NVIDIA nccl-tests issues

NVIDIA / nccl-tests

NCCL Tests

BSD 3-Clause "New" or "Revised" License

775 stars 226 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

undefined reference to ncclRedOpDestroy

#195 freshduer opened 6 months ago
2
all_reduce_perf between NVLINK connected H100 PCIe GPUs lower than A100 SXM4 GPUs

#194 chinthysl opened 6 months ago
0
NCCL Test hang when the number of nodes goes beyond 18, and CPU usage is very high

#193 chgdragon2023 opened 6 months ago
2
NCCL Test Does not work with GID 3 or GID 1, but it works fine for GID 0

#192 chgdragon2023 opened 7 months ago
0
nccl-tests result is only a half of ib_write_bw

#191 HeGaoYuan opened 7 months ago
0
hypercube out-of-bound errors with single-proc + `gpus-per-thread=4`, not with multi-proc + `gpus-per-thread=1`

#190 robogast opened 7 months ago
1
clarify that the measurement is unidirectional

#189 stas00 opened 7 months ago
11
misc/socket.cc:441 NCCL WARN socketFinalizeAccept: wrong type 4 != 3

#188 MiyazonoKaori closed 7 months ago
6
NCCL alltoall_perf hangs via PXN

#187 gavin1332 closed 7 months ago
1
how can i run nccl-test use max bandwidth

#186 liuxingbo12138 opened 8 months ago
0
misc/ibvwrap.cc:187 NCCL WARN Call to ibv_modify_qp failed with error Network is unreachable

#185 chgdragon2023 opened 8 months ago
3
Nsight Profiling: one ncclAllReduce takes too long

#184 yanminjia opened 8 months ago
0
Test NCCL failure common.cu:954 'unhandled cuda error" when test on >2 GPUs

#183 caopulan closed 8 months ago
4
Although it is an InfiniBand environment, it seems that the average Bandwidth is not as good as expected.

#182 gim4moon opened 8 months ago
4
AlltoAllGetBw is incorrect when used with multiple nodes

#181 sukoncon opened 8 months ago
1
./build/all_reduce_perf between nodes failed

#180 sleepwalker2017 opened 8 months ago
1
nccl-test is throwing timeout error on two nodes

#179 manomugdha opened 8 months ago
26
A100 - All reduce performance

#178 arul-lm opened 8 months ago
1
bus error

#177 bltcn closed 8 months ago
3
what does error in nccl-test output represent?

#176 blackgold opened 9 months ago
3
Two A100 nodes cannot reach ideal all-reduce performance

#175 lcw2 opened 9 months ago
4
No explanation on BusBW factor regarding alltoall in docs

#174 lappazos opened 9 months ago
0
Collecting latency data per coll.

#173 nv-udeodhar closed 9 months ago
0
Why need more than one iteration to check data?

#172 FarmerLiuAng closed 9 months ago
4
Issue Running NCCL Tests on Gentoo with Varying GPU Availability: CUDA failure common.cu:892 'invalid device ordinal'

#171 SweeneyJun closed 9 months ago
3
unhandled cuda error during test

#170 mlinmg closed 9 months ago
1
if the bandwidth results of the Nccl test are related to the number of nodes?

#169 PrometheusComing opened 9 months ago
2
Test in dockers of multi-node

#168 jiangxiaobin96 opened 10 months ago
0
all_reduce_perf(--op='sum') get wrong results when size is over specific value

#167 metaVariable closed 7 months ago
9
Test NCCL failure common.cu:958 'internal error - please report this issue to the NCCL developers / '

#166 kylematoba closed 8 months ago
10
Test CUDA failure common.cu:892 'invalid device ordinal'

#165 marabgol closed 11 months ago
11
Calculating "net_bw" in addition to "bus_bw"

#164 yehuday opened 12 months ago
0
when i am running this command : mpirun -np 1 ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 2. I found this

#163 james2v opened 1 year ago
2
Nccl test fails on 8 x V100- misc/socket.cc:483 NCCL WARN socketStartConnect: Connect to xxx failed : Software caused connection abort

#162 hacker-jerry closed 1 year ago
9
When I am running on multiple nodes, I can get the corresponding results when running on 3 nodes, and an exception will occur when more than 3 nodes are executed.

#161 songqimao opened 1 year ago
3
what do algobw actually mean when I run test with more than one nodes?speed between nodes or speed between gpus.

#160 wenjunlong closed 1 year ago
3
The difference between algbw and busbw

#159 allinsds opened 1 year ago
0
Testing git\n

#158 BhaviniMishra closed 1 year ago
0
New AlltoAllV (Imbalanced AlltoAll) benchmark.

#157 babusid opened 1 year ago
1
Two A800 nodes cannot reach ideal all-reduce performance

#156 joydchh opened 1 year ago
18
Debugging with cuda-gdb causes problems

#155 minihu-crypto opened 1 year ago
0
Bandwidth result not equal to ib_write_bw result

#154 Jiaao-Bai closed 1 year ago
3
`busbw` does not reflect the speed of hardware bottleneck in H800

#153 zhangmenghao opened 1 year ago
7
Origin of Poor Internode NCCL Performance

#152 vitduck closed 1 year ago
11
Don't call MPI_Comm_split if NCCL_TESTS_SPLIT_MASK is not set

#151 tstruk opened 1 year ago
2
all_reduce_perf fails on 2 nodes

#150 scvance closed 1 year ago
2
Expected bandwidth results? 8x A100 GPUs over NVLink

#149 acgandhi opened 1 year ago
10
test error: stuck when run test example

#148 zhengmq2010 opened 1 year ago
4
src/Makefile: remove unused variables

#147 yangxingwu opened 1 year ago
0
makefile: remove extra space

#146 yangxingwu closed 1 year ago
0

Previous Next