Comments (3)
Apologies, obviously an error on our end, we were using a flag -DCUDA_ARCH=12.0.0 erroneously. I wonder why it ever worked? Works after this is taken out.
Perhaps a follow-up question - is there an advantage of making separate compilations of relion for nodes having different gpus? We have P100s, V100s, and A100s, all supporting different compute capabilities. We can go for a single install with the lowest common denominator (P100 = sm60), but will we have better performance on A100s if we compile it there with sm80?
If yes, would multi-node jobs still work between nodes that have relion compiled with different cuda compute capabilities?
from relion.
will we have better performance on A100s if we compile it there with sm80?
Theoretically yes but when I tested this before, the difference was less than 5 %. Of course this depends on the card and the task (is the GPU really limiting?). You should test on your hardware and dataset.
would multi-node jobs still work between nodes that have relion compiled with different cuda compute capabilities?
Yes.
from relion.
Ok, given that we have many people running very different projects and so can't do focused optimization, 5 % potential gain we'll probably skip.
Many thanks for your input!
from relion.
Related Issues (20)
- failed to create cufft plan HOT 3
- No option to correct CTF during ReconstructTomograms in Relion5 GUI HOT 8
- Relion5 build : Could NOT find Torch home directory for Conda env HOT 1
- Issue with CTF estimation, RELION tries to convert particles even though they are in the right format HOT 1
- Crash with relion and GPU HOT 2
- Multibody refinement initial offset search parameters are multiplied by an extra factor of pixel size
- stack numbering not updated in relion_image_handler
- Relion5.0 tomo HOT 7
- CTFFIND5 support HOT 6
- Gain ref different size than tifs HOT 3
- Cannot SIZE object 50 because it does not exist
- Error running blush refinement HOT 2
- DynaMight missing python modul HOT 10
- How to get blush model wieght installed and sourced in a location accessible to all? HOT 1
- relion v4.0 :how to change size of relion's window HOT 1
- merging dataset in relion HOT 1
- Relion-5.0-beta-2-commit DynaMight visualization: TypeError: can't convert np.ndarray of type numpy.object
- Relion4.0.1_cu11.6_Refine_No orientation was found as better than any other HOT 5
- Dynamight is not working
- 3D classification fails on iteration 2, but only when multiple classes are requested HOT 10
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 relion.