Skip to content

Commit

Permalink
update gloo submodule (pytorch#101472)
Browse files Browse the repository at this point in the history 
Ran command: `git submodule update --remote -- third_party/gloo`

This is to pull in changes up to facebookincubator/gloo@31b1f02 to enhance gloo scalability for large distributed training jobs running into ephemeral port exhaustion issues.

The test failure in pytorch#101438 shows the need for that.

Pull Request resolved: pytorch#101472
Approved by: https://github.com/H-Huang
  • Loading branch information
@xunnanxu @pytorchmergebot
xunnanxu authored and pytorchmergebot committed May 17, 2023
1 parent 8c0b148 commit df6acf2
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion third_party/gloo
Submodule gloo updated 29 files
+7 −0 README.md
+4 −4 cmake/Dependencies.cmake
+5 −5 cmake/Modules/Findrccl.cmake
+2 −2 gloo/allgather.cc
+4 −0 gloo/allgather_ring.h
+2 −2 gloo/allgatherv.cc
+3 −3 gloo/common/error.h
+40 −0 gloo/rendezvous/prefix_store.cc
+6 −0 gloo/rendezvous/prefix_store.h
+25 −0 gloo/rendezvous/store.h
+2 −2 gloo/test/allgather_test.cc
+1 −1 gloo/test/allgatherv_test.cc
+7 −4 gloo/test/base_test.h
+7 −0 gloo/transport/tcp/CMakeLists.txt
+101 −0 gloo/transport/tcp/device.cc
+50 −0 gloo/transport/tcp/device.h
+46 −0 gloo/transport/tcp/error.cc
+87 −0 gloo/transport/tcp/error.h
+171 −0 gloo/transport/tcp/helpers.h
+121 −0 gloo/transport/tcp/listener.cc
+74 −0 gloo/transport/tcp/listener.h
+64 −0 gloo/transport/tcp/loop.cc
+32 −0 gloo/transport/tcp/loop.h
+71 −226 gloo/transport/tcp/pair.cc
+3 −26 gloo/transport/tcp/pair.h
+183 −0 gloo/transport/tcp/socket.cc
+97 −0 gloo/transport/tcp/socket.h
+2 −2 gloo/transport/tcp/tls/pair.cc
+1 −1 tools/amd_build/pyHIPIFY/cuda_to_hip_mappings.py

0 comments on commit df6acf2

Please sign in to comment.