Skip to content
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.

Sign up for GitHub

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

fix record event for operator type in new dygraph #44582

Merged
merged 2 commits into from
Jul 26, 2022
Merged

fix record event for operator type in new dygraph #44582

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ad4f44b
fix new dygraph record event for op
rainyfly Jul 25, 2022
c352a37
update unit test
rainyfly Jul 25, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion paddle/fluid/eager/auto_code_generator/eager_generator.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1386,7 +1386,7 @@ static std::string GenerateGradNodeCreationContent(
"%s"
" {\n"
" paddle::platform::RecordEvent node_creation_record_event(\"%s\", "
"paddle::platform::TracerEventType::Operator, 1);\n"
"paddle::platform::TracerEventType::OperatorInner, 1);\n"
"%s"
" if(require_any_grad) {\n"
" VLOG(6) << \" Construct Grad for %s \"; \n"
Expand Down
2 changes: 1 addition & 1 deletion paddle/fluid/eager/auto_code_generator/final_state_generator/python_c_gen.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -111,7 +111,7 @@ def FindParsingFunctionFromAttributeType(atype):


RECORD_EVENT_TEMPLATE = \
"paddle::platform::RecordEvent {}(\"{} {}\", paddle::platform::TracerEventType::Operator, 1);"
"paddle::platform::RecordEvent {}(\"{} {}\", paddle::platform::TracerEventType::UserDefined, 1);"


RETURN_INPLACE_PYOBJECT_TEMPLATE = \
Expand Down
2 changes: 1 addition & 1 deletion paddle/fluid/eager/backward.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -851,7 +851,7 @@ void Backward(
bool retain_graph) {
VLOG(3) << "Run in Backward";
paddle::platform::RecordEvent backward_record_event(
"backward", paddle::platform::TracerEventType::Operator, 1);
"backward", paddle::platform::TracerEventType::UserDefined, 1);
RunBackward(tensors, grad_tensors, retain_graph);
phi::autotune::AutoTuneStatus::Instance().Update();
}
Expand Down
4 changes: 2 additions & 2 deletions paddle/fluid/platform/profiler/chrometracing_logger.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -603,7 +603,7 @@ void ChromeTracingLogger::StartLog() {
std::string(
R"JSON(
{
"id": %d, "name": "%s", "totalGlobalMem": %u,
"id": %d, "name": "%s", "totalGlobalMem": %llu,
"computeMajor": %d, "computeMinor": %d,
"maxThreadsPerBlock": %d, "maxThreadsPerMultiprocessor": %d,
"regsPerBlock": %d, "regsPerMultiprocessor": %d, "warpSize": %d,
Expand Down Expand Up @@ -633,7 +633,7 @@ void ChromeTracingLogger::StartLog() {
std::string(
R"JSON(
{
"id": %d, "name": "%s", "totalGlobalMem": %u,
"id": %d, "name": "%s", "totalGlobalMem": %llu,
"computeMajor": %d, "computeMinor": %d,
"maxThreadsPerBlock": %d, "maxThreadsPerMultiprocessor": %d,
"regsPerBlock": %d, "regsPerMultiprocessor": %d, "warpSize": %d,
Expand Down
22 changes: 11 additions & 11 deletions python/paddle/fluid/tests/unittests/test_profiler_statistic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,9 +82,9 @@ def test_statistic_case1(self):
profiler.TracerEventType.Forward, 50, 110,
1000, 1001)

userdefined_node = HostPythonNode('Communication Time',
profiler.TracerEventType.UserDefined,
100, 110, 1000, 1001)
userdefined_node = HostPythonNode(
'Communication Time', profiler.TracerEventType.PythonUserDefined,
100, 110, 1000, 1001)

communication_node = HostPythonNode(
'Communication', profiler.TracerEventType.Communication, 105, 110,
Expand Down Expand Up @@ -209,7 +209,7 @@ def test_statistic_case1(self):
0, profiler.TracerEventType.Memcpy), 60)
self.assertEqual(
time_range_summary.get_cpu_range_sum(
profiler.TracerEventType.UserDefined), 25)
profiler.TracerEventType.UserDefined), 15)
self.assertEqual(
time_range_summary.get_cpu_range_sum(
profiler.TracerEventType.Communication), 5)
Expand Down Expand Up @@ -277,9 +277,9 @@ def test_statistic_case2(self):
profiler.TracerEventType.Forward, 50, 110,
1000, 1001)

userdefined_node = HostPythonNode('Communication Time',
profiler.TracerEventType.UserDefined,
100, 110, 1000, 1001)
userdefined_node = HostPythonNode(
'Communication Time', profiler.TracerEventType.PythonUserDefined,
100, 110, 1000, 1001)
allreduce_launchkernel0 = HostPythonNode(
'cudalaunchkernel', profiler.TracerEventType.CudaRuntime, 102, 104,
1000, 1001)
Expand Down Expand Up @@ -451,7 +451,7 @@ def test_statistic_case2(self):
0, profiler.TracerEventType.Memcpy), 60)
self.assertEqual(
time_range_summary.get_cpu_range_sum(
profiler.TracerEventType.UserDefined), 25)
profiler.TracerEventType.UserDefined), 15)
self.assertEqual(
time_range_summary.get_cpu_range_sum(
profiler.TracerEventType.Communication), 5)
Expand Down Expand Up @@ -518,9 +518,9 @@ def test_statistic_case3(self):
optimization_node = HostPythonNode(
'Optimization', profiler.TracerEventType.Optimization, 220, 300,
1000, 1001)
userdefined_node = HostPythonNode('Communication Time',
profiler.TracerEventType.UserDefined,
60, 70, 1000, 1001)
userdefined_node = HostPythonNode(
'Communication Time', profiler.TracerEventType.PythonUserDefined,
60, 70, 1000, 1001)

conv2d_node = HostPythonNode('conv2d',
profiler.TracerEventType.Operator, 25, 25,
Expand Down
3 changes: 2 additions & 1 deletion python/paddle/profiler/profiler_statistic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -514,7 +514,8 @@ def parse(self, nodetrees):
or 'memset' in host_statistic_node.name.lower():
self.add_memory_manipulation_item(host_statistic_node)
else:
self.add_userdefined_item(host_statistic_node)
if host_statistic_node.type == TracerEventType.PythonUserDefined:
self.add_userdefined_item(host_statistic_node)
self.add_kernel_item(host_statistic_nodes[0])

for threadid, root_statistic_node in node_statistic_trees.items():
Expand Down