experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
STT-tensorflow/tensorflow/core/distributed_runtime
History
Chris Kennelly c04bf06bfc Optimize calls to std::string::find() and friends for a single char. 
The character literal overload is more efficient.

PiperOrigin-RevId: 348124169
Change-Id: I55909265a8267017210eb0deff5091da20d8ed70
2020-12-17 17:48:51 -08:00
..
eager
Fix an issue of out of order execution. For a multi-device function, don't send a packed input to the function device until all underlying remote handles are ready on remote devices. Otherwise, on a remote worker, a remote component function execution request could be enqueued before a request for producing a function input.
2020-12-11 14:56:03 -08:00
rpc
Optimize calls to std::string::find() and friends for a single char.
2020-12-17 17:48:51 -08:00
base_rendezvous_mgr.cc
Fix cancellation race condition in BaseRendezvousMgr::RegisterCall
2020-06-19 13:04:53 -07:00
base_rendezvous_mgr.h
Fix cancellation race condition in BaseRendezvousMgr::RegisterCall
2020-06-19 13:04:53 -07:00
BUILD
Support aborting RING communication in multi worker collectives
2020-10-21 17:03:09 -07:00
call_options_test.cc
…
call_options.cc
…
call_options.h
…
cancellable_call.cc
Support aborting RING communication in multi worker collectives
2020-10-21 17:03:09 -07:00
cancellable_call.h
Support aborting RING communication in multi worker collectives
2020-10-21 17:03:09 -07:00
cluster_function_library_runtime_test.cc
Pass in GrpcWorkerEnv when creating GrpcWorkerCache.
2020-06-04 11:46:09 -07:00
cluster_function_library_runtime.cc
Change the function output type, either a Tensor for a local output or a TensorShape for a remote output, preparing for the support of function outputs placed on remote workers.
2020-08-04 19:13:03 -07:00
cluster_function_library_runtime.h
Use the original output indices when adding a component function output to RemoteMgr.
2020-08-19 14:41:05 -07:00
collective_param_resolver_distributed_test.cc
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00
collective_param_resolver_distributed.cc
Support aborting param resolution in multi worker collectives
2020-10-21 18:40:58 -07:00
collective_param_resolver_distributed.h
Support aborting param resolution in multi worker collectives
2020-10-21 18:40:58 -07:00
collective_rma_distributed_test.cc
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00
collective_rma_distributed.cc
Support aborting RING communication in multi worker collectives
2020-10-21 17:03:09 -07:00
collective_rma_distributed.h
Support aborting RING communication in multi worker collectives
2020-10-21 17:03:09 -07:00
device_resolver_distributed_test.cc
Use device attributes from group resolution
2020-09-09 10:53:49 -07:00
device_resolver_distributed.cc
Use device attributes from group resolution
2020-09-09 10:53:49 -07:00
device_resolver_distributed.h
Use device attributes from group resolution
2020-09-09 10:53:49 -07:00
graph_mgr.cc
[TF2XLA] Remove the serialization of CustomKernelCreator, since there is only one, and we won't add new ones
2020-09-22 10:59:59 -07:00
graph_mgr.h
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
local_master.cc
…
local_master.h
…
master_env.h
Fix two memory leaks and enable asan for C API remote tests.
2020-07-17 10:17:20 -07:00
master_interface.h
…
master_session.cc
fix typos in core directory
2020-10-29 02:52:55 +03:00
master_session.h
…
master_test.cc
…
master.cc
…
master.h
…
message_wrappers_test.cc
…
message_wrappers.cc
Support TensorProtos as Operation inputs, in order to support remote inputs passed as Tensors to EagerClusterFunctionLibraryRuntime::Run.
2020-03-09 17:10:38 -07:00
message_wrappers.h
…
partial_run_mgr_test.cc
…
partial_run_mgr.cc
…
partial_run_mgr.h
…
README.md
…
recent_request_ids_test.cc
Internal tests cleanup.
2020-10-27 13:24:35 -07:00
recent_request_ids.cc
…
recent_request_ids.h
…
remote_device_test.cc
Pass in GrpcWorkerEnv when creating GrpcWorkerCache.
2020-06-04 11:46:09 -07:00
remote_device.cc
replace PFLR DeviceGetContext hardcode with Device::IsRemoteCallAllowed
2020-10-22 20:49:03 +02:00
remote_device.h
…
rendezvous_mgr_interface.h
[Cleanup] Remove unused method RendezvousMgrInterface::CleanupAll().
2020-04-13 11:26:06 -07:00
request_id_test.cc
…
request_id.cc
…
request_id.h
…
rpc_collective_executor_mgr_test.cc
Make NcclManager part of CollectiveExecutorMgr
2020-09-17 14:35:16 -07:00
rpc_collective_executor_mgr.cc
Make NcclManager part of CollectiveExecutorMgr
2020-09-17 14:35:16 -07:00
rpc_collective_executor_mgr.h
Make NcclManager part of CollectiveExecutorMgr
2020-09-17 14:35:16 -07:00
rpcbench_test.cc
Internal tests cleanup.
2020-10-27 13:24:35 -07:00
scheduler.cc
…
scheduler.h
…
server_lib_test.cc
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
server_lib.cc
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
server_lib.h
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
session_mgr_test.cc
Garbage collect old WorkerSession when the restarted master task create new one.
2020-08-03 11:31:26 -07:00
session_mgr.cc
Garbage collect old WorkerSession when the restarted master task create new one.
2020-08-03 11:31:26 -07:00
session_mgr.h
Garbage collect old WorkerSession when the restarted master task create new one.
2020-08-03 11:31:26 -07:00
tensor_coding_test.cc
Internal tests cleanup.
2020-10-27 13:24:35 -07:00
tensor_coding.cc
…
tensor_coding.h
…
test_utils.h
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00
worker_cache_logger.cc
…
worker_cache_logger.h
…
worker_cache_partial.cc
…
worker_cache_partial.h
…
worker_cache_wrapper.h
…
worker_cache.h
…
worker_env.h
Fix two memory leaks and enable asan for C API remote tests.
2020-07-17 10:17:20 -07:00
worker_interface.h
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00
worker_session.cc
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
worker_session.h
When calling connect_to_cluser, if the options are identical and there is no renaming of local device, reuse existing local DeviceManager, otherwise we keep the old DeviceManager around to allow the old Tensor created to be usable.
2020-05-20 08:53:52 -07:00
worker.cc
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00
worker.h
Set a timeout to check health RPC
2020-10-21 13:02:25 -07:00

README.md

Distributed TensorFlow

This directory contains the initial open-source implementation of the distributed TensorFlow runtime, using gRPC for inter-process communication.

To learn how to use the distributed runtime to create a TensorFlow cluster, see the Distributed TensorFlow How-To.