How do I connect to HPC?

You can connect to HPC by SSH client with the following information. For HPC Coral: SSH (login-0-0): IP: 10.4.1.115 Port: 22 Web (login-0-1): If you need GUI interface, please go to: https://10.4.1.116:4443/nxwebplayer For HPC Jade: IP: 10.4.1.100 Port: 22

Read More

What is GPU Computing?

GPU accelerated computing offers unprecedented application performance by off-loading compute-intensive portions to the GPU. Each GPU consists of thousands of smaller and more efficient cores for handling multiple tasks. Applications are computed much faster in a parallel computing environment with GPU accelerating. By June 2016, our HPC cluster – Jade supports GPU accelerated computing with 2 NVIDIA K80 […]

Read More

What is CUDA-Aware MPI?

Message Passing Interface (MPI) is a standard API for communicating data between distributed processes. It is commonly used in HPC to build applications that can scale to multi-nodes. CUDA-Aware MPI is communicating data between difference GPU processors without staging through system memory. For more detailed information, please refer to the following webpages: An Introduction to CUDA-Aware […]

Read More

How do I run a CUDA program in Pearl?

There are two computing nodes “compute-2-2” and “compute-2-3” installed with one NVIDIA K40m GPU card respectively. You can submit your job to “gpu.q” for using GPU acceleration. Here is an example of submission script for running in a single node: #!/bin/sh #$ -S /bin/bash #$ -cwd #$ -V #$ -N p1 #$ -j y #$ -A username #$ -q […]

Read More

How do I know my application supporting GPU acceleration or not?

Most of the programs, packages and software require re-compilation for supporting GPU acceleration. You can check the list from NVIDIA to see whether your application is supported or not. GPU-ACCELERATED APPLICATIONS: https://www.nvidia.com/en-us/data-center/gpu-accelerated-applications/catalog/ If you need assistance, please contact unix_admin@um.edu.mo

Read More