Name: A100X Converged Accelerator 80GB GPU
Brand: NVIDIA
Price: 25499.95 USD
Availability: InStock

NVIDIA 900-21004-0030-000 A100X Converged Accelerator 80GB GPU

price

$25,499.95

In Stock: 16
Manufacturer: NVIDIA
Model or Part: 900-21004-0030-000
Condition: Refurbished
Weight: 3.00lb

Qty

Add to Cart

Buy Now

Add to Wish List

Shipping Information

United States Shipping

Orders over $200 qualify for Free Super-Saver Shipping! We ship at our discretion via the most cost-effective method using FedEx, UPS, or USPS.

To ensure your order is processed efficiently, all shipping charges must be pre-paid at checkout. We are unable to bill directly to third-party shipper accounts or use customer-provided labels for domestic shipments.

International Shipping

International orders begin processing as soon as payment is confirmed.

Pre-paid Shipping: For your convenience, you can select and pay for international shipping directly at checkout.
Carrier Accounts: Alternatively, we can bill transport costs directly to your FedEx account number. (Note: We cannot bill to DHL, TNT, or UPS accounts).
Customer Labels: We are also happy to ship internationally using pre-paid labels provided by you for any carrier.

Expedited and Rush Shipping

Need it faster? HardwareJet.com offers 1 or 2-day expedited shipping options during checkout for eligible items.

To get your order to the front of the line, expedited orders include a small rush fee in addition to the carrier's shipping charges. If an item is not available for expedited shipping, these options will not appear in your cart.

Product Information

Item Condition

New - Sealed or Open Box Condition
Excellent - Almost-New, Missing Original Packaging
Refurbished - Seller Refurbished, Good Condition, Previously Used Item

To bring you an even wider selection, some of our items are shipped directly from our trusted partners. Delivery times may vary for these items.

💬 Recently Purchased by Your Peers

Description

Overview

The NVIDIA A100X Converged Accelerator 80GB GPU (part number 900-21004-0030-000) delivers industry leading performance for compute intensive workloads. Built on the latest Hopper architecture, this accelerator combines massive tensor processing power with high bandwidth memory to accelerate AI training, inference, high performance computing (HPC) and data analytics in a single, unified platform. With 80 gigabytes of high speed HBM3 memory, the A100X provides the capacity and speed required for the most demanding models and simulations.

Key Features

80 GB of HBM3 memory delivering up to 2.5 TB/s memory bandwidth
Hopper GPU architecture with third generation Tensor Cores for AI acceleration
Support for FP64, FP32, TF32, BF16, FP16 and INT8 precision formats
NVLink and PCIe Gen5 connectivity for scalable multi-GPU configurations
Integrated Multi-Instance GPU (MIG) technology to partition the GPU into up to 7 isolated instances
Enhanced security features including hardware root of trust and secure boot

Technical Specifications

GPU Model: NVIDIA A100X Converged Accelerator
Memory Capacity: 80 GB HBM3
Memory Bandwidth: 2.5 TB/s
CUDA Cores: 10800
Tensor Cores: 336
Peak FP64 Performance: 19.5 TFLOPS
Peak FP32 Performance: 39.0 TFLOPS
Peak TF32 Performance: 156 TFLOPS
Peak BF16/FP16 Performance: 312 TFLOPS
PCIe Interface: PCIe Gen5 x16
NVLink: 3-way NVLink with 600 GB/s total bandwidth
Power Consumption: 400 W TDP
Form Factor: Dual slot, full height

Applications

Deep learning model training and large language model development
Real-time AI inference for autonomous systems and edge computing
Scientific simulations in physics, chemistry and climate modeling
High performance data analytics and graph processing
Financial risk modeling and quantitative analysis
Rendering and visual effects in media and entertainment

Whether you are building a supercomputer cluster, an AI research lab or a data center focused on next generation workloads, the NVIDIA A100X Converged Accelerator 80GB GPU provides the raw performance, memory capacity and flexibility to accelerate your most demanding applications.