-
Notifications
You must be signed in to change notification settings - Fork 741
Arm backend: Add devtools support to example #6959
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
Arm backend: Add devtools support to example #6959
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/6959
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 5193bbb with merge base 0fb2ccf ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@pytorchbot label "topic: not user facing" |
|
@digantdesai This changes files outside Arm folders and need a review from you, Thanks for your time! |
| using executorch::runtime::internal::EventTracerProfileOpScope; | ||
| #define EXECUTORCH_INTERNAL_PROF(NAME) \ | ||
| EventTracerProfileOpScope event_tracer_op_scope = \ | ||
| EventTracerProfileOpScope(event_tracer, NAME); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah AFAICT this is not meant to be used from outside, as you were suspecting.
So I would try using stuff from executorch/runtime/core/event_tracer.h like EventTracer::[start|stop]_profiling?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes that works, PR will be updated soon.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fixed!
New flags on run.sh --etdump Build in etdump and profiling, the etdump base64 coded and put in the log --debug_build Build debug instead of release --extra_build_flags Extra flags to pass to cmake this makes it for example possible to override the allocator pool size or other build time cmake flags. The devtools build has been updated so FLATCC_EXECUTABLE can be used to point out the executable. Signed-off-by: Zingo Andersen <[email protected]> Change-Id: Ic0fb1e48ee633c5fe91473bdc2db9e894b2fc4fa
ed1bdd7 to
3f06cec
Compare
|
Error seem unrealted to this patch e.g. unittest / linux / linux-job] - https://github.com/pytorch/executorch/actions/runs/12255701820/job/34189395827?pr=6959#logs ++ curl --silent --show-error --location --fail --retry 3 --output /tmp/vulkansdk.tar.gz https://sdk.lunarg.com/sdk/download/1.2.198.1/linux/vulkansdk-linux-x86_64-1.2.198.1.tar.gz |
New flags on run.sh
--etdump
Build in etdump and profiling, the etdump base64 coded and put in the log
--debug_build
Build debug instead of release
--extra_build_flags
Extra flags to pass to cmake this makes it for example possible to
override the allocator pool size or other build time cmake flags.
The devtools build has been updated so FLATCC_EXECUTABLE can be used to point out the executable as seem to be intended to be possible but was not working.