Zuricho / ParallelFold

Modified version of Alphafold to divide CPU part (MSA and template searching) and GPU part. This can accelerate Alphafold when predicting multiple structures
https://parafold.sjtu.edu.cn
147 stars 45 forks source link

这个是将1,2,3个序列,并行到0,1,2显卡完成吗? #8

Closed ChrisLou-bioinfo closed 2 years ago

ChrisLou-bioinfo commented 2 years ago

能否实现两个显卡为一个流程服务?看Alphafold issues讨论过,源码是不支持多显卡提升性能的。不知您的这个ParallelFold能否实现呢?

Zuricho commented 2 years ago

我们也不可以