aarongreig

Proposal/RFC: full precision math builtins

8

Hello :wave: I've been working on allowing clspv to generate SPIR-V that meets the full math builtin precision requirements in the OpenCL C spec. My approach to this has been...

Adding FP16 math builtins

5

This is a libclc issue but I think the discussion is also worth having here. I've been working on a couple of ways we could add fp16 implementations of the...

Pulls in changes from https://github.com/oneapi-src/unified-runtime/pull/1212

https://github.com/oneapi-src/unified-runtime/pull/1196

https://github.com/oneapi-src/unified-runtime/pull/805

https://github.com/intel/llvm/pull/13876

Consider adding forward progress device info enums

6

A set of new device info enums was [added to PI](https://github.com/intel/llvm/pull/13389/files#diff-952d2515bc7a8c91d37420f519ff726c0e6ce71f5a81bfd98e5c6fcfcaa3ada0R440) recently to support the [forward progress extension](https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/proposed/sycl_ext_oneapi_forward_progress.asciidoc). These do not need adapter implementations, but they will be removed as...

Correct level of indirection used in KernelSetArgPointer calls.

6

Also attempt to clarify the wording around this a bit. Addresses #558 LLVM testing https://github.com/intel/llvm/pull/12270

These were removed from the spec a while ago so this is dead code.

Replace PROGRAM_INFO_SOURCE with PROGRAM_INFO_IL.

1

Also make returns for this consistent across adapter implementations (i.e. actual IL if that's supported, otherwise return UNSUPPORTED_ENUMERATION). resolves #1138 LLVM testing https://github.com/intel/llvm/pull/12265