BRIN's Scientific Computation Services
Hardware – MAHAMERU 3
Gen 2 - Computing Science Nodes
Computating science nodes are consist of CPU and GPU nodes. In total, 792 cores CPU with 4.3 TB memory and 6 GPUs. Each node is connected through the Infiniband network.
CPU node
HPE Apollo r2200
(Intel(R) Xeon(R) CPU 2.10GHz)
- 28 nodes
- 1008 cores
- 3.5 TB total memory (128 GB/node)
GPU node
HPE DL 380 Gen 9
(Intel(R) Xeon(R) CPU E5-2695)
- 4 nodes
- 144 cores CPU
- Tesla P100 GPU
- 2 TB total memory (512GB/node)
Dual GPU node
Dell Power Edge R740XD
(Intel(R) Xeon Gold 6140 2.3GHz)
- 2 nodes
- 72 cores CPU
- Dual Tesla P100 GPU
- 768 GB total memory (384GB/node)
Gen 2 - Render Farm Nodes
Server hardware
- HPE Apollo r2200 (10 nodes CPU)
- Asus ESC4000 G4 (10 nodes quad GPU)
Gen 2 - Storage Cluster
- Dell Power Edge R740XD
- 2.6 PB in total, distributed in 20 nodes (132 TB/node)
- Divided into Hadoop cluster (10 nodes)* and storage cluster (10 nodes)*.
*currently in development.
Gen 1
Server hardware
- Fujitsu RX350 S7
- Fujitsu RX200 S7
- Fujitsu CX270 S1
Storage
- Storage: Fujitsu Eternus DX1/200 S3
Network
- Dual Gigabit Network
*Currently inactive, dev only.
