Leap Ahead of the Competition
with GPU-Accelerated Computing
Get faster time-to-results without the traditional equipment headaches
Get Ready for The Future with More Powerful, More Efficient Computing
Penguin Computing™ delivers targeted, modular and complementary AI & Analytics architectures for AI/ML and High Performance Data Analytic Pipeline. Our solutions shorten time to insight and discovery by removing the complexities involved in designing, deploying, and supporting customer’s AI & Analytics infrastructure.
Our GPU-accelerated compute delivers best of breed solution that powers our Technology Practices, especially in HPC Technology as well as AI & Analytic Technology. Our infrastructure offering includes both 19” EIA and 21” Tundra (Open Compute) infrastructure enabling higher density, and alternative non-air cooling for more compute per rack.
Penguin Computing team (2019 NVIDIA® HPC Preferred OEM Partner of the Year) is experienced with building both CPU and GPU-based systems as well as the storage subsystems required for this level of data analytics, the outcome of moving to a GPU-accelerated strategy is superior performance by all measures, faster compute time, and reduced hardware requirements.
GPU-Accelerated Servers
19″ EIA Servers
| 2U | Processor | PCIe Slots | GPU(s) Supported |
|---|---|---|---|
| Altus XE2318GT | AMD EPYC™ 9004 Series | 8x PCIe Gen5 x16 FHFL, 2x PCIe Gen5 x8 LP | NVIDIA H100 NVL, L40, L40S |
| Relion XE2318GT | 4th Gen Intel® Xeon® Scalable Processors | 8x PCIe Gen5 x16 FHFL, 2 x PCIe Gen5 x8 LP | NVIDIA H100 NVL, L40, L40S |
| 3U | Processor | PCIe Slots | GPU(s) Supported |
| Altus XE3314GTS | AMD EPYC™ 9004 Series | 6x PCIe Gen5 x16 LP Slots | NVIDIA HGX H100 80GB HBM3 SXM5 x4 NVIDIA HGX H200 141GB HBM3e SXM5 x4 |
| Relion XE3314GTS | 4th Gen Intel® Xeon® Scalable Processors | 6x PCIe Gen5 x16 LP Slots | NVIDIA HGX H100 80GB HBM3 SXM5 x4 NVIDIA HGX H200 141GB HBM3e SXM5 x4 |
| 4U | Processor | PCIe Slots | GPU(s) Supported |
| Altus XE4318GT | AMD EPYC™ 9004 Series | 8x PCIe Gen5 x16 FHFL, 2x PCIe Gen5 x8 LP | NVIDIA H100 NVL, L40, L40S |
| Relion XE4318GT | 4th Gen Intel® Xeon® Scalable Processors | 8x PCIe Gen5 x16 FHFL, 10x PCIe Gen5 x16 LP | NVIDIA H100 NVL, L40, L40S |
| 5U | Processor | PCIe Slots | GPU(s) Supported |
| Altus XE5318GTS | AMD EPYC™ 9004 Series | 12x PCIe Gen5 x16 Slots | NVIDIA HGX H100 80GB HBM3 SXM5 x8 NVIDIA HGX H200 141GB HBM3e SXM5 x8 |
| Relion XE5318GTS | 4th Gen Intel® Xeon® Scalable Processors | 12x PCIe Gen5 x16 LP Slots, 1x PCIe Gen4 x16 LP Slot | NVIDIA HGX H100 80GB HBM3 SXM5 x8 NVIDIA HGX H200 141GB HBM3e SXM5 x8 |
| Altus XE5318GTO | AMD EPYC™ 9004 Series | 12x PCIe Gen5 x16 Slots | AMD Instinct MI300X 192GB HBM3 OAM x8 |
| Relion XE5318GTO | 4th Gen Intel® Xeon® Scalable Processors | 12x PCIe Gen5 x16 Slots | AMD Instinct MI300X 192GB HBM3 OAM x8 |
21″ OCP Servers
| 1OU | Processor | PCIe Slots | GPU(s) Supported |
|---|---|---|---|
| Altus XO1314GT | AMD EPYC™ 9004 Series Processors | 4x PCIe Gen4 x16 FHFL (GPU), 2x PCIe Gen4 x16 (LP) | NVIDIA L40 |
| Relion XO1314GT | 4th Gen Intel® Xeon® Scalable Processors | 4x PCIe Gen4 x16 FHFL (GPU), 2x PCIe Gen4 x16 (LP) | NVIDIA L40 |
Selected applications supported by NVIDIA-based Penguin Computing GPU servers:
- Amber
- ANSYS Fluent
- Gaussian
- Gromacs
- LS-DYNA
- NAMD
- OpenFOAM
- Simulia Abaqus
- VASP
- WRF
Selected deep learning frameworks supported by NVIDIA-based Penguin Computing GPU servers:
- Caffe2
- Microsoft Cognitive Toolkit
- MXNET
- Pytorch
- TensorFlow
- Theano
Benefits of GPU-Accelerated Computing
- Computing Power/Speed A single GPU can offer the performance of hundreds of CPUs for certain workloads. In fact, NVIDIA, a leading GPU developer, predicts that GPUs will help provide a 1000X acceleration in compute performance by 2025.
- Efficiency/Cost Adding a single GPU-accelerated server costs much less in upfront, capital expenses and, because less equipment is required, reduces footprint and operational costs. Using libraries also allows organizations to use GPU acceleration without in-depth knowledge of GPU programming, reducing the investment of time required to achieve results.
- Flexibility The inherently flexible nature of GPU programmability allows new algorithms to be developed and deployed quickly across a variety of industries. According to Intersect360 Research, 70% of the most popular HPC applications, including 10 of the top 10, have built-in support for GPUs.
- Long-Term Benefits Adding GPU-accelerated computing now prepares you for the artificial intelligence (AI) revolution, which also relies in GPU-accelerated computing. This inevitable increase on the reliance on GPUs means that early adopters will enjoy not only greater computing power over time but have a greater margin of difference over time than competitors who do not migrate to GPU-accelerated computing.
