Networking and GPU Compute Solutions for Enterprise AI
NVIDIA DGX™ and NVIDIA HGX™ GPU Servers for AI Training and Inference
Choosing the Right Server GPU for the Job
Whether you are looking to train trillion parameter models or power your next start up, we offer solutions for every business type.
NVIDIA HGX H100
Enhances AI and HPC operations with scalable NVLink technology for superior GPU server connectivity and performance.
NVIDIA HGX H200
Delivers unmatched scalability and performance in AI and HPC with high-speed networking and robust security features.
NVIDIA HGX B100/B200
This GPU for AI achieves up to 15X higher inference performance over the previous generation for massive models such as GPT-MoE-1.8T.
NVIDIA DGX H100/H200
A foundational enterprise AI solution, offering substantial performance improvements with advanced H100 Tensor Core GPUs.
NVIDIA DGX B200
Specializes in deep learning inference, achieving up to 15X greater performance in real-time AI processing.
Server GPU Performance Comparison
| Feature | HGX H100 | HGX H200 | HGX B100/B200 | DGX H100/H200 | DGX B200 |
|---|---|---|---|---|---|
| GPU | 8x HGX H100 | 8x HGX H200 | 8x HGX B100/B200 | 8x NVIDIA H100 Tensor Core | 8x NVIDIA Blackwell GPUs |
| GPU Memory | 640 GB total | 1128 GB total | 1.5 TB total | 640/1,128 GB total | 1.5 TB total |
| Networking | Fourth-generation NVLink 900GB/s, Third-generation NVSwitch | NVIDIA NVLink® 900GB/s, PCIe Gen5 128GB/s | Fifth generation NVIDIA NVLink and Fourth generation NVIDIA NVSwitch™ | 4x OSFP ports, 2x dual-port QSFP112 NVIDIA ConnectX-7 VPI Up to 400Gb/s InfiniBand/Ethernet | 4x OSFP ports, 2x dual-port QSFP112 NVIDIA BlueField-3 DPU Up to 400Gb/s InfiniBand/Ethernet |
| Performance | FP8 Tensor Core 31,662 TFLOPS | FP8 Tensor Core 31, 664 TFLOPS | FP8 B200: 72 petaFLOPS B100: 56 petaFLOPS | 32 petaFLOPS | Training 72 petaFLOPS, Inference 144 petaFLOPS |
Engineering Expertise
Our team of thermal, electrical, mechanical, and networking engineers are skilled in designing solutions designed to your specific requirements.
Solution Architects
AMAX's solution architects optimize IT configurations for performance, scalability, and industry-specific reliability.
Networking
AMAX designs custom networking topologies to enhance connectivity and performance in AI and HPC environments.
Thermal Management
AMAX implements innovative cooling technologies that boost performance and efficiency in dense computing setups.
Compute Optimization
AMAX ensures maximum performance through benchmarking and testing, aligning hardware and software for AI workloads.
From Design to Deployment
AMAX's approach to AI solutions begins with intelligent design, emphasizing the creation of high-performance computing and network infrastructures tailored to AI server applications. We guide each project from concept to deployment, ensuring systems are optimized for both efficiency and future scalability.
AMAX is NVIDIA DGX Elite
Being NVIDIA DGX AI Compute Systems Elite places AMAX among a select group of only 22 partners across North America and Latin America. Our partnership with NVIDIA underscores our commitment to delivering cutting-edge AI server solutions.
NVIDIA DGX™ H200
AI performance is accelerated through NVIDIA's advanced DGX platform.
- 1,128GB GPU memory enhances AI processing.
- Dual Intel Xeon Platinum 8480C processors.
- Up to 400Gb/s InfiniBand/Ethernet connectivity.
- Supports DGX OS, Ubuntu, Red Hat Linux.