Usage Rates
Compute
Compute resource rate: $0.0123 per SU
Resources:
- standard compute nodes have 48 cores per node; there are 74 nodes in total
- big memory nodes have 1.5 TiB of memory (RAM) per node; there are 2 nodes in total
- GPU nodes have 4 GPU devices (cards) per node; there are 12 nodes in total
| Picotte Compute Rates | ||
|---|---|---|
| Resource type | Slurm partition | SU per unit resource |
| Std. compute | def | 1 per core-hour |
| Big memory | bm | 68 per TiB-hour |
| GPU | gpu | 43 per GPU device-hour |
Example: Using all 4 GPU devices on a GPU node for 1 hour consumes 172 SU, for a total charge of $0.0123 * 172 = $2.12
NOTE: all resource usage above is computed based on resources reserved for the actual, not requested, lifetime of a job. Examples:
- A job requests 4 GPU devices for 1 hour, but runs only on one GPU device for 1 hour. While the actual usage is 1 GPU-hour, the resources allocated to the job are 4 GPU-hours. The billable amount is 4 GPU-hours = 172 SU. This is because those resources are made unavailable to others.
- A job requests 1 GPU device for 4 hours, but completes in 1 hour. The actual usage is 1 GPU-hour. The charge is for 1 GPU-hour = 43 SU.
Persistent Storage
Storage rate: ~~1.48 SU per TiB-hour~~ 1081 SU per TiB-month
To compare to Proteus (see above), this is equivalent to ~~$3.06 per TiB-week~~ $13.30 per TiB-month ~= $3.32 per TiB-week.
Example: storing 5 TiB of data for 1 month → $0.0123 * 1081 * 5 = $66.48
Usage Reports
To view your projects' usage reports, please see: Migrating to Picotte#Usage Reports and Billing Amounts