Search results

Suggestion to adjust license wording for consistency with standard BSD-style phrasing

Status: Open.

#385 In NVIDIA/nccl-tests;

· wjjahah opened on May 29, 2026

Why are different buffers used for each iteration when -n > 1 in nccl-tests?

Status: Open.

#384 In NVIDIA/nccl-tests;

· lidxhaha opened on May 29, 2026

alltoall: NCCL_WIN_COLL_SYMMETRIC incorrectly triggers symk GIN init, causing AlltoAll DevComm GIN connection to be silently skipped

Status: Open.

#381 In NVIDIA/nccl-tests;

· YeSho-cpp opened on May 13, 2026

NCCL all-reduce bandwidth drops in VMs when using more than 2 passthrough ConnectX-7 NICs

Status: Open.

#380 In NVIDIA/nccl-tests;

· mjf-89 opened on Apr 30, 2026

NCCL_IB_DISABLE=1 not work

Status: Open.

#373 In NVIDIA/nccl-tests;

· haolujun opened on Feb 25, 2026

h100 only has 70GB on bus

Status: Open.

#371 In NVIDIA/nccl-tests;

· poptree opened on Feb 7, 2026

running multi-node with Exclusive Process

Status: Open.

#369 In NVIDIA/nccl-tests;

· angainor opened on Jan 21, 2026

Is it applicable to the L20 card?

Status: Open.

#368 In NVIDIA/nccl-tests;

· jiaozenghui opened on Jan 20, 2026

H20-3e 400Gx8 Roce Actual performance experimental data

Status: Open.

#367 In NVIDIA/nccl-tests;

· Rankole opened on Jan 10, 2026

ALL_Reduce_Perf with NVLS in one node(8GPUs)

Status: Open.

#366 In NVIDIA/nccl-tests;

· networkResearcher opened on Jan 9, 2026

Segfault while running benchmarks

Status: Open.

#363 In NVIDIA/nccl-tests;

· access2rohit opened on Dec 12, 2025

Low NCCL Performance in ubuntu kernel 6.8.x

Status: Open.

#362 In NVIDIA/nccl-tests;

· fazlan1990 opened on Dec 10, 2025