The Trasgo group manages the following computational resources associated to different scientific funded projects.
Trasgo Heterogenous Cluster
Interconnection network
40 Gbps Ethernet (1 common HUB)
Computation servers
- Gorgon: AMD Epyc 7713, 2 CPUs with 64 cores each one, 512 GB RAM. 4 hardware accelerators: 1 GPU Nvidia A100 (thanks to NVIDIA Hardware Donation Program) + 3 GPU Nvidia Titan Black (thanks to NVIDIA Research and Education Center Programs).
- Manticore: Intel Xeon Scalable Platinum 8260, 2 CPUs with 24 cores (48 threads) each one, 256 GB RAM. 4 hardware accelerators: 2 GPU Nvidia Volta V-100, 2 GPU AMD GFX900.
- Medusa: Intel Xeon Scalable Silver 4208, 2 CPUs with 8 cores (16 threads) each one, 64 GB RAM, 2 hardware accelerators: 1 GPU Nvidia Titan Z (thanks to NVIDIA Research and Education Center Programs), 1 FPGA Intel Stratix 10.
- Chimera: Intel Xeon E5-2620, 2 CPUs with 6 cores each one, 32 GB RAM. Hardware accelerators: 1 Intel XeonPhi (Knights Corner) MIC P4304-2, 1 GPU Nvidia Titan Black.
- Hydra: Intel Xeon E5-2609, 1 CPU with 6 cores, 62 GB RAM. 4 hardware accelerators: 2 GPUs Nvidia Tesla K40, and 2 GPUs Nvidia Titan Black.
- Heracles: Dell PowerEdge R815 with 4 CPUs AMD Opteron with 16 threads each one, 256 GB RAM.
Frontends
2 servers to access, compile and launch works to the computation servers, shared disk for user accounts, backups, etc.
Acknowledgments
These computational resources were partially supported by funded research projects: Junta de Castilla y León and European Union (FEDER), projects VA226P20 and VA082P17, and Ministerio de Ciencia, Innovación y Universidades, project TIN2017-88614-R, and Ministerio de Economía y Competitividad, project TIN2014-58876-P.
These computational resources were also partially supported by grants from NVIDIA (NVIDIA A100 thanks to the Academic Hardware Grant Program, NVIDIA Titan Blank and Titan Z GPUs thanks to the NVIDIA Research and Education Center Programs).