AgilityAI GPU Servers
Custom Designed GPU Servers for HPC, Training and Inferencing

Major Components inside the NVIDIA DGX A100 System
At the core, the NVIDIA DGX A100 system leverages the NVIDIA A100 GPU, designed to efficiently accelerate large complex AI workloads as well as several small workloads, including enhancements and new features for increased performance over the V100 GPU. The A100 GPU incorporates 40 GB high-bandwidth HBM2 memory, larger and faster caches, and is designed to reduce AI and HPC software and programming complexity.



IBM Power System AC922
IBM POWER9 is the first chip with PCIe Gen4 (2x the bandwidth of PCIe Gen3). The Power AC922 has PCIe Gen4 and other advanced I/O interconnects including CAPI 2.0, OpenCAPI and NVIDIA® NVLink™. Unlike x86-based servers, on the Power AC922 the NVIDIA® NVLink™ enables CPU to GPU connectivity delivering 5.6x the data throughput for today’s data-intensive and AI workloads. The Power AC922 supports up to 6 NVIDIA® Tesla V100 GPUs (16GB or 32GB).
Key Features of IBM Power System AC922
- Faster I/O – up to 5.6x more I/O bandwidth than x86 servers
- PCIe Gen4 – 2x the bandwidth of PCIe Gen3
- POWER9 processor—the latest POWER processor, designed for AI
- Advanced GPUs – up to 6 NVIDIA® Tesla® V100 GPUs with NVLink
- Coherence – share RAM across CPUs & GPUs
- Built scalable from one server to supercomputer


AgilityAI RG408P-SA AMD PCIe Gen4 GPU Server
Key Features
The Applied Data Systems RG408P-SA is a state-of-the-art PCIe Gen4 based GPU server utilizing AMD EPYC Rome Processors. Available with up to eight GPUs. Comes pre-installed with Ubuntu, CUDA, cuDNN, TensorFlow and PyTorch. Combine with ExtremeStor, our high- speed parallel file system-based storage solution for maximum performance.
- Up to Eight GPUs with direct connect architecture for maximum performance
- Up to 8TB DDR4 ECC memory across 32x DIMM slots
- Supports 280w AMD EPYC Rome processors, up to 64-cores each
- Nine PCIe Gen4 x16 slots available (no PCIe switches)
- Up to 24 hot swap 2.5″ drive slots; Four NVMe drives plus Four SATA standard
- Titanium Level, 96% efficient, redundant, 2+2, power supplies
AgilityAI RG204SX-SA AMD SXM4 GPU Server
Key Features
The Applied Data Systems RG204SX-SA is a state-of-the-art SXM4 based GPU server utilizing AMD EPYC Rome Processors. Available with up to four SXM4 GPUs. Comes pre-installed with Ubuntu, CUDA, cuDNN, TensorFlow and PyTorch. Combine with ExtremeStor, our high- speed parallel file system-based storage solution for maximum performance.
- Up to Four NVIDIA Tesla A100 GPUs with high speed SXM4 NV-Link architecture for maximum performance
- Up to 8TB DDR4 ECC memory across 32x DIMM slots
- Supports 280w AMD EPYC Rome processors, up to 64-cores each
- Four PCIe Gen4 x16 slots available (no PCIe switches)
- Four hot swap 2.5″ drive slots; Four NVMe/SATA drives
- Titanium Level, 96% efficient, redundant, 2x, power supplies
