-
Notifications
You must be signed in to change notification settings - Fork 787
[SYCL][CUDA] Allow "ptx" as device object format #10801
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
This allows one to compile SYCL device code to plain PTX directly: -fsycl-targets=nvptx64-nvidia-cuda -fsycl-device-obj=ptx -S
This PR is intended to address #8797 |
@stefanatwork, could you add a test? |
@mdtoguchi Could you please explain what the test looks like ? Thanks. |
The test would use the option I also noticed that there is no documentation of this new argument - do we need to update the -help information or user guide? |
It is a good idea to add the new argument to the doc/help! Not familiar with driver test, I will ping @npmiller @jchlanda. Thanks. |
WRT the test, you could look at how different phases are inspected in: https://github.com/intel/llvm/blob/sycl/clang/test/Driver/sycl-spirv-obj.cpp#L1 (that's with the use of |
Could this new flag be used in the "normal" SYCL flow, not to produce PTX, but to do the device code compilation per-object, before the linking stage? |
This pull request is stale because it has been open 180 days with no activity. Remove stale label or comment or this will be automatically closed in 30 days. |
This pull request was closed because it has been stalled for 30 days with no activity. |
I've resurrected this patch here: #17390 |
This allows one to compile SYCL device code to plain PTX directly: -fsycl-targets=nvptx64-nvidia-cuda -fsycl-device-obj=ptx -S