The ERISXdl (ERIS eXtreme Deep Learning) GPU platform is opening for early adopters. The system currently includes 40 NVIDIA Tesla V100 with an aggregate of more than 200,000 CUDA cores. Several deep learning frameworks are available:
- Pytorch
- TensorFlow
- mxnet
- theano
The platform supports Docker for containerized environments with access to the NVIDIA AI Catalog as well as Slurm scheduler for resource management.
The platform will be open for early adopters for free until it's ready for production. The cost for ERISXdl usage on production will be around $ 0.008 per minute GPU.
Please take a look at these instructions to get started. If you have any questions or suggestions please email hpcsupport@partners.org.