Skip to content

Conversation

@anmyachev
Copy link
Contributor

@anmyachev anmyachev commented Oct 18, 2024

This is another preparatory step to switching to an upstream profiler.

Closes #2512

@anmyachev anmyachev linked an issue Oct 18, 2024 that may be closed by this pull request
from .benchmark_testing import do_bench, assert_close, perf_report, Benchmark, USE_IPEX_OPTION, BENCHMARKING_METHOD # type: ignore # noqa: F401

if USE_IPEX_OPTION:
if USE_IPEX_OPTION or BENCHMARKING_METHOD == "UPSTREAM_PYTORCH_PROFILER":
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For elapsed_time I still want to leave the upstream driver, since it does not contain additional function calls for profiling, which introduce their overhead, but will not be used, since there is a different profiling technology.

@anmyachev anmyachev marked this pull request as ready for review October 18, 2024 13:01
@anmyachev anmyachev merged commit 4c5296d into main Oct 21, 2024
6 checks passed
@anmyachev anmyachev deleted the amyachev/upstream-prof-driver branch October 21, 2024 09:17
anmyachev added a commit that referenced this pull request Dec 17, 2024
The same way as for Triton benchmarks, let's see how it works.

Ref: #2510


https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/12379668850
(upstream profiler)

Signed-off-by: Anatoly Myachev <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

3 participants