Close
Menu
Cisco SMARTnet Authorized Reseller
85
HPE P35962-001 NVIDIA A100 40GB PCIe Accelerator View larger
Previous Next

HPE P35962-001 NVIDIA A100 40GB PCIe Accelerator

New

High-performance 

×
   
 


  • New

Detailed Specifications

HPE P35962-001

NVIDIA A100 40 GB PCIe Accelerator

TheHPE P35962-001 NVIDIA A100 40 GB PCIe Accelerator is a high-end computational GPU designed to dramatically accelerate AI, machine learning, and high-performance computing tasks. Built onNVIDIA's Ampere GA100 architecture, it delivers massive computing horsepower with 40 GB of HBM2 memory and advanced GPU features. This Accelerator is ideal for data centers, research institutions, and AI-driven enterprises seeking blazing-fast training and inference. Its dual-slot PCIe form factor supports seamless integration in standard server chassis, while passive cooling ensures optimized thermal design. Whether deployed for deep neural network training, scientific modeling, or compute-heavy simulations, the A100 GPU delivers industry-leading performance and future-proof scalability.

Features

Designed for demanding computational workloads, the NVIDIA A100 GPU in thisHPE-branded version delivers unparallelled throughput for data-intensive tasks. Featuring 40 GB of high-bandwidth HBM2 memory delivering up to ~1.6 TB/s bandwidth, it enables massive datasets to be processed rapidly—critical for large AI models and HPC applications Under the hood, the A100 offers exceptional compute capability: 6,912 CUDA cores, 432 third-generation Tensor Cores, and peak performance reaching ~312 teraflops for Tensor operations, with robust FP64, FP32, and INT precision support . Leveraging PCIe Gen4 ×16 interface, it achieves optimized host connectivity with accelerated data transfer speeds 

Advanced Options

Built as a full-height, dual-slot card with passive cooling, the A100 requires strong system airflow, integrates easily into data center servers, and supports NVLink bridging between adjacent GPUs for high-bandwidth interconnects (~600 GB/s) . Equipped with ECC memory protection andNVIDIA GPU Boost for adaptive clock optimization, this Accelerator ensures both reliability and sustained peak performance in production-grade environments Engineered for scale, the NVIDIA A100 excels in both raw speed and operational endurance. With its 40 GB HBM2 memory capable of ~1.6 TB/s bandwidth, the GPU expertly handles massive data movements and model loads with minimal bottleneck 

Product Key Features

  • NVIDIAAmpere GA100 architecture with massive GPU compute power

  • 40 GB HBM2 memory delivering up to ~1.6 TB/s bandwidth

  • Supports PCIe Gen4 ×16 connectivity with passive cooling design

  • Delivers up to ~312 TFLOPS Tensor compute via CUDA and Tensor Cores

  • ECC-enabled for error-resilient memory and PCIe data integrity

  • NVLink bridge support for GPU-to-GPU communication (~600 GB/s)

Reliability & Performance

. The GA100 architecture—featuring thousands of CUDA and 3rd-gen Tensor Cores—powers deep learning and HPC workloads at up to ~312 TFLOPS, delivering unprecedented AI throughput in a single card For mission-critical reliability, the card includes memory-level ECC and supports NVIDIA GPU Boost to dynamically adjust clock speeds within thermal and power budgets . Its passive dual-slot design integrates smoothly into air-cooled server environments, while NVLink support enables high-speed GPU-to-GPU data transfer, essential for large-scale multi-GPU training setups Dual-slot dimensions and full-height form factor position the A100 as a robust workhorse in modern data centers. This platform not only delivers extraordinary performance but is also designed for long-term resilience, making it a solid investment for evolving compute infrastructures.

Key Features :

  • NVIDIAGA100 GPU—Ampere architecture

  • 40 GB HBM2 memory (~1.6 TB/s bandwidth)

  • PCIe Gen4 ×16, passive dual-slot form factor

  • ~6,912 CUDA cores, 432 Tensor Cores

  • Peak Tensor performance up to ~312 TFLOPS

  • ECC memory protection

  • NVLink support (~600 GB/s inter-GPU bandwidth)

  • NVIDIA GPU Boost dynamic clock scaling

SpecificationDetails
Model Number P35962-001
GPU Model NVIDIA A100 (Ampere GA100) Accelerator
GPU Memory 40 GB HBM2
Memory Bandwidth ~1,555–1,600 GB/s
Compute Performance Up to ~312 TFLOPS Tensor compute
CUDA Cores / Tensor Cores 6,912 CUDA cores / 432 Tensor Cores
PCIe Interface PCIe Gen4 ×16
Form Factor Full-height, full-length; dual-slot passive cooling
NVLink Support Yes – up to 600 GB/s total interconnect bandwidth
ECC Support Enabled by default
Power Consumption 250 W (TDP)
Cooling Type Passive heatsink (requires system airflow)
Warranty (Refurbished) Typically 90-day
Key Technologies ECC, GPU Boost, NVLink bridge, Multi-Instance GPU (MIG)

  • All packages are shipped same day if the order is placed before 3PM CST

    We use UPS/FedEx to ship the packages with different service levels offered by UPS/FedEx

    Overnight Shipping take 1 business day
    Expedited Shipping take about 2-3 business days
    Ground shipping take about 3-5 business days

    You can select the desired service level at Checkout and we will make sure the package is shipped as per the selected service level.

  • We Accept

    i) PayPal
    ii) All Credit & Debit Cards
    iii) Purchase Orders
    iv) Check
    v) Bank Transfer/ACH/Wire Transfer

    To process the order through Paypal/CC, you can simply checkout and process the payment.

    If you want to issue us Purchase Order/Check/Bank Transfer, Please contact us at the following details

    Call: 469.458.9636
    Email: sales@dcsupplies.net

  • If you have any queries, please contact us at the following details and we will be happy to help

    Call
    469.458.9636
    469.458.9638

    Email
    sales@dcsupplies.net

Close