-
Notifications
You must be signed in to change notification settings - Fork 38
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
Serialize parameters before kernel launch #1471
Labels
Milestone
Comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
For driver debugging it is useful to isolate Triton kernels to the underlying SPV code and run that SPV code in a controlled environment. We have introduced the
SPIRVRunner
which shares code with the Intel Triton backend, allowing us to run the generated SPV code for a kernel using the same arguments. The runner is not generalized and requires manually setting parameters to match the kernel params. If we could serialize the parameters for a kernel in some structured format just before the kernel launches, we could improve generalization of the runner.Add serialization (to JSON? + pytorch format tensors) of
args
dictionary here:https://github.com/intel/intel-xpu-backend-for-triton/blob/llvm-target/third_party/intel/backend/driver.py#L393
in
XPULauncher.__call__
.Add corresponding deserialization to
SPIRVRunner.cpp
.The text was updated successfully, but these errors were encountered: