Open gancho-ivanov opened 2 years ago
I am also interested in this! Linux ARM64 is used more and more for HPC. +1 for better support!
In my experience Linux ARM64 is not supported at the moment.
Creating conda environment on my Ubuntu 20.04.5 ARM64 fails with:
$ conda env create -f RoseTTAFold-linux.yml
Collecting package metadata (repodata.json): done
Solving environment: failed
ResolvePackageNotFound:
- gnutls==3.6.15=he1e5248_0
- torchaudio==0.9.0=py38
- intel-openmp==2021.2.0=h06a4308_610
- libgfortran-ng==7.5.0=h14aa051_19
- libunistring==0.9.10=h27cfd23_0
- mkl-service==2.3.0=py38h27cfd23_1
- pytorch-cluster==1.5.9=py38_torch_1.9.0_cu111
- xz==5.2.5=h7b6447c_0
- libpng==1.6.37=hbc83047_0
- ffmpeg==4.3=hf484d3e_0
- libidn2==2.3.1=h27cfd23_0
- bzip2==1.0.8=h7b6447c_0
- pytorch-spline-conv==1.2.1=py38_torch_1.9.0_cu111
- biocore::blast-legacy=2.2.26
- libffi==3.3=he6710b0_2
- libiconv==1.15=h63c8f33_5
- hhsuite
- pytorch-geometric==1.7.2=py38_torch_1.9.0_cu111
- zstd==1.4.9=haebb681_0
- libstdcxx-ng==9.3.0=hd4cf53a_17
- certifi==2021.5.30=py38h06a4308_0
- mkl==2021.2.0=h06a4308_296
- libgomp==9.3.0=h5101ec6_17
- sqlite==3.36.0=hc218d9a_0
- cffi==1.14.5=py38ha65f79e_0
- ncurses==6.2=he6710b0_1
- libtasn1==4.16.0=h27cfd23_0
- torchvision==0.10.0=py38_cu111
- brotlipy==0.7.0=py38h497a2fe_1001
- psipred=4.01
- libuv==1.40.0=h7b6447c_0
- freetype==2.10.4=h5ab3b9f_0
- ninja==1.10.2=hff7bd54_1
- zlib==1.2.11=h7b6447c_3
- pysocks==1.7.1=py38h578d9bd_3
- openh264==2.1.0=hd408876_0
- cudatoolkit==11.1.74=h6bb024c_0
- lame==3.100=h7b6447c_0
- lcms2==2.12=h3be6417_0
- ld_impl_linux-64==2.35.1=h7274673_9
- numpy==1.20.2=py38h2d18471_0
- pytorch==1.9.0=py3.8_cuda11.1_cudnn8.0.5_0
- nettle==3.7.3=hbbd107a_1
- libgcc-ng==9.3.0=h5101ec6_17
- pytorch-sparse==0.6.10=py38_torch_1.9.0_cu111
- ca-certificates==2021.5.25=h06a4308_1
- readline==8.1=h27cfd23_0
- pillow==8.2.0=py38he98fc37_0
- gmp==6.2.1=h2531618_2
- python==3.8.10=h12debd9_8
- pandas==1.2.5=py38h1abd341_0
- libtiff==4.2.0=h85742a9_0
- jpeg==9b=h024ee3a_2
- scikit-learn==0.24.2=py38ha9443f7_0
- numpy-base==1.20.2=py38hfae3a4d_0
- cryptography==3.4.7=py38ha5dfef3_0
- chardet==4.0.0=py38h578d9bd_1
- libgfortran4==7.5.0=h14aa051_19
- libwebp-base==1.2.0=h27cfd23_0
- markupsafe==2.0.1=py38h497a2fe_0
- setuptools==52.0.0=py38h06a4308_0
- mkl_random==1.2.1=py38ha9443f7_2
- pytorch-scatter==2.0.7=py38_torch_1.9.0_cu111
- mkl_fft==1.3.0=py38h42c9631_2
- tk==8.6.10=hbc83047_0
- openssl==1.1.1k=h27cfd23_0
- pip==21.1.3=py38h06a4308_0
- lz4-c==1.9.3=h2531618_0
- blas==1.0=mkl
Hello Rosetta community!
I'd like to ask whether it is safe to run RoseTTAFold on Linux ARM64 machines ? Does anyone have positive experience with it ?
I am talking about cloud friendly environments like AWS Graviton, Oracle/Azure/Google Ampere and Huawei's TaiShan. I am not asking about RasberryPi!
Thank you! Gancho