https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117576
Bug ID: 117576
Summary: -mveclibabi=aocl: Support vectorized array functions
and maybe also sincos
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: pheeck at gcc dot gnu.org
CC: burnus at gcc dot gnu.org, hubicka at gcc dot gnu.org,
jamborm at gcc dot gnu.org, rguenth at gcc dot gnu.org
Target Milestone: ---
Target: x86_64-*-*
GCC currently supports emitting calls to vectorized math functions from the AMD
AOCL LibM library (this is enabled by specifying the -mveclibabi=aocl cmdline
option). However, GCC doesn't support all available functions.
>From commit message of r15-5204-g99ec0eb32a0350:
> Compared to the list of functions in AOCL LibM docs I left out these
> vectorized function families:
>
> - sincos and all functions working with arrays ... Because these
> functions have pointer arguments and that would require a bigger
> rework of ix86_veclibabi_aocl(). Also, I'm not sure if GCC even ever
> generates calls to these functions.
> - linearfrac ... Because these functions are specific to the AMD
> library. There's no equivalent glibc function nor GCC internal
> function nor GCC built-in.
> - powx, sqrt, fabs ... Because GCC doesn't vectorize these functions
> into calls and uses instructions instead.
It could be interesting to add the missing support for emmiting calls to the
array functions. There is a short remark on this by Richi in pr56504 and a
short remark by Honza in
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668603.html. It could
also be interesting to add support for calls to vectorized sincos. Honza also
comments on this in the linked mail.
--- Misc notes: ---
The array functions differ from other vectorized functions by having pointers
to arrays of scalar types as parameters instead of having vector types as
parameters. See an example here:
https://github.com/amd/aocl-libm-ose/blob/master/src/optimized/vec/vrda_sin.c
Extensive list of AMD vectorized math functions can be found here:
https://docs.amd.com/r/en-US/57404-AOCL-user-guide/AOCL-LibM?section=fast-scalar-and-vector-variants
The library can be downloaded here (on this page, there is also a link to its
source files):
https://www.amd.com/en/developer/aocl/libm.html
