As a powerful research infra, we recently got a NVIDIA HGX-A100 GPU machine.
The GPU server has eight A100 GPU and each of GPU memory is 80Gbyte (total 8x80 = 640GB!).
It is valued approximately $150,000, and supported by RIST-POSCO OpenLab.
We plan to use this strong machine for large-scale video analysis and enhancement researches!
Now, our lab has our own big cluster system that consists of a centeralized storage server with 6 shared GPU client servers. The GPU servers hold the GPUs of RTX 3090 (24Gb), Titan RTX (24Gb), A6000 (48Gb), and A100 (80Gb).
Our cluster enables to share the same environment regardless of which server students log into.