NVIDIA H100 Solutions| SabrePC

NVIDIA HGX H200 SXM5 Solutions

ES4-8005921

Dual 3rd Gen Xeon NVIDIA HGX H200 3U Server

Processor: 2x 4th/5th Gen Intel Xeon Scalable Family
Drive Bays: 8x 2.5" SATA/NVMe Hot-Swap
Supports: 4x NVIDIA H200 SXM GPUs
Rack Height: 3U

ES4-7992039

Dual 4th/5th Gen Xeon NVIDIA HGX H200 6U Server

Processor: 2x 4th/5th Gen Intel Xeon Scalable family
Drive Bays: 8x 2.5" Hot-Swap
Supports: 8x NVIDIA H200 SXM GPUs
Rack Height: 6U

ES4-8040017

Dual EPYC 9004 NVIDIA HGX H200 6U Server

Processor: 2x AMD EPYC 9004-Series
Drive Bays: 24x 2.5" Hot-Swap
Supports: 8x NVIDIA H200 SXM GPUs
Rack Height: 8U

Servers and Workstations Supporting NVIDIA H100 PCIe

ES2-6107209

Quad NVIDIA H100 PCIe AMD EPYC 9004 2U Server

1x AMD EPYC 9004 CPU
Up to 3TB DDR5 ECC Memory
Supports 4x NVIDIA H100 PCIe
2x 2.5" & 4x 3.5” Hotswap Drive Bays
3-Year Warranty Included

EWS-6159933

4x NVIDIA H100 Dual Intel Xeon Scalable Rackmount Workstation

2x 4th Gen Intel Xeon Scalable
Up to 4TB DDR5 ECC Memory
Supports 4x NVIDIA H100 PCIe
8x 3.5” Hotswap Drive Bays
3-Year Warranty Included

ES4-6131647

8x NVIDIA H100 Dual AMD EPYC 9004 4U Server

2x AMD EPYC 9004 CPUs
Up to 6TB DDR5 ECC Memory
Supports 8x NVIDIA H100 PCIe
8x 2.5/3.5" Hotswap Drive Bays
3-Year Warranty Included

NVIDIA HGX H100 SXM5 Servers

ES4-6164107

NVIDIA HGX H100 Dual Intel Xeon Scalable 4U Server

2x 4th/5th Gen Intel Xeon Scalable
Up to 8TB DDR5 ECC Memory
NVIDIA HGX H100 System Board
4x H100 80GB SXM5 + NVLink
6x 2.5” Hotswap Drive Bays
3-Year Warranty Included

ES4-6163990

NVIDIA HGX 8x H100 Dual Intel Xeon Scalable 8U

Starting at

$270,653.52

2x 4th Gen Intel Xeon Scalable
Up to 6TB DDR5 ECC Memory
NVIDIA HGX H100 System Board
8x H100 80GB SXM5 + NVLink
24x 2.5” Hotswap Drive Bays
3-Year Warranty Included

Configure

ES4-6101416

NVIDIA HGX 8x H100 Dual AMD EPYC 9004 8U

2x AMD EPYC 9004
Up to 6TB DDR5 ECC Memory
NVIDIA HGX H100 System Board
8x H100 80GB SXM5 + NVLink
24x 2.5” Hotswap Drive Bays
3-Year Warranty Included

The NVIDIA H100 Chip

NVIDIA is fabricating the most advanced GPU chip to date, set to release Q3 2022. The NVIDIA H100, built on the new Hopper Architecture, can be had as an SXM5 chip module or as a PCIe card to supersede the Ampere A100. The H100 supports mixed-precision processing with capabilities to train AI on FP8 for 6x performance gains over the last generation. NVIDIA’s Hopper Architecture can power your next SabrePC-built complex AI workstations and/or servers.

Train Faster and Harder Together

NVIDIA H100s are designed to work together to offer the best performance. SabrePC recognizes AI is becoming essential in all industries as companies manufacture intelligence to process every engagement, product, and recommendation. Talk to our industry experts to develop the best solutions for you as we become an extension of your team. Train and develop AI models faster and harder as we work together to build the perfect system for you. We offer EDU discounts and B2B bulk pricing.

NVLink Chip to Chip

NVIDIA’s vision of the H100 reveals the future of AI with NVLink Chip-to-Chip to connect not only GPUs together but also NVIDIA’s new CPUs and SmartNICs for unprecedented bidirectional bandwidth. NVIDIA introduces the H100 CNX and NVIDIA Grace Hopper. C2C designs will allow direct communication between components to free up CPU, PCIe lane, and memory. Each iteration of NVIDIA’s H100 will be the most powerful and most efficient GPU solution ever.