Name: NVIDIA GPU A100 SXM4 80 Go avec dissipateur thermique (699-2G506-0210-320)
Brand: NVIDIA
SKU: 699-2G506-0210-320
Availability: InStock

Spécifications techniques

Fabricant

NVIDIA

Numéro de pièce MFG

699-2G506-0210-320

Mémoire GPU

80GB HBM2e

Interface mémoire

5120-bit HBM2e (High Bandwidth Memory 2e)

Bande passante mémoire

2,039GB/s

Consommation d'énergie

400 W

Ports de sortie

4x DisplayPort 1.4a

Garantie et assistance

-Garantie Premium d’un an

-année garantie sur les pièces et la main d'oeuvre

-année couverture de prêt

-année assistance technique

Assistance technique gratuite

1 (800) 801-8432

96 units in Stock. Lot QTY discount available. Please request a quote if you are interested in a QTY purchase (QTY discount applies in batches of 8, 16, 24, etc.)

⚠️

Export Restriction Notice

This product is subject to U.S. Export Administration Regulations (EAR). The NVIDIA A100 SXM4 80GB may not be sold, shipped, exported, re-exported, transferred, redistributed, or incorporated into any system or device intended for China, including Hong Kong and Macau.

The NVIDIA A100 80GB SXM4 is a high-performance data-center GPU engineered to accelerate the most demanding AI, high-performance computing (HPC), and data analytics workloads. Built on the Ampere architecture, the A100 delivers massive computational throughput, advanced multi-instance GPU (MIG) capabilities, and industry-leading memory bandwidth to meet the needs of modern large-scale computing environments.

Designed for integration into NVIDIA HGX platforms via the SXM4 form factor, the A100 delivers optimized power and supports high-bandwidth interconnects such as NVLink and NVSwitch, making it the premier solution for enterprise AI training, inference, scientific simulation, and hyperscale deployment.

Features

80GB High-Bandwidth HBM2e Memory

Provides 80 GB of ECC-protected HBM2e, enabling extensive models and datasets.
Offers up to 2 TB/s memory bandwidth, among the fastest of any GPU.

NVIDIA Ampere Architecture

Built with 3rd-generation Tensor Cores and improved CUDA cores for accelerated AI, HPC, and data analytics.
Significantly boosts training and inference performance over previous generations.

3rd-Generation Tensor Cores

Specialized hardware for accelerating AI workloads such as training, inference, matrix operations, and scientific computing.
Supports FP64 Tensor Core acceleration, BFLOAT16, TF32, FP16, INT8, and INT4 precision.

Multi-Instance GPU (MIG)

Allows a single A100 GPU to be partitioned into up to seven isolated GPU instances.
Enables maximum utilization and secure multi-tenant operations in data-center environments.

High-Speed Interconnects

Supports bidirectional NVLink at 600 GB/s, enabling multi-GPU scaling with minimal latency.
Integrates into HGX A100 systems with NVSwitch for large-scale GPU clusters.

SXM4 Form Factor

Optimized for server deployment with superior power delivery, thermals, and interconnect bandwidth.
Not a PCIe card — designed for OEM and hyperscale server platforms.

Unmatched Compute Performance

Excellent for AI model training, large-scale inference, simulation, computational science, and big data analytics.
Delivers multi-petaflop AI performance when deployed across multiple GPUs.

ECC Reliability & Secure Operation

Full error correction code (ECC) protection across memory and L2 cache.
Built for mission-critical enterprise workloads requiring accuracy and stability.

Docs & Drivers

NVIDIA A100 SXM4 80GB Spec Sheet

Critiques

Ajouter votre avis

Nom

Masquer mon nom

Messagerie électronique

Évaluation

Veuillez nous faire savoir ce que vous avez aimé ou non dans ce produit afin que les autres acheteurs puissent prendre une décision plus éclairée.

NVIDIA GPU A100 SXM4 80 Go avec dissipateur thermique (699-2G506-0210-320)

Modules complémentaires recommandés

Autres produits à considérer

Vous aimerez peut-être aussi