[SYCL] Intel SYCL runtime support for AutoGPTQ #638
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hello @PanQiWei . @qwopqwop200 , @fxmarty .Greetings from Intel. Thanks for creating AutoGPTQ.
Considering adoption of more SYCL runtime across different vendors and specifically for Intel accelerators, this is an effort to upstream the cuda kernels for (gemm/non gemm quants) to SYCL runtime . This is an approach taken to further extend Intel accelerator support as well as provide an alternative compiler runtime for non cuda specific devices.
The proposal is to extend SYCL runtime through DPCT toolchain and to support a standard runtime for different vendors.
Since the idea is to provide better sycl compiler characterization, this PR would help enable the framework for a wide range of accelerators working on SYCL runtime.S
An initial kernel prototype is added, and further extension kernels will be ported to support SYCL backend in next couple of days. (Draft mode).
An extended idea: (if community is interested) SYCL maintenance will be extended by myself (team) and external contributors and depending upon support received we plan to integrate nv/amd and other vendor support for SYCL runtime for AutoGPTQ.