10000 Tags · Hamidreza-Ramezani/nccl · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Tags: Hamidreza-Ramezani/nccl

Tags

v2.7.8-1

Toggle v2.7.8-1's commit message 
2.7.8-1

Fix collective mismatch error when using ncclSend/ncclRecv

v2.7.6-1

Toggle v2.7.6-1's commit message 
2.7.6-1

Fix crash when NVswitch is not visible inside a VM.

v2.7.5-1

Toggle v2.7.5-1's commit message 
2.7.5-1

Minor fixes for A100 platforms.
Add a WARN for invalid GroupEnd call.

v2.7.3-1

Toggle v2.7.3-1's commit message 
2.7.3-1

Add support for A100 GPU and related platforms.
Add support for CUDA 11.
Add support for send/receive operations (beta).

v2.6.4-1

Toggle v2.6.4-1's commit message 
2.6.4-1

Add support for network collectives.
Add support for XML topology dump/injection.
Add text values for GDR and P2P Levels, including "NVL".
Add speed detection for PCI, Infiniband and Ethernet cards.
Add CPU detection for ARM and AMD CPUs.
Add support for adaptive routing on Infiniband.
Change NET plugin API to v3 : merge PCI path and GPU pointer
  capability into a single structure and add other properties.

v2.5.7-1

Toggle v2.5.7-1's commit message 
2.5.7-1

v2.5.6-2

Toggle v2.5.6-2's commit message 
2.5.6-2

Fix PPC64 Debian packaging

v2.5.6-1

Toggle v2.5.6-1's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Verified
Learn about vigilant mode 
2.5.6-1 (NVIDIA#255)

Add LL128 Protocol.

Rewrite the topology detection and tree/ring creation (NVIDIA#179). Improve
tree performance by sending/receiving from different GPUs. Add
model-based tuning to switch between the different algorithms and
protocols.

Rework P2P/SHM detection in containers (NVIDIA#155, NVIDIA#248).

Detect duplicated devices and return an error (NVIDIA#231).

Add tuning for GCP

v2.4.8-1

Toggle v2.4.8-1's commit message 
2.4.8-1

Fix NVIDIA#209: improve socket transport performance
  Split transfers over multiple sockets
  Launch multiple threads to drive sockets
  Detect AWS NICs and set nsockets/nthreads accordingly

v2.4.7-1

Toggle v2.4.7-1's commit message 
NCCL 2.4.7-1

    Performance tweaks for PowerPC builds only;
      Set default NCCL_MIN_NRINGS to 4
      Disable PCI-E NUMA distance detection
0