[all-commits] [llvm/llvm-project] 81967c: [AMDGPU] Document that only naturally aligned atom...

[Fabian Ritter via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 81967cfb4ca816c59a1c9719f4aa773730686310
      https://github.com/llvm/llvm-project/commit/81967cfb4ca816c59a1c9719f4aa773730686310
  Author: ritter-x2a <9519134+ritter-x2a at users.noreply.github.com>
  Date:   2026-06-10 (Wed, 10 Jun 2026)

  Changed paths:
    M llvm/docs/AMDGPUUsage.rst

  Log Message:
  -----------
  [AMDGPU] Document that only naturally aligned atomics of up to 64 bits are supported by the AMDGPU backend (#200167)

We get an error from AtomicExpandPass if those constraints are not satisfied.
The 64-bit limit is set [here, in AMDGPUISelLowering.cpp](https://github.com/llvm/llvm-project/blob/5cac2751fb9cf3112d16717b278e40d07dd6cfdc/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp#L645).

This patch also introduces a new "Unsupported IR Constructs" section to the AMDGPUUsage doc, where we can document more such cases.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications