[PATCH] D113256: [AArch64][ARM] Enablement of Cortex-A710 Support

Dave Green via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sun Nov 7 08:48:09 PST 2021


dmgreen added inline comments.


================
Comment at: llvm/include/llvm/Support/AArch64TargetParser.def:159-160
                   AArch64::AEK_RCPC | AArch64::AEK_SSBS))
+AARCH64_CPU_NAME("cortex-a710", ARMV9A, FK_NEON_FP_ARMV8, false,
+                 (AArch64::AEK_MTE | AArch64::AEK_PAUTH | AArch64::AEK_FLAGM |
+                  AArch64::AEK_SB | AArch64::AEK_I8MM | AArch64::AEK_FP16FML |
----------------
lenary wrote:
> dmgreen wrote:
> > Natural order would be better I think, where the new A710 is added after the A78.
> > 
> > FlagM should already be included as a part of 8.4, so isn't needed here.
> > Should BFloat16 be added?
> > FlagM should already be included as a part of 8.4, so isn't needed here.
> 
> FlagM is not part of the `AARCH64_ARCH("armv9-a", ARMV9A ...)`  definition, nor part of the equivalents for armv8.4a and armv8.5a, so it was added explicitly here.
> 
> > Should BFloat16 be added?
> 
> Yes
> 
> 
> 
FeatureFlagM is included in HasV8_4aOps. So should already be included (much like something like dotprod).

Why AEK_FLAGM isn't part of ARMV8_4A I don't know. And why we have two maps for the same set of information...


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113256/new/

https://reviews.llvm.org/D113256



More information about the llvm-commits mailing list