Skip to content
This repository has been archived by the owner on Feb 1, 2020. It is now read-only.

Issues: facebookarchive/TorchMPI

10 Open 4 Closed
10 Open 4 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Non-productive luajit processes spawned
#52 opened Jul 6, 2018 by kazuhiro-itakura
Configuring incomplete, errors occurred!
#51 opened May 29, 2018 by pankajkumar
Interface Gloo collectives
#11 opened Mar 6, 2017 by nicolasvasilache
Implement a simple mnist parameterserver + data parallel + model parallel example
#10 opened Mar 6, 2017 by nicolasvasilache
Implement broadcast p2p cross nodes
#8 opened Mar 6, 2017 by nicolasvasilache
Investigate perf of async offload to threadpool and threadpool perf in general
#7 opened Mar 5, 2017 by nicolasvasilache
reduce CPU involvement in p2p GPU collectives
#6 opened Mar 5, 2017 by nicolasvasilache
Write a proper wrapper assigning ranks to nodes for DGX-1 based systems
#4 opened Mar 5, 2017 by nicolasvasilache
Investigate NCCL communicator deadlocks when more than 1 rank per GPU
#3 opened Mar 5, 2017 by nicolasvasilache
Investigate NCCL communicator deadlocks under threadpool
#2 opened Mar 5, 2017 by nicolasvasilache
ProTip! Follow long discussions with comments:>50.