Comments (14)
Hi same problem here, I have the following card NVIDIA GeForce RTX 2080 Ti , I know 11Mb is not a lot, but maybe there is a workaround?
I have many CPUs', but I am unsure how to make the modeling itself with CPUs, can anybody give accurate instructions if this is possible?
Thanks!!
Fabian
Trying to solve a big protein/RNA complex...
from rosettafold2na.
RuntimeError: CUDA out of memory. Tried to allocate 1.54 GiB (GPU 0; 10.75 GiB total capacity; 6.28 GiB already allocated; 1.44 GiB free; 7.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
from rosettafold2na.
How large is the structure you are trying to predict?
from rosettafold2na.
How large is the structure you are trying to predict?
The complex I want to predict consists of a protein with 872 residues and a DNA with 12bp.
from rosettafold2na.
I had the same problem, and my GPU mem is 8GB. Does it have to run on a multi-GPU server?
from rosettafold2na.
I have encountered this same issue. It was running on a 3090 GPU with 24GB mem.
from rosettafold2na.
Running on a GPU with more memory solved the problem for me when I encountered this error.
from rosettafold2na.
Running on a GPU with more memory solved the problem for me when I encountered this error.
I forced this program to use CPU, which can avoid this error, but it was too slow.
from rosettafold2na.
As others have pointed out, for large complexes the memory requirements might be high, using a higher memory card (or CPU only) may be necessary.
I have a memory-optimized version I will try to push in the next couple of weeks (need to make sure results are the same).
from rosettafold2na.
More of a FYI: I'm getting CUDA out of memory
for a complex of a 505 aa dimer (1010 aa in total) with a 76 bases RNA molecule (x2 = 152 in total) on an NVIDIA V100 with 32GB memory, which is somewhat disappointing as I was hoping to also model considerable larger complexes than that in the future.
from rosettafold2na.
If anyone is encountering this error, I just solved it yesterday by:
export 'PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:256'
and then running the script as usual. I have a 16GB GPU and trying to solve for 1030nts long RNA structure.
Thanks,
Kamil
from rosettafold2na.
met the same problem, hope a new version of multi-gpu in the future
from rosettafold2na.
FYI I tested a similar system to the one described by @mf-rug (2x500aa + 76b RNA) on a 16GB GPU, using @kcygan recommendation of setting the PYTORCH_CUDA_ALLOC_CONF environment. I still got the CUDA out of memory error.
I also tried suggestions from stackoverflow
- adding garbage_collection_threshold:0.6 to the PYTORCH_CUDA_ALLOC_CONF
- setting torch.cuda.empty_cache() at the start of the _run_model() function
Alas none of these helped.
However running predict.py on 8 CPUs with 64GB memory on an HPC completed in ~ 5hrs. Good enough for me in this instance.
from rosettafold2na.
If you're struggling to run RosettaFold2NA locally, feel free to try https://www.tamarind.bio/rosettafold2na. Tamarind is an online platform for bioinformatics tools that offers structural biology workflows, including RoseTTAFold2NA, for free.
from rosettafold2na.
Related Issues (20)
- reprocess_rnac.pl - Out of memory HOT 1
- Please provide pre-processed files for RNA
- Error: (803.7) [makeblastdb] Blast-def-line-set.E.title HOT 3
- Cannot load Graphbolt C++ library HOT 2
- About whether RoseTTAFold2NA can generate multiple models?
- Downloading pre-trained weights
- Regression: run_RF2NA.sh - R:RNA.fa No such file or directory HOT 2
- NVTX functions not installed HOT 2
- A SOLUTION for Package Dependency Issues 📦 HOT 2
- run_RF2NA.sh: Logging issues HOT 2
- make_rna_msa.sh: line 9: [: missing `]' HOT 2
- UnboundLocalError: local variable 'is_ignore' referenced before assignment HOT 3
- dimaio/new_config: Conda environment needs to be activated
- UserWarning: Using torch.cross without specifying the dim arg is deprecated. HOT 1
- Unaccepted characters after Rfam-database-check HOT 4
- Antibody structure prediction HOT 1
- Is the training dataset exclusive and is it available?
- hardware requirements
- RNA Secondary Structure
- How the weight file is generated, and whether the code has been open-sourced
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 rosettafold2na.