Inventec P9000G6 P9000G7 HGX B200 B300 NVL8 180GB Air-Cooled (8) Intel Granite Rapids (2) AI Graphics Card GPU Server

Inventec P9000G6 P9000G7 HGX B200 B300 NVL8 180GB Air-Cooled (8) Intel Granite Rapids (2) AI Graphics Card GPU Server

$1,999,000.00
Sale price  $1,999,000.00 Regular price 
Skip to product information
Inventec P9000G6 P9000G7 HGX B200 B300 NVL8 180GB Air-Cooled (8) Intel Granite Rapids (2) AI Graphics Card GPU Server

Inventec P9000G6 P9000G7 HGX B200 B300 NVL8 180GB Air-Cooled (8) Intel Granite Rapids (2) AI Graphics Card GPU Server

$1,999,000.00
Sale price  $1,999,000.00 Regular price 

• Support dual Intel® Xeon® 6 processor with E-cores and P-cores

• Powered by the NVIDIA HGX™ B300 system with air cooling solution

• Best GPU Communication with NVIDIA NVLink™ Bridge

• Support DDR5 DIMM, 6400 MT/s @ 1DPC, 5200 MT/s @ 2DPC

• Accommodate up to 8 U.2 NVMe SSD

• Inventec-designed retimer board boosts CPU-GPU data flow

• Increase power efficiency and reliability by decoupling 12V and 54V power source

• Modular design for scalability and serviceability

Features & Compatibility

High-Speed Interconnects for Smooth AI Execution Efficient data movement between system components is essential for maximizing AI performance. The P9000G7 (AC) supports DDR5 memory up to 6400 MT/s at 1DPC and 5200 MT/s at 2DPC, enabling high memory bandwidth for CPU-side processing and fast access to training data. NVIDIA NVLink™ Bridge tightly connects GPUs, enabling fluid collaboration and high-throughput processing across multi-GPU workloads. To fur ther enhance overall system responsiveness, the P9000G7 (AC) includes an Inventec-designed retimer board that improves signal integrity and accelerates data exchange between CPU and GPU modules, ensuring smooth performance in data-intensive environments. Designed for Uptime and Maintenance Efficiency In AI infrastructure, agility and uptime are as critical as raw performance. That is why the P9000G7(AC ) is engineered with a modular, service-centric approach. Its toolless design provides quick access to all critical components to improve serviceability and reduce maintenance effort. The clear separation between the host and GPU modules enhances air flow zoning and simplifies cooling management, ensuring stable performance and operational agility across dense computing deployments. The Backbone of Next-Generation AI Infrastructure.

Form Factor

10U Rackmount

W x H x D: 448x 441x 850 mm

(17.63x 17.36x 33.46inch)

Processor

5th Gen AMD EPYC™ Processors up to 400W

GPU Module

NVIDIA HGX™ B200 (AC)

PCIe Slot

(from MLB)

Option(1) 1x FHHLDW

Option(2) 2x FHHLSW

Option(3) 1x OCP 3.0 SFF + 1x FHHLSW

PCIe Slot

(from Switch)

8x Low Profile card (PCIeG5x16)

Storage

(from Switch)

8x U.2 NVMe SSD bays

Storage

(from MLB)

4x U.2 NVMe SSD bays from CPU (option)

2x M.2 NVMe SSD

Management Port

1x RJ45 for BMC(AST2600) remote management

Cooling

Air cooling

Power Supply

3300W 54V ATS PSU, support 4+2 redundancy 2400W 12V PSU, support 1+1 redundancy

Support S5 power and thermal for DPU

Fan

15x 8086 Fan for GPU cooling

5x 8056 Fan for CPU cooling

You may also like