[PATCH] D156301: [WIP] Support FP global atomics in AMDGPUAtomicOptimizer.

[Jay Foad via Phabricator via llvm-commits]

foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUAtomicOptimizer.cpp:652
+  case AtomicRMWInst::FMin:
+    return APFloat::getLargest(APFloat::IEEEsingle(), false);
+  }
----------------
arsenm wrote:
> This would be -infinity
No, the identity should be +inf for fmin and -inf for fmax.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D156301/new/

https://reviews.llvm.org/D156301