Comments (3)
oh my carelessness!
from nccl.
The message you get ("invalid device function") means that one of the GPU cannot execute the compiled code.
Indeed, the GTX760 has a compute capability of 3.0 (https://developer.nvidia.com/cuda-gpus), and the default NVCC_GENCODE in the Makefile only compiles for 3.5 and later.
In the Makefile recompile with :
NVCC_GENCODE="-gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_52,code=sm_52"
then copy build lib to /usr/lib
It work!
from nccl.
You can also set your LD_LIBRARY_PATH instead of installing in /usr/lib :
export LD_LIBRARY_PATH=$PWD/build/lib:$LD_LIBRARY_PATH
./build/test/single/all_reduce_test <size>
from nccl.
Related Issues (20)
- Why do I have to retrieve the uniqueID again before running ncclCommInitRank again? HOT 3
- NCCL communication aborted: unhandled system error HOT 6
- internal error - please report this issue to the NCCL developers HOT 1
- Why does NCCL only use 8 channels on NVLSTree algorithm? HOT 1
- Question about NCCL_NET_MAX_REQUESTS HOT 1
- nccl-tests get stuck and free(): invalid next size (fast) error with 2.19.3 and 2.18.5, but no error with 2.16.5 HOT 21
- [question] libibverbs: Warning: couldn't load driver 'libipathverbs-rdmav25.so': libipathverbs-rdmav25.so: cannot open shared object file: No such file or directory HOT 2
- Is limiting maxCta equivalent to setting maxChannels? HOT 1
- NCCL 2.19.3
- Why does redecescatter Ring algo's result is better than NVLS's result in one H100 server? HOT 4
- Two A100 nodes cannot reach ideal all-reduce performance
- How to calculate the total latency for different collectives? HOT 1
- Getting random segmentation fault HOT 13
- torch.distributed.DistBackendError: NCCL error on 2.18.5 version when using lightning with distributed strategy HOT 2
- transport/net_ib.cc:1295 NCCL WARN NET/IB : Got completion from peer with error 4, opcode 0, len 0, vendor err 81 (Send) localGid ::ffff: remoteGid ::ffff: HOT 5
- NCCL with L3 BGP unnumbered, lo always uses ip 127.0.0.1 HOT 2
- The total nvlink bandwidth obtained by NCCL topo is not the actual total bandwidth HOT 2
- [Bug] NCCL all_reduce failed with A800 when NCCL_ALGO uses Ring HOT 16
- Why none cross NIC has better performance ?
- problem about ncclKerns which has been replaced
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from nccl.