frasercrmck wrote: > These should use the regular builtin fmin / fmax.
Do you mean the AMD implementations, or the CLC ones too? Note there's no vector support for `__builtin_fmin` which is why I chose `__builtin_elementwise_min`. They appear to generate the same code so maybe I'm misunderstanding the difference between the two builtins. > > > I note that the comments around the use of canonicalize mention sNAN, which > > isn't required by the spec. > > The spec is quite badly written on what's expected of snans here, and the > conformance test doesn't test what is written in the spec (hoping to fix that > [here](https://github.com/KhronosGroup/OpenCL-CTS/pull/2285) Thanks for the link. I was going by [7.2](https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_C.html#inf-nan-and-denormalized-numbers) but now I see there's also a [footnote](https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_C.html#_footnoteref_43). https://github.com/llvm/llvm-project/pull/128506 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
