| .. |
|
control_collectives
|
|
|
|
control_plane
|
Fix clang builds by adding headers (#160252)
|
2025-08-11 09:03:14 +00:00 |
|
cuda
|
[AsyncTP] Fixes AsyncMM (#162040)
|
2025-09-08 10:53:59 +00:00 |
|
quantization
|
|
|
|
symm_mem
|
[SymmMem] Better tuning of A2AV based on accurate node boundary (#162003)
|
2025-09-09 04:18:17 +00:00 |
|
Backend.cpp
|
|
|
|
Backend.hpp
|
[fr] [xpu] Add FlightRecorder support for ProcessGroupXCCL (#158568)
|
2025-08-22 09:03:35 +00:00 |
|
Backoff.cpp
|
|
|
|
Backoff.hpp
|
|
|
|
c10d.h
|
|
|
|
comm.cpp
|
|
|
|
comm.hpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
debug.cpp
|
|
|
|
debug.h
|
|
|
|
default_comm_hooks.cpp
|
|
|
|
default_comm_hooks.hpp
|
|
|
|
error.h
|
|
|
|
exception.h
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
FakeProcessGroup.hpp
|
Allow controlling PG backend and options via init_device_mesh (#159371)
|
2025-08-05 12:44:14 +00:00 |
|
FileStore.cpp
|
[ROCm/Windows] Fix build failures and support some BLAS calls (#161981)
|
2025-09-03 20:26:14 +00:00 |
|
FileStore.hpp
|
c10d/Store: add clone feature (#150966) (#150966) (#151045)
|
2025-04-11 04:00:23 +00:00 |
|
FlightRecorder.cpp
|
Add missing fstream include to fix std::ofstream compilation error (#162421)
|
2025-09-09 05:14:32 +00:00 |
|
FlightRecorder.hpp
|
[fr] [xpu] Add FlightRecorder support for ProcessGroupXCCL (#158568)
|
2025-08-22 09:03:35 +00:00 |
|
FlightRecorderCuda.cpp
|
[c10d][fr] Split cuda and non-cuda fr logic into two cpp file (#154929)
|
2025-06-03 07:00:14 +00:00 |
|
FlightRecorderDetail.hpp
|
[fr] [xpu] Add FlightRecorder support for ProcessGroupXCCL (#158568)
|
2025-08-22 09:03:35 +00:00 |
|
Functional.cpp
|
support scalar tensor for functional all_gather (#149913)
|
2025-07-25 22:38:08 +00:00 |
|
Functional.hpp
|
[c10d] support dynamic shapes for all_to_all_single_autograd (#157521)
|
2025-07-08 23:19:59 +00:00 |
|
GlooDeviceFactory.cpp
|
gloo: support ibverbs in cmake (#153425)
|
2025-05-13 17:09:00 +00:00 |
|
GlooDeviceFactory.hpp
|
Reapply "ProcessGroupGloo: support lazy_init (#150801)" (#151031)
|
2025-04-11 01:58:35 +00:00 |
|
GroupRegistry.cpp
|
|
|
|
GroupRegistry.hpp
|
|
|
|
HashStore.cpp
|
Revert "Make distributed modules importable even when backend not built (#159889)" (#162568)
|
2025-09-10 04:29:42 +00:00 |
|
HashStore.hpp
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
init.cpp
|
[4/N][SymmMem] Add get_remote_tensor + move up get_buffer and get_signal_pad (#161533)
|
2025-09-01 07:02:06 +00:00 |
|
logger.cpp
|
[DDP] add one option to allow skipping all reduce unused parameters (#151503)
|
2025-04-17 23:30:19 +00:00 |
|
logger.hpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
logging.cpp
|
|
|
|
logging.h
|
|
|
|
NanCheck.cu
|
|
|
|
NanCheck.hpp
|
|
|
|
NCCLUtils.cpp
|
[c10d] Prototype of group_split for dist2 work (#157716)
|
2025-07-14 21:04:12 +00:00 |
|
NCCLUtils.hpp
|
[c10d] Prototype of group_split for dist2 work (#157716)
|
2025-07-14 21:04:12 +00:00 |
|
Ops.cpp
|
[Lint] Update clang-format to 19.1.4 (#153889)
|
2025-05-20 14:12:46 +00:00 |
|
ParamCommsUtils.cpp
|
|
|
|
ParamCommsUtils.hpp
|
|
|
|
PrefixStore.cpp
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
PrefixStore.hpp
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
ProcessGroup.cpp
|
[c10d] Fix setGroupName and setGroupDesc in group_split and merge_remote_group (#159429)
|
2025-07-30 19:55:55 +00:00 |
|
ProcessGroup.hpp
|
[c10d] Cleanup split_group logic using the newly built splitGroup (#158488)
|
2025-07-29 03:27:11 +00:00 |
|
ProcessGroupGloo.cpp
|
[C10d][Gloo] Enable complex datatype support in ProcessGroupGloo (#156633)
|
2025-09-05 21:24:36 +00:00 |
|
ProcessGroupGloo.hpp
|
[fr] [xpu] Add FlightRecorder support for ProcessGroupXCCL (#158568)
|
2025-08-22 09:03:35 +00:00 |
|
ProcessGroupGlooCuda.cpp
|
Pass shared_ptr by value (#161834)
|
2025-08-30 18:00:37 +00:00 |
|
ProcessGroupGlooDetail.hpp
|
[C10d][Gloo] Enable complex datatype support in ProcessGroupGloo (#156633)
|
2025-09-05 21:24:36 +00:00 |
|
ProcessGroupMPI.cpp
|
[c10d] Add _allgather_base , reduce_scatter , and _reduce_scatter_base into ProcessGroupMPI to enable FSDP with MPI backend (#150162)
|
2025-04-14 19:31:38 +00:00 |
|
ProcessGroupMPI.hpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
ProcessGroupNCCL.cpp
|
[C10d][Gloo] Enable complex datatype support in ProcessGroupGloo (#156633)
|
2025-09-05 21:24:36 +00:00 |
|
ProcessGroupNCCL.hpp
|
[nccl symm mem] don't use arg for mempool, correctly use symmetric registration in hooks (#161238)
|
2025-08-25 03:09:32 +00:00 |
|
ProcessGroupUCC.cpp
|
Replace size() checks with empty() (#153805)
|
2025-05-19 16:20:57 +00:00 |
|
ProcessGroupUCC.hpp
|
|
|
|
ProcessGroupWrapper.cpp
|
|
|
|
ProcessGroupWrapper.hpp
|
|
|
|
PyProcessGroup.hpp
|
[c10d] Cleanup split_group logic using the newly built splitGroup (#158488)
|
2025-07-29 03:27:11 +00:00 |
|
python_comm_hook.cpp
|
[BE][Ez] Update deprecated pybind11 functions (#154798)
|
2025-06-01 06:17:50 +00:00 |
|
python_comm_hook.h
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
RankLocal.hpp
|
|
|
|
reducer_cuda.cpp
|
|
|
|
reducer_timer.hpp
|
|
|
|
reducer.cpp
|
[Reducer] Remove custom handling of view tensors for MTIA (#157882)
|
2025-07-11 17:56:45 +00:00 |
|
reducer.hpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
sequence_num.cpp
|
|
|
|
sequence_num.hpp
|
|
|
|
socket_fmt.h
|
|
|
|
socket.cpp
|
Revert "Replace C array with std::array in formatSockAddr (#159812)"
|
2025-08-06 14:55:48 +00:00 |
|
socket.h
|
|
|
|
Store.cpp
|
|
|
|
Store.hpp
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
TCPStore.cpp
|
[TCPStore] Allow ping to be retried (#159165)
|
2025-07-25 20:03:00 +00:00 |
|
TCPStore.hpp
|
c10d/Store: add nonblocking mode to queue_pop (#151485)
|
2025-04-18 02:14:50 +00:00 |
|
TCPStoreBackend.cpp
|
|
|
|
TCPStoreBackend.hpp
|
c10d/Store: add queues (#150969)
|
2025-04-11 19:24:17 +00:00 |
|
TCPStoreLibUvBackend.cpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
TraceUtils.h
|
[c10d][gloo] Integrate vendor generic FR into gloo (#152614)
|
2025-06-03 16:12:54 +00:00 |
|
Types.cpp
|
[C10d][Gloo] Enable complex datatype support in ProcessGroupGloo (#156633)
|
2025-09-05 21:24:36 +00:00 |
|
Types.hpp
|
[C10d][Gloo] Enable complex datatype support in ProcessGroupGloo (#156633)
|
2025-09-05 21:24:36 +00:00 |
|
UCCTracing.cpp
|
Avoid std::chrono::system_clock (#153135)
|
2025-05-08 16:30:29 +00:00 |
|
UCCTracing.hpp
|
|
|
|
UCCUtils.cpp
|
|
|
|
UCCUtils.hpp
|
|
|
|
UnixSockUtils.hpp
|
|
|
|
Utils.cpp
|
|
|
|
Utils.hpp
|
[BE][11/16] fix typos in torch/ (torch/csrc/distributed/) (#156321)
|
2025-06-23 02:57:50 +00:00 |
|
WinSockUtils.hpp
|
|
|
|
Work.cpp
|
Revert "Make distributed modules importable even when backend not built (#159889)" (#162568)
|
2025-09-10 04:29:42 +00:00 |
|
Work.hpp
|
Work: block_current_stream API (#156883)
|
2025-07-08 23:55:46 +00:00 |
dda071587f