2374a563cc86f95474f1ec98852c93842e93ab36
examples/alphafold2.md
... | ... | @@ -144,4 +144,6 @@ It is possible to run alphafold through a web portal. See |
144 | 144 | https://colab.research.google.com/github/deepmind/alphafold/blob/main/notebooks/AlphaFold.ipynb . |
145 | 145 | |
146 | 146 | ### Known issues |
147 | -Known issues: This version may not run on some newer GPUs that require CUDA 11.1 or later. An update is coming that should correct this. |
|
... | ... | \ No newline at end of file |
0 | +- Some newer GPUs that require CUDA 11.1 or later may not be able to run this version of Alphafold. An update is planned that should correct this. |
|
1 | +- Unified memory across GPUs does not appear to work in the current version. |
|
2 | +- The `ptxas` executable is required to be in PATH in some cases, but not all. We can not redistribute this binary since it is part of the NVIDIA SDK. It must be installed separetely and added to the environment PATH variable, typically in `/usr/local/cuda/bin`. Version 11.0.3 works well in our hands, but other versions should work. [You can download the SDK here : https://developer.nvidia.com/cuda-toolkit-archive](https://developer.nvidia.com/cuda-toolkit-archive) |