Enable architecture selection for DPCTL_TARGET_CUDA#2096
Enable architecture selection for DPCTL_TARGET_CUDA#2096ndgrigorian merged 17 commits intomasterfrom
DPCTL_TARGET_CUDA#2096Conversation
|
Deleted rendered PR docs from intelpython.github.com/dpctl, latest should be updated shortly. 🤞 |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_8 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_9 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_10 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_17 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_22 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_30 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_34 ran successfully. |
|
Array API standard conformance tests for dpctl=0.21.0dev0=py310h93fe807_37 ran successfully. |
This PR proposes to change
DPCTL_TARGET_CUDACMake option from a boolean to a string allowing users to specify a CUDA architecture (e.g.sm_80). If not specified, it defaults tosm_50.The specified architecture is used to construct a SYCL alias target (e.g.
nvidia_gpu_sm_80) and passed via-fsycl-targetsoption, following OneAPI for NVIDIA GPUsAdditionally removing
DPCTL_TARGET_CUDAenv handling logic