Edit the GPU documentation with the new approach for accessing the GPUs

[kylemonahan]

Summary: Let's update our HPC docs with the GPU changes that @DelilahYM sent out to HPC users, below. Just adding this issue to remind us to do this, and so we can discuss these proposed changes.

Here are a few places we could change:

https://rtguides.it.tufts.edu/hpc/slurm/batchjob.html

• On this link, "copy" is spelled incorrectly as "cope" and we should call out the constraints here, as this is the first time users see the GPU in the docs.

https://rtguides.it.tufts.edu/hpc/slurm/interactive.html

• On this link, we should call out the constraints as above, and we should consider linking to the GPU resources in case they missed that: https://rtguides.it.tufts.edu/hpc/compute/gpu.html

GPU changes that Delilah sent out, for reference:

Here are the important changes GPU users need to be aware and make adjustment in SLURM job scripts or commands:

Starting May 1st, 2025:

    “gpu” partition users need to request A100 gpus with constraints specified.
        --gres=gpu:a100:1 --constraint=”a100-80G” (for one A100-80GB GPU)
        --gres=gpu:a100:1 --constraint=”a100-40G” (for one A100-40GB GPU)
        --gres=gpu:a100:1 (for one of any A100s)