HPC options Cambridge
https://docs.hpc.cam.ac.uk/hpc/user-guide/policies.html#service-level-3-free-usage
Documentation
https://docs.hpc.cam.ac.uk/hpc/
university-wide CSD3 computing cluster (https://docs.hpc.cam.ac.uk/hpc/), which has 90 nodes each with 4x A100 80GB GPUs
Sign up here
https://www.hpc.cam.ac.uk/rcs-application
students can get 3000 GPU hours per quarter for free https://docs.hpc.cam.ac.uk/hpc/user-guide/policies.html#service-level-3-free-usage
There is a queueing/job submission system using SLURM
https://www.cst.cam.ac.uk/local/sys/resources/gpu
We provide shared servers to allow researchers and ACS / Part III students to quickly get started developing and testing code that needs GPUs (and/or a lot of CPU resource). The shared GPU development servers have a single GPU and are intended for testing and development work when you will only intermittently need access to a GPU, running your experimental code for up to a few minutes at most, to keep the impact on other users to a minimum. Once your code is working and needs to be run for more than a few minutes at a time, you should move to another system (preferably the High Performance Computing Service); see below.
PhD students, researchers and faculty can use the shared server dev-gpu-1.cl.cam.ac.uk which has a single A100 GPU (or the CPU development server, dev-cpu-1.cl.cam.ac.uk, when you don't need a GPU).
ACS (MPhil / Part III) students can use the shared server dev-gpu-acs.cl.cam.ac.uk which has a single P100 GPU (or the CPU development server, dev-cpu-acs.cl.cam.ac.uk, when you don't need a GPU).
Unfortunately this service is not available for Part II projects.
If you have not used SSH to connect to a departmental server before, see our SSH documentation. You cannot log in using just a password; you will probably need to set up a VPN and Kerberos.
https://www.hpc.cam.ac.uk/rcs-application
I know that the department also has some resources available internally, but I know very little about what is available here, somewhere to start:
https://www.cst.cam.ac.uk/local/sys/resources/gpu