experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Refactor to extract XPlane to tool converters.

Browse Source 
PiperOrigin-RevId: 321865397
Change-Id: Ifa496f60df649e68a4bc647c1417a065cdcf141a
This commit is contained in:
A. Unique TensorFlower 2020-07-17 15:19:04 -07:00 committed by TensorFlower Gardener
parent ec25e31817
commit a44a11793c
17 changed files with 199 additions and 92 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/core/profiler/convert/BUILD
View File

@ -102,6 +102,7 @@ cc_library(
"//tensorflow/core/profiler/protobuf:steps_db_proto_cc",
"//tensorflow/core/profiler/protobuf:tf_function_proto_cc",
"//tensorflow/core/profiler/utils:diagnostics",
"//tensorflow/core/profiler/utils:hardware_type_utils",
"//tensorflow/core/profiler/utils:html_utils",
"//tensorflow/core/profiler/utils:math_utils",
"//tensorflow/core/profiler/utils:op_metrics_db_utils",
@ -128,6 +129,7 @@ cc_library(
"//tensorflow/core/profiler/protobuf:steps_db_proto_cc",
"//tensorflow/core/profiler/utils:diagnostics",
"//tensorflow/core/profiler/utils:event_span",
"//tensorflow/core/profiler/utils:hardware_type_utils",
"//tensorflow/core/profiler/utils:html_utils",
"//tensorflow/core/profiler/utils:math_utils",
"//tensorflow/core/profiler/utils:tf_op_utils",
@ -303,7 +305,6 @@ cc_library(
"//tensorflow/core/profiler/protobuf:xplane_proto_cc",
"//tensorflow/core/profiler/rpc/client:save_profile",
"//tensorflow/core/profiler/utils:xplane_schema",
"//tensorflow/core/profiler/utils:xplane_utils",
"@com_google_absl//absl/container:flat_hash_set",
"@com_google_absl//absl/strings",
],
@ -505,6 +506,7 @@ cc_library(
"//tensorflow/core/profiler/protobuf:xplane_proto_cc",
"//tensorflow/core/profiler/utils:tf_xplane_visitor",
"//tensorflow/core/profiler/utils:xplane_schema",
"//tensorflow/core/profiler/utils:xplane_utils",
"//tensorflow/core/profiler/utils:xplane_visitor",
"@com_google_absl//absl/algorithm:container",
"@com_google_absl//absl/container:flat_hash_map",

6
tensorflow/core/profiler/convert/op_stats_to_input_pipeline_analysis.cc
View File

@ -38,6 +38,7 @@ limitations under the License.
#include "tensorflow/core/profiler/protobuf/steps_db.pb.h"
#include "tensorflow/core/profiler/utils/diagnostics.h"
#include "tensorflow/core/profiler/utils/event_span.h"
#include "tensorflow/core/profiler/utils/hardware_type_utils.h"
#include "tensorflow/core/profiler/utils/html_utils.h"
#include "tensorflow/core/profiler/utils/math_utils.h"
#include "tensorflow/core/profiler/utils/tf_op_utils.h"
@ -553,12 +554,13 @@ StepSummary ComputeStepTimeSummaryInMs(
}
InputPipelineAnalysisResult ConvertOpStatsToInputPipelineAnalysis(
const OpStats& op_stats, const HardwareType& hardware_type) {
const OpStats& op_stats) {
InputPipelineAnalysisResult result =
ComputeGenericInputPipelineAnalysisResult(
op_stats.step_db().step_sequence());
PopulateStepDiagnostics(op_stats, result.mutable_diagnostics());
result.set_hardware_type(HardwareType_Name(hardware_type));
result.set_hardware_type(HardwareType_Name(
ParseHardwareType(op_stats.run_environment().device_type())));
GenerateHostResult(op_stats.host_op_metrics_db(), &result);
InputPipelineAnalysisRecommendation recommendation = GenerateRecommendation();

2
tensorflow/core/profiler/convert/op_stats_to_input_pipeline_analysis.h
View File

@ -59,7 +59,7 @@ BottleneckAnalysis ComputeBottleneckAnalysis(
any_step_details);
InputPipelineAnalysisResult ConvertOpStatsToInputPipelineAnalysis(
const OpStats& op_stats, const HardwareType& hardware_type);
const OpStats& op_stats);
// Returns true if explanation for "All Others" time is also included in
// input_statement.

9
tensorflow/core/profiler/convert/op_stats_to_overview_page.cc
View File

@ -31,6 +31,7 @@ limitations under the License.
#include "tensorflow/core/profiler/protobuf/steps_db.pb.h"
#include "tensorflow/core/profiler/protobuf/tf_function.pb.h"
#include "tensorflow/core/profiler/utils/diagnostics.h"
#include "tensorflow/core/profiler/utils/hardware_type_utils.h"
#include "tensorflow/core/profiler/utils/html_utils.h"
#include "tensorflow/core/profiler/utils/math_utils.h"
#include "tensorflow/core/profiler/utils/op_metrics_db_utils.h"
@ -316,14 +317,13 @@ std::string EagerRecommendationHtml(double host_op_time_eager_percent,
return recommendation;
}
OverviewPage ConvertOpStatsToOverviewPage(const OpStats& op_stats,
HardwareType hardware_type) {
OverviewPage ConvertOpStatsToOverviewPage(const OpStats& op_stats) {
OverviewPage overview_page;
*overview_page.mutable_run_environment() =
ComputeRunEnvironment(op_stats.run_environment());
*overview_page.mutable_analysis() = ComputeAnalysisResult(op_stats);
*overview_page.mutable_input_analysis() =
ConvertOpStatsToInputPipelineAnalysis(op_stats, hardware_type);
ConvertOpStatsToInputPipelineAnalysis(op_stats);
BottleneckAnalysis bottleneck = ComputeBottleneckAnalysis(
overview_page.input_analysis().input_time_breakdown(),
overview_page.input_analysis().step_details());
@ -331,7 +331,8 @@ OverviewPage ConvertOpStatsToOverviewPage(const OpStats& op_stats,
bottleneck, op_stats.device_op_metrics_db().precision_stats());
SetCommonRecommendation(
bottleneck.input_classification(), bottleneck.input_statement(), "",
hardware_type, TfFunctionRecommendationHtml(op_stats.tf_function_db()),
ParseHardwareType(op_stats.run_environment().device_type()),
TfFunctionRecommendationHtml(op_stats.tf_function_db()),
EagerRecommendationHtml(
overview_page.analysis().host_op_time_eager_percent(),
overview_page.analysis().device_op_time_eager_percent()),

3
tensorflow/core/profiler/convert/op_stats_to_overview_page.h
View File

@ -54,8 +54,7 @@ OverviewPageAnalysis ComputeAnalysisResult(const OpStats& op_stats);
OverviewPageRunEnvironment ComputeRunEnvironment(
const RunEnvironment& run_environment);
OverviewPage ConvertOpStatsToOverviewPage(const OpStats& op_stats,
HardwareType hardware_type);
OverviewPage ConvertOpStatsToOverviewPage(const OpStats& op_stats);
// Returns a html which provides tf-function related recommendation.
std::string TfFunctionRecommendationHtml(const TfFunctionDb& tf_function_db);

2
tensorflow/core/profiler/convert/op_stats_to_tf_stats_test.cc
View File

@ -80,7 +80,7 @@ TEST(OpStatsToTfStats, GpuTfStats) {
kKernel3DurationNs, /*on_device=*/true, kKernel3,
&device_plane, &stream2);
const OpStats op_stats = ConvertXSpaceToOpStats(space);
const OpStats op_stats = ConvertXSpaceToOpStats(space, {OP_METRICS_DB});
const TfStatsDatabase tf_stats = ConvertOpStatsToTfStats(op_stats);
// TfOp1, TfOp2, Idle

28
tensorflow/core/profiler/convert/xplane_to_memory_profile.cc
View File

@ -35,6 +35,7 @@ limitations under the License.
#include "tensorflow/core/profiler/protobuf/xplane.pb.h"
#include "tensorflow/core/profiler/utils/tf_xplane_visitor.h"
#include "tensorflow/core/profiler/utils/xplane_schema.h"
#include "tensorflow/core/profiler/utils/xplane_utils.h"
#include "tensorflow/core/profiler/utils/xplane_visitor.h"
namespace tensorflow {
@ -481,6 +482,23 @@ void ProcessMemoryProfileProto(int64 max_num_snapshots,
}
}
template <typename Proto>
Status ConvertProtoToJson(const Proto& proto_output, std::string* json_output) {
protobuf::util::JsonPrintOptions json_options;
json_options.always_print_primitive_fields = true;
auto status = protobuf::util::MessageToJsonString(proto_output, json_output,
json_options);
if (!status.ok()) {
// Convert error_msg google::protobuf::StringPiece (or absl::string_view) to
// tensorflow::StringPiece.
auto error_msg = status.message();
return errors::Internal(
"Could not convert proto to JSON string: ",
absl::string_view(error_msg.data(), error_msg.length()));
}
return Status::OK();
}
} // namespace
MemoryProfile ConvertXPlaneToMemoryProfile(const XPlane& host_plane,
@ -490,5 +508,15 @@ MemoryProfile ConvertXPlaneToMemoryProfile(const XPlane& host_plane,
return memory_profile;
}
Status ConvertXSpaceToMemoryProfileJson(const XSpace& xspace,
std::string* json_output) {
if (const XPlane* host_plane =
FindPlaneWithName(xspace, kHostThreadsPlaneName)) {
MemoryProfile memory_profile = ConvertXPlaneToMemoryProfile(*host_plane);
TF_RETURN_IF_ERROR(ConvertProtoToJson(memory_profile, json_output));
}
return Status::OK();
}
} // namespace profiler
} // namespace tensorflow

3
tensorflow/core/profiler/convert/xplane_to_memory_profile.h
View File

@ -16,6 +16,7 @@ limitations under the License.
#ifndef TENSORFLOW_CORE_PROFILER_CONVERT_XPLANE_TO_MEMORY_PROFILE_H_
#define TENSORFLOW_CORE_PROFILER_CONVERT_XPLANE_TO_MEMORY_PROFILE_H_
#include "tensorflow/core/platform/status.h"
#include "tensorflow/core/platform/types.h"
#include "tensorflow/core/profiler/protobuf/memory_profile.pb.h"
#include "tensorflow/core/profiler/protobuf/xplane.pb.h"
@ -29,6 +30,8 @@ namespace profiler {
MemoryProfile ConvertXPlaneToMemoryProfile(const XPlane& host_plane,
int64 max_num_snapshots = 1000);
Status ConvertXSpaceToMemoryProfileJson(const XSpace& xspace,
std::string* json_output);
} // namespace profiler
} // namespace tensorflow

71
tensorflow/core/profiler/convert/xplane_to_op_stats.cc
View File

@ -106,7 +106,8 @@ void SetRunEnvironment(int32 accelerator_count, RunEnvironment* env) {
}
void ProcessHostPlane(const XPlane* host_plane, bool use_device_step_events,
OpMetricsDb* op_metrics_db, StepEvents* step_events) {
const OpStatsConfig& config, OpMetricsDb* op_metrics_db,
StepEvents* step_events) {
absl::flat_hash_map<int64, TfOp> tf_ops =
CollectTfOpsFromHostThreadsXPlane(*host_plane);
OpMetricsDbCombiner combiner(op_metrics_db);
@ -114,9 +115,11 @@ void ProcessHostPlane(const XPlane* host_plane, bool use_device_step_events,
plane.ForEachLine([&](const XLineVisitor& line) {
ConsumeTfMetricsDbData(
ConvertHostThreadsXLineToTfMetricsDbData(line, tf_ops), &combiner);
CombineStepEvents(ConvertHostThreadsXLineToStepEvents(
line, use_device_step_events, *step_events),
step_events);
if (config.contains(STEP_DB)) {
CombineStepEvents(ConvertHostThreadsXLineToStepEvents(
line, use_device_step_events, *step_events),
step_events);
}
});
}
@ -138,7 +141,8 @@ void PropagateXSpaceDiagnosticsToOpStats(const XSpace& space,
}
}
OpStats ConvertXSpaceToOpStats(const XSpace& space) {
OpStats ConvertXSpaceToOpStats(const XSpace& space,
const OpStatsConfig& config) {
const XPlane* host_plane = FindPlaneWithName(space, kHostThreadsPlaneName);
std::vector<const XPlane*> device_planes =
FindPlanesWithPrefix(space, kGpuPlanePrefix);
@ -152,34 +156,45 @@ OpStats ConvertXSpaceToOpStats(const XSpace& space) {
std::vector<KernelReport> reports;
for (const XPlane* device_trace : device_planes) {
if (!op_stats.has_perf_env()) {
*op_stats.mutable_perf_env() = GetPerfEnvFromXPlane(*device_trace);
if (config.contains(OP_METRICS_DB)) {
if (!op_stats.has_perf_env()) {
*op_stats.mutable_perf_env() = GetPerfEnvFromXPlane(*device_trace);
}
const PerfEnv& perf_env = op_stats.perf_env();
OpMetricsDb device_op_metrics_db = ConvertDeviceTraceXPlaneToOpMetricsDb(
*device_trace, perf_env.peak_tera_flops_per_second(),
perf_env.peak_hbm_bw_giga_bytes_per_second());
op_metrics_db_combiner.Combine(device_op_metrics_db);
}
if (config.contains(STEP_DB)) {
CombineStepEvents(ConvertDeviceTraceXPlaneToStepEvents(*device_trace),
&step_events);
}
if (config.contains(KERNEL_STATS_DB)) {
KernelStatsDb kernel_stats_db = ConvertDeviceTraceXPlaneToKernelStatsDb(
*device_trace, /*on_kernel_fn=*/{});
reports.insert(reports.begin(), kernel_stats_db.reports().begin(),
kernel_stats_db.reports().end());
}
const PerfEnv& perf_env = op_stats.perf_env();
OpMetricsDb device_op_metrics_db = ConvertDeviceTraceXPlaneToOpMetricsDb(
*device_trace, perf_env.peak_tera_flops_per_second(),
perf_env.peak_hbm_bw_giga_bytes_per_second());
op_metrics_db_combiner.Combine(device_op_metrics_db);
CombineStepEvents(ConvertDeviceTraceXPlaneToStepEvents(*device_trace),
&step_events);
KernelStatsDb kernel_stats_db = ConvertDeviceTraceXPlaneToKernelStatsDb(
*device_trace, /*on_kernel_fn=*/{});
reports.insert(reports.begin(), kernel_stats_db.reports().begin(),
kernel_stats_db.reports().end());
}
GroupKernelReports(&reports, op_stats.mutable_kernel_stats_db());
SortKernelsByTotalDurationDesc(op_stats.mutable_kernel_stats_db());
// Convert a host plane.
if (config.contains(KERNEL_STATS_DB)) {
GroupKernelReports(&reports, op_stats.mutable_kernel_stats_db());
SortKernelsByTotalDurationDesc(op_stats.mutable_kernel_stats_db());
}
bool has_device = !device_planes.empty();
if (host_plane) {
ProcessHostPlane(host_plane, has_device,
// Convert a host plane.
if (host_plane && config.contains(OP_METRICS_DB)) {
ProcessHostPlane(host_plane, has_device, config,
op_stats.mutable_host_op_metrics_db(), &step_events);
}
StepEvents nonoverlapped_step_events = ToNonOverlappedStepEvents(step_events);
*op_stats.mutable_step_db() =
ConvertStepEventsToStepDb(has_device, nonoverlapped_step_events);
*op_stats.mutable_device_op_metrics_db()->mutable_precision_stats() =
ComputePrecisionStats(nonoverlapped_step_events);
if (config.contains(STEP_DB)) {
StepEvents nonoverlapped_step_events =
ToNonOverlappedStepEvents(step_events);
*op_stats.mutable_step_db() =
ConvertStepEventsToStepDb(has_device, nonoverlapped_step_events);
*op_stats.mutable_device_op_metrics_db()->mutable_precision_stats() =
ComputePrecisionStats(nonoverlapped_step_events);
}
return op_stats;
}

12
tensorflow/core/profiler/convert/xplane_to_op_stats.h
View File

@ -16,14 +16,24 @@ limitations under the License.
#ifndef TENSORFLOW_CORE_PROFILER_CONVERT_XPLANE_TO_OP_STATS_H_
#define TENSORFLOW_CORE_PROFILER_CONVERT_XPLANE_TO_OP_STATS_H_
#include "absl/container/flat_hash_set.h"
#include "tensorflow/core/profiler/protobuf/op_stats.pb.h"
#include "tensorflow/core/profiler/protobuf/xplane.pb.h"
namespace tensorflow {
namespace profiler {
enum OpStatsKind {
OP_METRICS_DB,
STEP_DB,
KERNEL_STATS_DB,
};
using OpStatsConfig = absl::flat_hash_set<OpStatsKind>;
// NOTE: call GroupTfEvents before if OpStats.step_db needs to be generated.
OpStats ConvertXSpaceToOpStats(const XSpace& space);
OpStats ConvertXSpaceToOpStats(const XSpace& space,
const OpStatsConfig& config);
// Propagate and dedup the diagnostics in XSpace and add to OpStats.
void PropagateXSpaceDiagnosticsToOpStats(const XSpace& space,

18
tensorflow/core/profiler/convert/xplane_to_op_stats_test.cc
View File

@ -61,8 +61,8 @@ TEST(ConvertXPlaneToOpStats, PerfEnv) {
*device_plane.GetOrCreateStatMetadata("compute_cap_minor"),
absl::StrCat(kComputeCapMinor));
GroupTfEvents(&space, /*event_group_name_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space);
GroupTfEvents(&space, /*group_metadata_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space, {OP_METRICS_DB});
const PerfEnv& perf_env = op_stats.perf_env();
EXPECT_NEAR(141, perf_env.peak_tera_flops_per_second(), kMaxError);
EXPECT_NEAR(900, perf_env.peak_hbm_bw_giga_bytes_per_second(), kMaxError);
@ -76,8 +76,8 @@ TEST(ConvertXPlaneToOpStats, RunEnvironment) {
XPlaneBuilder device_plane2(
GetOrCreateGpuXPlane(&space, /*device_ordinal=*/1));
GroupTfEvents(&space, /*event_group_name_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space);
GroupTfEvents(&space, /*group_metadata_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space, {});
const RunEnvironment& run_env = op_stats.run_environment();
EXPECT_EQ("GPU", run_env.device_type());
@ -106,8 +106,8 @@ TEST(ConvertXPlaneToOpStats, CpuOnlyStepDbTest) {
{{StatType::kStepId, kStepId}});
CreateXEvent(&host_plane_builder, &tf_executor_thread, "matmul", 30, 70);
GroupTfEvents(&space, /*event_group_name_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space);
GroupTfEvents(&space, /*group_metadata_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space, {OP_METRICS_DB, STEP_DB});
const StepDatabaseResult& step_db = op_stats.step_db();
EXPECT_EQ(step_db.step_sequence_size(), 1);
@ -143,8 +143,8 @@ TEST(ConvertXPlaneToOpStats, GpuStepDbTest) {
CreateXEvent(&device_plane_builder, &stream, "matmul", 50, 40,
{{StatType::kCorrelationId, kCorrelationId}});
GroupTfEvents(&space, /*event_group_name_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space);
GroupTfEvents(&space, /*group_metadata_map=*/nullptr);
OpStats op_stats = ConvertXSpaceToOpStats(space, {OP_METRICS_DB, STEP_DB});
const StepDatabaseResult& step_db = op_stats.step_db();
EXPECT_EQ(step_db.step_sequence_size(), 1);
@ -161,7 +161,7 @@ TEST(ConvertXPlaneToOpStats, PropagateAndDedupErrors) {
*space.add_errors() = kError;
*space.add_errors() = kError;
OpStats op_stats = ConvertXSpaceToOpStats(space);
OpStats op_stats = ConvertXSpaceToOpStats(space, {});
EXPECT_EQ(1, op_stats.diagnostics().errors_size());
EXPECT_EQ(kError, op_stats.diagnostics().errors(/*index=*/0));

53
tensorflow/core/profiler/convert/xplane_to_profile_response.cc
View File

@ -30,18 +30,15 @@ limitations under the License.
#include "tensorflow/core/profiler/convert/xplane_to_op_stats.h"
#include "tensorflow/core/profiler/convert/xplane_to_trace_events.h"
#include "tensorflow/core/profiler/profiler_service.pb.h"
#include "tensorflow/core/profiler/protobuf/hardware_types.pb.h"
#include "tensorflow/core/profiler/protobuf/input_pipeline.pb.h"
#include "tensorflow/core/profiler/protobuf/kernel_stats.pb.h"
#include "tensorflow/core/profiler/protobuf/memory_profile.pb.h"
#include "tensorflow/core/profiler/protobuf/op_stats.pb.h"
#include "tensorflow/core/profiler/protobuf/overview_page.pb.h"
#include "tensorflow/core/profiler/protobuf/tf_stats.pb.h"
#include "tensorflow/core/profiler/protobuf/trace_events.pb.h"
#include "tensorflow/core/profiler/protobuf/xplane.pb.h"
#include "tensorflow/core/profiler/rpc/client/save_profile.h"
#include "tensorflow/core/profiler/utils/xplane_schema.h"
#include "tensorflow/core/profiler/utils/xplane_utils.h"
namespace tensorflow {
namespace profiler {
@ -55,12 +52,6 @@ const absl::string_view kKernelStats = "kernel_stats";
const absl::string_view kMemoryProfile = "memory_profile";
const absl::string_view kXPlane = "xplane";
HardwareType HardwareTypeFromRunEnvironment(const RunEnvironment& run_env) {
if (run_env.device_type() == "GPU") return HardwareType::GPU;
if (run_env.device_type() == "CPU") return HardwareType::CPU_ONLY;
return HardwareType::UNKNOWN_HARDWARE;
}
template <typename Proto>
void AddToolData(absl::string_view tool_name, const Proto& tool_output,
ProfileResponse* response) {
@ -69,23 +60,6 @@ void AddToolData(absl::string_view tool_name, const Proto& tool_output,
tool_output.SerializeToString(tool_data->mutable_data());
}
template <typename Proto>
Status ConvertProtoToJson(const Proto& proto_output, std::string* json_output) {
protobuf::util::JsonPrintOptions json_options;
json_options.always_print_primitive_fields = true;
auto status = protobuf::util::MessageToJsonString(proto_output, json_output,
json_options);
if (!status.ok()) {
// Convert error_msg google::protobuf::StringPiece (or absl::string_view) to
// tensorflow::StringPiece.
auto error_msg = status.message();
return errors::Internal(
"Could not convert proto to JSON string: ",
absl::string_view(error_msg.data(), error_msg.length()));
}
return Status::OK();
}
// Returns the tool name with extension.
std::string ToolName(absl::string_view tool) {
if (tool == kTraceViewer) return "trace.json.gz";
@ -115,21 +89,18 @@ Status ConvertXSpaceToProfileResponse(const XSpace& xspace,
// Trace viewer is the only tool, skip OpStats conversion.
if (tools.size() == 1) return Status::OK();
}
OpStats op_stats = ConvertXSpaceToOpStats(xspace);
HardwareType hw_type =
HardwareTypeFromRunEnvironment(op_stats.run_environment());
OpStats op_stats =
ConvertXSpaceToOpStats(xspace, {OP_METRICS_DB, STEP_DB, KERNEL_STATS_DB});
if (tools.contains(kOverviewPage)) {
OverviewPage overview_page_db =
ConvertOpStatsToOverviewPage(op_stats, hw_type);
OverviewPage overview_page_db = ConvertOpStatsToOverviewPage(op_stats);
AddToolData(ToolName(kOverviewPage), overview_page_db, response);
if (tools.contains(kInputPipeline)) {
AddToolData(ToolName(kInputPipeline), overview_page_db.input_analysis(),
response);
}
} else if (tools.contains(kInputPipeline)) {
InputPipelineAnalysisResult input_pipeline_analysis =
ConvertOpStatsToInputPipelineAnalysis(op_stats, hw_type);
AddToolData(ToolName(kInputPipeline), input_pipeline_analysis, response);
AddToolData(ToolName(kInputPipeline),
ConvertOpStatsToInputPipelineAnalysis(op_stats), response);
}
if (tools.contains(kTensorflowStats)) {
TfStatsDatabase tf_stats_db = ConvertOpStatsToTfStats(op_stats);
@ -139,15 +110,11 @@ Status ConvertXSpaceToProfileResponse(const XSpace& xspace,
AddToolData(ToolName(kKernelStats), op_stats.kernel_stats_db(), response);
}
if (tools.contains(kMemoryProfile)) {
if (const XPlane* host_plane =
FindPlaneWithName(xspace, kHostThreadsPlaneName)) {
MemoryProfile memory_profile = ConvertXPlaneToMemoryProfile(*host_plane);
std::string json_output;
TF_RETURN_IF_ERROR(ConvertProtoToJson(memory_profile, &json_output));
TF_RETURN_IF_ERROR(SaveGzippedToolDataToTensorboardProfile(
req.repository_root(), req.session_id(), req.host_name(),
ToolName(kMemoryProfile), json_output));
}
std::string json_output;
TF_RETURN_IF_ERROR(ConvertXSpaceToMemoryProfileJson(xspace, &json_output));
TF_RETURN_IF_ERROR(SaveGzippedToolDataToTensorboardProfile(
req.repository_root(), req.session_id(), req.host_name(),
ToolName(kMemoryProfile), json_output));
}
return Status::OK();
}

1
tensorflow/core/profiler/utils/BUILD
View File

@ -47,6 +47,7 @@ cc_library(
deps = [
"//tensorflow/core:lib",
"//tensorflow/core/profiler/protobuf:hardware_types_proto_cc",
"@com_google_absl//absl/strings",
],
)

8
tensorflow/core/profiler/utils/hardware_type_utils.cc
View File

@ -74,5 +74,13 @@ double GetFlopMaxThroughputPerSM(const DeviceCapabilities& device_cap) {
device_cap.clock_rate_in_ghz();
}
HardwareType ParseHardwareType(absl::string_view device_type) {
if (device_type == "GPU" || device_type == "Nvidia GPU")
return HardwareType::GPU;
if (device_type == "CPU") return HardwareType::CPU_ONLY;
if (device_type == "TPU") return HardwareType::TPU;
return HardwareType::UNKNOWN_HARDWARE;
}
} // namespace profiler
} // namespace tensorflow

3
tensorflow/core/profiler/utils/hardware_type_utils.h
View File

@ -16,6 +16,7 @@ limitations under the License.
#ifndef TENSORFLOW_CORE_PROFILER_UTILS_HARDWARE_TYPE_UTILS_H_
#define TENSORFLOW_CORE_PROFILER_UTILS_HARDWARE_TYPE_UTILS_H_
#include "absl/strings/string_view.h"
#include "tensorflow/core/profiler/protobuf/hardware_types.pb.h"
namespace tensorflow {
@ -25,6 +26,8 @@ namespace profiler {
// streaming multiprocessor.
double GetFlopMaxThroughputPerSM(const DeviceCapabilities& device_cap);
HardwareType ParseHardwareType(absl::string_view device_type);
} // namespace profiler
} // namespace tensorflow

2
tensorflow/python/profiler/internal/BUILD
View File

@ -119,6 +119,8 @@ tf_python_pybind_extension(
deps = [
"//tensorflow/core:lib",
"//tensorflow/core/profiler:profiler_service_proto_cc",
"//tensorflow/core/profiler/convert:op_stats_to_tf_stats",
"//tensorflow/core/profiler/convert:xplane_to_op_stats",
"//tensorflow/core/profiler/convert:xplane_to_profile_response",
"//tensorflow/core/profiler/convert:xplane_to_trace_events",
"//tensorflow/core/profiler/lib:profiler_session_headers",

66
tensorflow/python/profiler/internal/profiler_wrapper.cc
View File

@ -20,9 +20,16 @@ limitations under the License.
#include "pybind11/pytypes.h"
#include "tensorflow/core/platform/host_info.h"
#include "tensorflow/core/platform/types.h"
#include "tensorflow/core/profiler/convert/op_stats_to_input_pipeline_analysis.h"
#include "tensorflow/core/profiler/convert/op_stats_to_overview_page.h"
#include "tensorflow/core/profiler/convert/op_stats_to_tf_stats.h"
#include "tensorflow/core/profiler/convert/xplane_to_memory_profile.h"
#include "tensorflow/core/profiler/convert/xplane_to_op_stats.h"
#include "tensorflow/core/profiler/convert/xplane_to_profile_response.h"
#include "tensorflow/core/profiler/convert/xplane_to_trace_events.h"
#include "tensorflow/core/profiler/lib/profiler_session.h"
#include "tensorflow/core/profiler/protobuf/input_pipeline.pb.h"
#include "tensorflow/core/profiler/protobuf/kernel_stats.pb.h"
#include "tensorflow/core/profiler/rpc/client/capture_profile.h"
#include "tensorflow/core/profiler/rpc/client/save_profile.h"
#include "tensorflow/core/profiler/rpc/profiler_server.h"
@ -32,6 +39,10 @@ namespace py = ::pybind11;
namespace {
using ::tensorflow::profiler::KERNEL_STATS_DB;
using ::tensorflow::profiler::OP_METRICS_DB;
using ::tensorflow::profiler::STEP_DB;
tensorflow::ProfileRequest MakeProfileRequest(
const tensorflow::string& logdir, const tensorflow::string& session_id,
const tensorflow::string& host) {
@ -162,4 +173,59 @@ PYBIND11_MODULE(_pywrap_profiler, m) {
tensorflow::MaybeRaiseRegisteredFromStatus(status);
return content;
});
m.def("xspace_to_trace_events", [](const py::bytes& serialized_xspace_proto) {
tensorflow::string content;
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
tensorflow::profiler::ConvertXSpaceToTraceEventsString(xspace, &content);
return py::bytes(content);
});
m.def("xspace_to_overview_page",
[](const py::bytes& serialized_xspace_proto) {
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
tensorflow::profiler::OverviewPage overview_page =
tensorflow::profiler::ConvertOpStatsToOverviewPage(
ConvertXSpaceToOpStats(xspace, {OP_METRICS_DB, STEP_DB}));
return py::bytes(overview_page.SerializeAsString());
});
m.def("xspace_to_input_pipeline",
[](const py::bytes& serialized_xspace_proto) {
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
tensorflow::profiler::InputPipelineAnalysisResult input_pipeline =
tensorflow::profiler::ConvertOpStatsToInputPipelineAnalysis(
ConvertXSpaceToOpStats(xspace, {OP_METRICS_DB, STEP_DB}));
return py::bytes(input_pipeline.SerializeAsString());
});
m.def("xspace_to_tf_stats", [](const py::bytes& serialized_xspace_proto) {
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
tensorflow::profiler::TfStatsDatabase tf_stats_db =
tensorflow::profiler::ConvertOpStatsToTfStats(
ConvertXSpaceToOpStats(xspace, {OP_METRICS_DB}));
return py::bytes(tf_stats_db.SerializeAsString());
});
m.def("xspace_to_kernel_stats", [](const py::bytes& serialized_xspace_proto) {
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
tensorflow::profiler::OpStats op_stats =
ConvertXSpaceToOpStats(xspace, {KERNEL_STATS_DB});
return py::bytes(op_stats.kernel_stats_db().SerializeAsString());
});
m.def("xspace_to_memory_profile",
[](const py::bytes& serialized_xspace_proto) {
tensorflow::profiler::XSpace xspace;
xspace.ParseFromString(serialized_xspace_proto);
std::string json_output;
tensorflow::profiler::ConvertXSpaceToMemoryProfileJson(xspace,
&json_output);
return py::bytes(json_output);
});
};