All-in-One Deep learning framework, customer oriented optimization for your own purpose

HPCMATE provides optimized Deep Learning Framework. Customer can select whatever combination of software components.

Best for

• HPCMATE DL framework - Docker, Slurm, Spark, Jupyterhub, MXNET, Caffe, CNTK, Torch, Theano, CUDA, NCCL, DL4J, OpenACC, OpenCL, ROCm, etc
• Fully automated build, deployment and validation
• Support customized Deep learning Framework - Docker container, Spark Streaming, Conda Environments
• Optimized system level tunning for the best performance of Deep learning Framework