How to request an interactive GPU on ARC
Jump to navigation
Jump to search
Currently, there are two partitions on ARC that have usable CUDA GPUs, gpu-v100, and bigmem.
The arc.hardware tool can be used to see that.
$ arc.hardware
Node specifications per partition:
================================================================================
Partition | Node CPUs Memory GPUs Node list
| count /node (MB) /node
--------------------------------------------------------------------------------
bigmem | 2 80 3000000 fm[1-2]
| 1 40 4127000 a100:4 mmg1
| 1 40 8256000 a100:2 mmg2
.........
gpu-v100 | 13 40 753000 v100:2 fg[1-13]
.........
1 GPU and 4 CPUs on the gpu-v100 partition for 1 hour (16 GB of RAM):
salloc -N1 -n1 -c4 --mem=16GB --gres=gpu:1 -p gpu-v100 -t 1:00:00
1 GPU and 4 CPUs on the bigmem partition for 1 hour (256 GB of RAM):
$ salloc -N1 -n1 -c4 --mem=256gb --gres=gpu:1 -p bigmem -t 1:00:00
Use the nvidia-smi command to check the GPU:
$ nvidia-smi
Fri Jun 3 11:35:14 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02 Driver Version: 470.57.02 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA A100-PCI... Off | 00000000:17:00.0 Off | 0 |
| N/A 39C P0 42W / 250W | 0MiB / 40536MiB | 32% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+