PPD Accelerated Computing Facility has a number of machines with various Nvidia GPUs: Titan RTX, Titan V, Titan X and Tesla P100, as well as an Arria 10 FPGA board for OpenCL development. For access please contact
Dr Abdeslem Djaoui
The machines can be used for various purposes from prototyping, benchmarking to production work using single or multiple GPUs.
The list of hosts and their specifications follows:
| HEPACC01 | |
|---|
| CPU | DUAL Intel Xeon E5-26702.6GHz 8 core processors |
| RAM | 64 GB |
| GPU | 2x NVIDIA TITAN V |
| GPU memory | 12 GB HBM2 |
| GPU cores | 5120 CUDA Computing Cores (1.5 GHz Boost Clock) |
|
HEPACC02 | |
| CPU | DUAL E5-2680v4 - Intel 14 Core Xeon 2.4GHz |
| RAM | 128 GB DDR4 Memory |
| GPU | 2x NVIDIA TITAN V |
| GPU memory | 2x 12 GB HBM2 |
| GPU cores | 5120 CUDA Computing Cores (1.5 GHz Boost Clock) |
|
HEPACC03 | |
| CPU | DUAL E5-2640v4 - Intel 10 Core Xeon 2.4GHz |
| RAM | 128 GB DDR4 |
| GPU | 4x Tesla P100 |
| GPU memory | 4x 16 GB HBM2 memory |
| GPU cores | 3584 CUDA Computing Cores each (1.5 GHz Boost Clock) |
| Interconnect | SXM2 NVLink interconnect between the 4 GPUs |
| SSD | 200GB |
|
HEPACC04, HEPACC05 and HEPACC06 |
| CPU | DUAL Xeon Silver 4116 2.1GHz, 3.0GHz Turbo,12 Core |
| RAM | 192GB (24x8GB) DDR4 |
| GPU | 2x Titan RTX GPU Computing Cards |
| GPU memory | 24GB GDDR6 memory |
| GPU cores | 4608 CUDA + 576 Tensor Cores |
| SSD | DUAL 1TB SATA |
|
HEPACC07 | |
| CPU | DUAL Intel Xeon E5-2650 v4 2.2GHz, 12 Core |
| RAM | 64 GB RDIMM |
| GPU | 2x Titan X Pascal GPU Computing Cards
|
| GPU memory | 12 GB GDDR5X |
| GPU cores | 3584 CUDA cores |
| SSD | 960GB |
A number of application containers (e.g. TensorFlow) are available on each of these nodes.
The FPGA node for OpenCL development containing a Bittware A10PL4 FPGA board is not available for general use, but access can be arranged upon request.
