System configuration
CIIRCluster comprises several types of nodes: login nodes, GPU nodes with different GPGPU cards or pure compute nodes with 256 or 512 GB of RAM. When specifying resources for your job, you can use slurm flags (see resource allocation) mentioned in last row.
Warning
All heavy computing have to be done via slurm on GPU/Compute nodes.
GPU-node with 4 x 1080 Ti
| *GPU-node with 1080 Ti | ** 4 so far, total of 5 planned ** |
|---|---|
| CPUs | 2x Xeon(R) Gold 6140 CPU; 18 Cores/CPU; 2.3GHz; 24,75MB L3 cache) |
| GPUs | 4 x GTX 1080 Ti; 11 GB |
| RAM | 192 GB DDR4, ECC |
| Network | 10 Gbps |
| Node-local storage | 480 GB SSD mounted as /lscratch |
| Slurm flags | intel, skylake, 1080ti |
GPU-node with K40
| GPU-node with K40 | ** 1 so far, total of 4 planned ** |
|---|---|
| CPUs | 2x Intel Xeon E5-2690 v4 CPU, 14 Cores/CPU; 2.6GHz; 35 MB SmartCache |
| GPUs | 1x or 2x Tesla K40m80 Ti; 11 GB |
| RAM | 256 GB DDR4, ECC |
| Network | 10 Gbps |
| Node-local storage | 480 GB SSD mounted as /lscratch |
| Slurm flags | intel, broadwell, K40 |
Compute-node
| Compute-node | ** 1 so far, total of 8 planned ** |
|---|---|
| CPUs | 2x Intel Xeon E5-2690 v4 CPU, 14 Cores/CPU; 2.6GHz; 35 MB SmartCache |
| RAM | 256 or 512 GB DDR4, ECC |
| Network | 10 Gbps |
| Node-local storage | 480 GB SSD mounted as /lscratch |
| Slurm flags | intel, broadwell |
GPU-node with 8-10 x GPU
| *GPU-node | ** 1 in preparation ** |
|---|---|
| CPUs | 2x Xeon(R) Gold 6140 CPU; 18 Cores/CPU; 2.3GHz; 24,75MB L3 cache) |
| GPUs | 4 x GTX 1080 Ti; 11 GB |
| RAM | 192 GB DDR4, ECC |
| Network | 10 Gbps |
| Node-local storage | 480 GB SSD mounted as /lscratch |
| Slurm flags | intel, skylake, 1080ti |
Node List
| Node | Partition | CPU | GPU | RAM |
|---|---|---|---|---|
| 01 | gpu | 2x Intel Xeon E5-2690 | GTX 1080 Ti; 11 GB | 256 GB |
| 02 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 03 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 04 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 05 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 06 | bigmem | 2x Intel Xeon E5-2690 | 512 GB | |
| 07 | bigmem | 2x Intel Xeon E5-2690 | 512 GB | |
| 08 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 09 | compute | 2x Intel Xeon E5-2690 | 256 GB | |
| 10 | gpu | 2x Intel Xeon E5-2690 | Tesla K40m80 Ti; 11 GB | 256 GB |
| 11 | gpu | 2x Intel Xeon E5-2690 | 2x Tesla K40m80 Ti; 11 GB | 256 GB |
| 12 | gpu | 2x Intel Xeon E5-2690 | 2x Tesla K40m80 Ti; 11 GB | 256 GB |
| 14 | gpu | 2x Xeon(R) Gold 6140 CPU | 4x GTX 1080 Ti; 11 GB | 187 GB |
| 15 | gpu | 2x Xeon(R) Gold 6140 CPU | 4x GTX 1080 Ti; 11 GB | 172 GB |
| 16 | gpu | 2x Xeon(R) Gold 6140 CPU | 4x GTX 1080 Ti; 11 GB | 187 GB |
| 17 | gpu | 2x Xeon(R) Gold 6140 CPU | 4x GTX 1080 Ti; 11 GB | 187 GB |
To view up-to-date information about the nodes, run sinfo:
sinfo -o "%9n %9R %.4c %.6m %.8d %14f %G"