yezhengmao/nccl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

0 Releases 63 Tags

RSS Feed
  • v2.13.4-1 19ab67d172

    2.13.4-1

    Ghost released this 2022-07-11 23:10:34 +08:00 | 61 commits to master since this release

    Optimize CUDA graph launch; avoid launching a CPU callback for
    intra-node operations.
    Simplify kernel common code to improve the latency of send/recv
    operations.
    Strengthen CUDA streams semantics.
    Change NET API to v6, to add dmabuf support.
    Add ncclGetLastError() function.
    Add ncclRemoteError code and use it for remote network errors.
    Support the use of a different NCCL_NET parameter per communicator.
    Add support for SHM and P2P transfers using cudaMemcpy.

    Downloads