's a wider supported vector mode by means of zero-extending. It might
need a cost model for this. I think the reporter is aiming at this.
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88670
[Bug 88670] [meta-bug] generic vector extension issues
