Search results

[Issue]: Compilation failure with CUDA 11.8

Status: Open.

#1972 In NVIDIA/nccl;

· chhwang opened on Jan 6, 2026

[Question]: NCCL memory registration

question

Status: Open.

#1970 In NVIDIA/nccl;

· alokprasad opened on Jan 3, 2026

[Issue]: [nccl4py] Random hang in point-to-point communication with ring topology (16 ranks)

Status: Open.

#1967 In NVIDIA/nccl;

· wujiahao15 opened on Dec 30, 2025

[Question]: Example Usage and Integration Questions about NCCL DSL

question

Status: Open.

#1965 In NVIDIA/nccl;

· wangfakang opened on Dec 26, 2025

[Question]: When will the AllGather PAT algorithm with full topology support be further updated?

question

Status: Open.

#1964 In NVIDIA/nccl;

· predawn-love opened on Dec 25, 2025

[Question]: Is cuMemAlloc without memset safe?

question

Status: Open.

#1962 In NVIDIA/nccl;

· visualxu opened on Dec 24, 2025

[Issue]: Start timestamp is 0.0 for certain reported events when using NCCL example profiler

Status: Open.

#1961 In NVIDIA/nccl;

· LeviMee opened on Dec 24, 2025

[Issue]: New file in 2.29.2 release: git_version.h

Status: Open.

#1960 In NVIDIA/nccl;

· lahwaacz opened on Dec 24, 2025

[Issue]: GIN gdaki mode hardcode 4K MTU in doca modify qp

Status: Open.

#1957 In NVIDIA/nccl;

· cxz66666 opened on Dec 23, 2025

[Issue]: NCCL error: /pytorch/torch/csrc/distributed/c10d/NCCLUtils.hpp:268, NCCL version 2.22.3 ncclSystemError: System call (e.g. socket, malloc) or external library call failed or device error. Last error: socketStartConnect: Connect to IP<51423> failed

Status: Open.

#1956 In NVIDIA/nccl;

· guotong1988 opened on Dec 23, 2025

[Question]: The performance in gin alltoall is lower than that in normal alltoall.

question

Status: Open.

#1955 In NVIDIA/nccl;

· myh1222 opened on Dec 23, 2025

[Question]: Why do sendbuffer and revcbuffer exist simultaneously?

question

Status: Open.

#1954 In NVIDIA/nccl;

· jiangaojie opened on Dec 22, 2025