kkraus14 commented on PR #34972: URL: https://github.com/apache/arrow/pull/34972#issuecomment-1511730719
For AMD GPUs there's HIP Events (https://docs.amd.com/bundle/HIP-API-Guide-v5.2/page/group___event.html) which operate the same as CUDA Events. There is OpenCL / SYCL events that appear similar as well: https://developer.codeplay.com/products/computecpp/ce/2.11.0/guides/sycl-for-cuda-developers/migrating-from-cuda-to-sycl#platform-api-equivalence. For SYCL, there is the ability to automatically handles data dependencies: https://developer.codeplay.com/products/computecpp/ce/2.11.0/guides/sycl-guide/multiple-kernels. Though I suspect the raw pointers that we use for passing the data aren't sufficient for this. I'm not sure of other programming models that we'd like to capture. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
