jthackray wrote: > Should I look at the ABI to understand if these tests are lowering properly? > Like in here: > https://github.com/ARM-software/abi-aa/blob/main/atomicsabi64/atomicsabi64.rst#id30
Yes, https://github.com/ARM-software/abi-aa/blob/main/atomicsabi64/atomicsabi64.rst#bit-floating-point-types shows what the instructions should be before and after. In PR #137703, you can see in the diffs that this changes to a single `ldfmax` instruction (note, not `ldfmaxnm`, as these were added in #125686). - `atomicrmw fmax` lowers to `ldfmaxnm` instruction - `atomicrmw fmaximum` lowers to `ldfmax` instruction and obviously the fmin/fminimum are similar. https://github.com/llvm/llvm-project/pull/137702 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
