NYU HPC Nvidia Clusters
- ssh -Y kc3585@greene.hpc.nyu.edu
- cd cuda
- source setup
- On every connection to cluster run:
source setup
- Download XQuartz on mac (or nothing on Vital)
- ssh -Y greene-solo (kc3585@greene.hpc.nyu.edu)
- (Don't do this) Start interactive session with X11 forwarding enabled
srun --x11 --pty /bin/bash
- Vector Addition
- start x11 server (wait for resources)
- compile program (this creates an executable)
nvcc -o VA-GPU-11 VA-GPU-11.cu
- run executable by scheduling the job
srun --x11 --nodes=1 --gres=gpu:1 --pty VA-GPU-11
- Dot Product
- compile program
nvcc -o Dot-GM Dot-GM.cu
- run executable by scheduling the job
srun --x11 --nodes=1 --gres=gpu:1 --pty Dot-GM
- compile program
- make audio
- compile the program
- make test-audio
- test the program
- make go-cpu
- run on the CPU
- make go-cuda
- run on GPU