diff options
| author | Jonathan Wright <jonathan.wright@arm.com> | 2021-02-04 23:00:00 +0000 |
|---|---|---|
| committer | Jonathan Wright <jonathan.wright@arm.com> | 2021-04-30 18:40:22 +0100 |
| commit | f546e0d3d0316aa76a45de1f548591bde7308c41 (patch) | |
| tree | 91cb2cb9a9545cb19a8601ade47e8496895f8e5a /gcc/value-range.h | |
| parent | 5672fe9da4ab4e8787c288b64008251065c67c98 (diff) | |
| download | gcc-f546e0d3d0316aa76a45de1f548591bde7308c41.zip gcc-f546e0d3d0316aa76a45de1f548591bde7308c41.tar.gz gcc-f546e0d3d0316aa76a45de1f548591bde7308c41.tar.bz2 | |
aarch64: Use RTL builtins for vmull[_high]_p8 intrinsics
Rewrite vmull[_high]_p8 Neon intrinsics to use RTL builtins rather
than inline assembly code, allowing for better scheduling and
optimization.
gcc/ChangeLog:
2021-02-05 Jonathan Wright <joanthan.wright@arm.com>
* config/aarch64/aarch64-simd-builtins.def: Add pmull[2]
builtin generator macros.
* config/aarch64/aarch64-simd.md (aarch64_pmullv8qi): Define.
(aarch64_pmull_hiv16qi_insn): Define.
(aarch64_pmull_hiv16qi): Define.
* config/aarch64/arm_neon.h (vmull_high_p8): Use RTL builtin
instead of inline asm.
(vmull_p8): Likewise.
Diffstat (limited to 'gcc/value-range.h')
0 files changed, 0 insertions, 0 deletions