| ffa63e57bdd703ae051ae849af5b5a272fca2223 |
|
25-Jan-2018 |
Sanjoy Das <sanjoy@google.com> |
[TF:XLA] Replace most of HloProfilePrinter by a protocol buffer
This change replaces the meat of HloProfilePrinter with a protobuf
HloProfilePrinterData. The original plan was to serialize HloProfilePrinter
into C++ source code and put that in a .cc file along with the string for the
xla::ProgramShape. However, since we now directly serialize xla::ProgramShape
into a .o file, for consistency I think we should do the same thing for
HloProfilePrinter (instead of adding yet another output file to tfcompile).
The change itself is fairly simple, it is large mostly due to the mass renaming
I had to do.
PiperOrigin-RevId: 183158192
/external/tensorflow/tensorflow/compiler/xla/service/hlo_execution_profile_test.cc
|
| 4b636957604faa3361a799dd9d8749a6b85afff7 |
|
22-Nov-2017 |
Sanjoy Das <sanjoy@google.com> |
Place HloProfilePrinter and HloProfileIndexMap in Executable
This refactoring will later allow XlaCompiledCpuFunction to pull out the
HloProfilePrinter from Executable and use that to display the hlo execution
profile. A de/serialized HloProfilePrinter will let AOT compiled binaries
display their Hlo execution profile.
PiperOrigin-RevId: 176689528
/external/tensorflow/tensorflow/compiler/xla/service/hlo_execution_profile_test.cc
|
| 58f31aab00b916a5351781ddd5499219ae3bb210 |
|
18-Nov-2017 |
A. Unique TensorFlower <gardener@tensorflow.org> |
Don't prepend '%' in HloInstruction constructor. And update the ToString and logging.
PiperOrigin-RevId: 176168081
/external/tensorflow/tensorflow/compiler/xla/service/hlo_execution_profile_test.cc
|
| 58c1aaf77721268a4ef87ebd2ab520a6d5a62f79 |
|
14-Nov-2017 |
Sanjoy Das <sanjoy@google.com> |
Split up HloExecutionProfile into a set of re-usable components
The end goal is to have Hlo profiling support in XlaJitCompiledCpuFunction and
eventually AOT compiled XlaCompiledCpuFunction.
This change leaves the HloExecutionProfile interface mostly intact -- internally
it uses the new split out components to do what it did before. However, in
future CLs:
- I'll extract out a HloExecutionProfilePrototype that contains the
HloProfilePrinter, the OwningHloProfilePrinterStaticData and the
HloToProfileIndex. This will then live in the Executable (if profiling is
enabled).
- The HloExecutionProfile for a specific execution will have a pointer to the
parent HloExecutionProfilePrototype, which it will use to paginate
profile_counters_.
- The CPU backend will use the HloToProfileIndex in the
HloExecutionProfilePrototype to map hlo instructions to profile counter
offsets. This will make the indices in the generated code "line up" with the
indices that the HloProfilePrinter expects.
These changes will allow the XlaJitCompiledCpuFunction (and later AOT) clients
to pass in an appropriately sized zeroed buffer to the generated function and
then pass that same buffer to the appropriate HloProfilePrinter to get a
textual Hlo profile.
PiperOrigin-RevId: 175613737
/external/tensorflow/tensorflow/compiler/xla/service/hlo_execution_profile_test.cc
|