Comments (1)
sjeaugey
commented on August 15, 2024
While NCCL 1.3 was only relying on a single ring, it has evolved to use multiple rings to use all NVLinks and network cards, so that a rank no longer has a unique index.
More recently, NCCL also started to use trees for small/medium size operations.
So the ring NCCL creates is neither unique nor the best way to communicate for e.g. a halo exchange. Halo exchanges should be handled by point-to-point operations when they are implemented.
Also closing as this is an old request ; feel free to reopen / follow up.
from nccl.
Related Issues (20)
- Is it possible to swap the calling order of `initTransportsRank` and `ncclTunerPluginLoad` HOT 1
- NCCL Logs Communicator Query HOT 1
- work request complete err: status 5 and vendor err 249 HOT 7
- Is there someway to measure gpu i/o usage or allreduce waiting time? HOT 1
- About sync in nvls algorithm
- NCCL Tree allreduce test cannot reach the theoretical bus bandwidth on 2 nodes with 4 nics HOT 7
- how does NCCL support peer-to-peer connections across NUMA nodes without the features of NICs and NVLinks? HOT 2
- How can I test IB bandwidth when NCCL is running?
- Single or double ring HOT 1
- Missing header file HOT 7
- Why does NVLSTree Allreduce perform worse than Ring Allreduce? HOT 1
- Encountering Random Segmentation Fault During NCCL-Tests HOT 14
- Ring broadcast
- inter-node nvls process when ib sharp not supported HOT 1
- NCCL Error on Multi-Node Training with Mixed GPU Setup HOT 2
- why intra-node ring only search one-direction bw?
- NCCL test, Tree is slower than Ring HOT 2
- Use nsight system to profile nccl p2p, I find something confused ...
- Scheduling to Packet Sending Connection from ncclEnqueueCheck to finally sending the packet in misc/socket.cc::socketProgressOpt
- standardize code style
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.