frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Profiler] Weaken ordering check during post processing.

Browse Source 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/78563

The profiler assembles a call hierarchy by replaying recorded events. There is an assert to ensure that the events form a well structured tree; however many of the inputs are from external sources and small differences (e.g. recording time in a lower precision) leads to traces which violate that assumption. For now this is acceptable; the post processing can handle resolving these descrepencies. As a result, I am relaxing the assert to only test event types where we expect the framework to be able to enforce these strong structural requirements.

Differential Revision: [D36787787](https://our.internmc.facebook.com/intern/diff/D36787787/)

Approved by: https://github.com/suo
This commit is contained in:
PyTorch MergeBot
2022-06-01 18:02:08 +00:00
parent 5aa2ed1922
commit 6a4997e66a
2 changed files with 1 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
test/cpp/lite_interpreter_runtime/test_mobile_profiler.cpp
View File

@ -70,7 +70,7 @@ TEST(MobileProfiler, ModuleHierarchy) {
ASSERT_TRUE(checkMetaData("aten::add", metadata_name, "top(C)::<unknown>.aten::add", trace_file));
}
TEST(MobileProfiler, DISABLED_Backend) {
TEST(MobileProfiler, Backend) {
std::string filePath(__FILE__);
auto testModelFile = filePath.substr(0, filePath.find_last_of("/\\") + 1);
testModelFile.append("test_backend_for_profiling.ptl");

11
torch/csrc/profiler/collection.cpp
View File

@ -392,19 +392,8 @@ void build_tree(std::vector<std::shared_ptr<Result>>& events) {
auto start_tid = event->start_tid_;
auto frame = stacks.at(start_tid);
// Handle any events which are allowed to end without an explicit end call.
// (E.g. a user scope.)
while (frame.get() != event.get()) {
TORCH_INTERNAL_ASSERT(frame != nullptr);
c10::visit(
c10::overloaded(
[](const op_fields& i) {
TORCH_INTERNAL_ASSERT(
i.scope_ == at::RecordScope::USER_SCOPE, (int)i.scope_);
},
[](const ExtraFields<EventType::Backend>& i){},
[](const auto&) { TORCH_INTERNAL_ASSERT(false); }),
frame->extra_fields_);
frame->finished_ = true;
TORCH_INTERNAL_ASSERT(!frame->parent_.expired());
frame = frame->parent_.lock();