Azure/msccl-executor-nccl issues and pull requests

#55 - Fix algorithm loading order

Pull Request - State: closed - Opened by yzygitzh about 2 months ago

#54 - Add Azure CI integration test

Pull Request - State: closed - Opened by seagater about 2 months ago

#53 - Fix NPKit rank reference

Pull Request - State: closed - Opened by yzygitzh about 2 months ago - 1 comment

#52 - Fix NPKit shutdown timing

Pull Request - State: closed - Opened by yzygitzh about 2 months ago - 1 comment

#51 - Fix README

Pull Request - State: closed - Opened by yzygitzh 2 months ago

#50 - Compile error: -DENABLE_NPKIT_NET_CHECK_LATENCY

Issue - State: open - Opened by ChenYuHo 2 months ago - 4 comments

#49 - Add build-time flag for precision clipping for half

Pull Request - State: closed - Opened by seagater 2 months ago - 3 comments

#48 - Add build-time flag to control number of instructions

Pull Request - State: closed - Opened by seagater 2 months ago - 6 comments

#47 - Sync msccl-executor-nccl to nccl-2.23.4-1

Pull Request - State: closed - Opened by seagater 3 months ago - 3 comments

#46 - Fix size matching in MSCCL

Pull Request - State: closed - Opened by yzygitzh 5 months ago

#45 - Fix NPKit context index calculation

Pull Request - State: closed - Opened by yzygitzh 9 months ago

#44 - Add resilient ib plugin

Pull Request - State: open - Opened by Andyli1007 9 months ago

#43 - Replace cudaFree with ncclCudaFree

Pull Request - State: closed - Opened by yzygitzh 10 months ago

#42 - How to run custom algos in single process?

Issue - State: closed - Opened by TonyWu199 11 months ago - 1 comment

#41 - Msccl v2.19 integrate (#40)

Pull Request - State: closed - Opened by Andyli1007 about 1 year ago - 2 comments

#40 - Msccl v2.19 integrate

Pull Request - State: closed - Opened by Andyli1007 about 1 year ago

#39 - Msccl v2.19

Pull Request - State: closed - Opened by Andyli1007 about 1 year ago

#38 - use ncclFp8E4M3 can't get correct result

Issue - State: closed - Opened by kanghui0204 about 1 year ago - 2 comments

#37 - Enable NIC Failure Recovery

Pull Request - State: open - Opened by Andyli1007 about 1 year ago

#36 - Support dynamic channels in MSCCL

Pull Request - State: closed - Opened by yzygitzh about 1 year ago

#35 - fix share algo path issue

Pull Request - State: closed - Opened by Andyli1007 about 1 year ago

#34 - fix the typo of npkit_events id

Pull Request - State: closed - Opened by Andyli1007 about 1 year ago

#33 - remove platform detection logic

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#32 - Remove buggy and redundant logs

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#31 - fix the correctless issue when using cuda graph in sm90 env

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#30 - fix readme typo

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#29 - Fix readme

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#28 - add ndv5 algo files and azure vm detection logic

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#27 - Enable npkit for msccl

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#26 - enable npkit for msccl

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#25 - Fix asm volatile issue

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#24 - Enable CGA for SM90

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#23 - Fix msccl-algorithm folder copy

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#22 - clean up code

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#21 - Msccl v2.18

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#20 - fix the work index low value issue in Cuda graph enabled scenario.

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#19 - Hotfix

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#18 - Fix compiler warnings

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#17 - fix the logic issue of chunks calculation for cpu proxy

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#16 - fix the logic issue of chunks calculation for cpu proxy

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#15 - improve the build capability for the arch lower than 80

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#14 - fix the build capability issue under arch 800

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#13 - enable make install & deb package for msccl

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#13 - enable make install & deb package for msccl

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#12 - enable make install & deb package

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#12 - enable make install & deb package

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#11 - Create msccl-algorithms folder by default

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#11 - Create msccl-algorithms folder by default

Pull Request - State: closed - Opened by yzygitzh over 1 year ago

#6 - Enable msccl capability (#5)

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#6 - Enable msccl capability (#5)

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#5 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#5 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#4 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#4 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#2 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#2 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#1 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

#1 - Enable msccl capability

Pull Request - State: closed - Opened by Andyli1007 over 1 year ago

GitHub / Azure/msccl-executor-nccl issues and pull requests