Resources
GPU
We have 4 GPU servers accessible to the entire CL department: Habra, Glados, Tars and Rattle. Glados access is restricted to only one group of users. All of them are accessible via ssh
ssh shortname@hostname.cl.uzh.ch
| Machine | Running | #GPUs | #CPUs | RAM | Local storage |
|---|---|---|---|---|---|
| Habra | 🟢 | 8x RTX 4090 | 2x AMD EPYC 9654 96-Core | 770G DDR4 | |
| Glados | 🟢 | -- | -- | -- | |
| Tars | 🔴 | 8x RTX 3090 | 2x AMD EPYC 7742 64-Core | 512G DD4 | |
| Rattle | 🔴 | 8x TitanX | 2x Intel(R) Xeon (R) CPU E5-2650 | 128G DDR4 |
- All servers have NFS storage mounted (
/home,/projects/and/projects-data/ /scratch/shortnameis local to the machine. That is, anything user does here is not accessible outside the machine. So, the user has to copy the data either via scp. See s3it docs for a simple explanation.
CPU
| Machine | Running |
|---|---|
| Hex | 🔴 |
| Bender | 🔴 |
| Flexo | 🔴 |
Why are the systems down?
All the GPU systems have been setup and running locally. Due to network issues we are unable to reach the servers. Our IT team is in touch with ZI to resolve the issues.