pytorch/test/cpp
Salil Desai e2dc60c6cb [Vulkan + Profiler] Add Timestamp Adjustment Algorithm (#90672)
@bypass-github-export-checks

This change ensures that vulkan event start/end times are correctly synced with their parent CPU times.

This sometimes requires increasing CPU event durations (to fully contain their child events) and delaying CPU event start times (to prevent overlaps), so this should not be used unless Vulkan events are being profiled and it is ok to use this modified timestamp/duration information instead of the the original information.

Differential Revision: [D39893109](https://our.internmc.facebook.com/intern/diff/D39893109/)

**NOTE FOR REVIEWERS**: This PR has internal Meta-specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D39893109/)!
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90672
Approved by: https://github.com/kimishpatel
2022-12-19 20:01:07 +00:00
..
api Changing the use from ASSERT_EQ to ASSERT_FLOAT_EQ on nn_utils test. (#83693) 2022-11-15 04:10:52 +00:00
c10d Allow Process Group to support multiple backends (#88330) (#90997) 2022-12-16 23:15:00 +00:00
common
dist_autograd
jit Clean up dependancy for flatbuffer_loader (#86041) 2022-12-08 03:48:04 +00:00
lazy [LTC] Make ComputePostOrder accept const T pointers (#88773) 2022-11-10 18:34:19 +00:00
lite_interpreter_runtime [Vulkan + Profiler] Add Timestamp Adjustment Algorithm (#90672) 2022-12-19 20:01:07 +00:00
monitor
profiler Nested profiling support for Linux-perf Profiler (#87904) 2022-11-02 14:51:53 +00:00
rpc Refactor distribuetd to use absolute header path (#85780) 2022-09-30 05:13:50 +00:00
tensorexpr Fix the performance issue that the for-loop before ExternallCall could not be parallelized. (#85056) 2022-10-07 07:36:28 +00:00
__init__.py