SBG Nodes¶
We have 3 Lenovo ThinkSystem SR670 nodes for GPU jobs plus additional nodes purchased due to contributions from researchers. Two nodes contain 4 x Nvidia Tesla V100 cards, and one contains 4 x Nvidia Ampere A100 cards.
| SBG2-3 | Lenovo ThinkSystem SR670 |
|---|---|
| Processor | 2 x 16 Core Intel Xeon Gold 6142 (Skylake) |
| Cores/Node | 32 |
| RAM | 384GB |
| Accessible RAM | ~365GB |
| TMP Size | 1.2TB |
| Interconnect | 25Gb Ethernet |
| GPU | 4 x NVIDIA Tesla V100 |
| GPU architecture | Volta |
| Form Factor | PCIe |
| Tensor Cores | 640 |
| CUDA Cores | 5,120 |
| GPU Memory | 16GiB per GPU (32GiB in sbg3) |
| CUDA Compute | 7.0 (CUDA version 9 or greater required) |
| SBG5 | Lenovo ThinkSystem SR670 |
|---|---|
| Processor | 2 x 24 Core Intel Xeon Platinum 8268 |
| Cores/Node | 48 |
| RAM | 384GB |
| Accessible RAM | ~365GB |
| TMP Size | 1.5TB |
| Interconnect | 25Gb Ethernet |
| GPU | 4 x NVIDIA Tesla A100 |
| GPU architecture | Ampere |
| Form Factor | PCIe |
| Tensor Cores | 432 3rd Generation |
| CUDA Cores | 6,912 |
| GPU Memory | 40GiB per GPU |
| CUDA Compute | 8.0 (CUDA version 11 or greater required) |
Accessing the GPU nodes¶
Access is permitted to QMUL researchers upon request. Note that access to GPU nodes is not permitted for Undergraduate and MSc students. Please raise a support ticket by emailing its-research-support@qmul.ac.uk with a brief overview of intended use, an example of a typical job submission script, and links to any software repositories, so that we can verify that the jobs will use the GPUs correctly. Please see the using GPUs section for advice on submitting GPU jobs.