High Performance Computing or HPC is a process of optimizing computing performance in the cloud that can process data more effectively and productively. The following are the components in the HPC Hardware and Software that we provide, including:

Hardware

1. High performance GPU server, i.e. server with high core count CPU, RAM >= 128 GB per-server, multiple GPUs per-server.
2. High bandwidth and low latency interconnect between the GPU server. For example: HPE Slingshot interconnect, Fujitsu Tofu interconnect, etc.

Software

1. Operating System with support for the high bandwidth-low latency interconnect; both in the device driver and application library.
2. Parallel processing library to be used in the HPC server cluster such as MPI (Message Passing Interface).
3. GPU-accelerated library support for math calculation and AI, for example: Nvidia Cuda, AMD RoCM, etc.
4. Math library optimized for the server platform, such as: AMD AoCL, Intel OneAPI math kernel library, etc.

We provide HPC products and services in collaboration with well-known vendors such as HPE.