+49 (0)331 – 7045331
info@auktionspunkt.de

High-Performance-Hardware - AI-Server, GPU, LLM, Quantencomputing

High-Performance-Hardware - AI-Server, GPU, LLM, Quantencomputing

Los 3: Nvidia Tesla A100 80GB GPU

2026-04-01T08:00:00Z

Startpreis: €1,000

Geschätzer Preis: €3 500 - €7 000

Los 3:

vorherige Zurück zum Katalog nächste

Nvidia Tesla A100 80GB GPU

Die Auktion beginnt in __ Tagen und __ Stunden __ Minuten

Startpreis: €1,000

Geschätzer Preis: €3 500 - €7 000

Beschreibung

for AI Server KI Server GPU-Server
PCIe Tensor Core GPU , Data Center / AI / HPC. 

Technical Specifications

  • Model: NVIDIA A100 Tensor Core GPU, PCIe 4.0 x16, dual-slot, full-height, full-length, passive cooling.
  • Part Numbers: 900-21001-0020-000 (Board), 699-21001-0230-600 (A100 80GB HBM2 PCIe Accelerator).
  • Serial-Number: 1324321150635
  • Architecture: Ampere, 7 nm, data center GPU for AI, HPC, and data analysis.
  • Memory: 80 GB HBM2e, ECC, very high bandwidth (approx. 2 TB/s, 5120-bit interface).
  • Computing performance (typical 80GB PCIe variant): up to 19.5 TFLOPS FP32 (TF32), up to 624 TFLOPS Tensor performance (FP16/BF16 with sparsity).
  • TDP: 300 W, designed for servers with adequate air cooling (passive heatsink).
  • Interface: PCI Express 4.0 x16, compatible with many current Intel Xeon and AMD EPYC server platforms.

Features:

  • Tensor Cores (3. Gen.) für AI-Training und -Inference.
  • Multi-Instance GPU (MIG), partitioning into multiple logical GPUs for multi-tenant workloads.
  • Full ECC support for memory and registers.

 

Applications

  • Training and fine-tuning of large AI/ML models (e.g., LLMs, computer vision, NLP) with high VRAM requirements.
  • High-performance inference workloads, including multi-tenant via MIG (e.g., multiple clients/models on a single card).
  • Classic HPC applications (simulations, numerical mathematics, scientific computing).
  • GPU-accelerated data analysis, data warehousing, ETL, and analytics frameworks.

 

Suchbegriffe:
NVIDIA DGX, NVIDIA, NVIDIA V100, NVIDIA P100, Product Line, Ubuntu LTS, Llama3, Llama4, Mistral, Falcon, Gemma, Google DeepMind, Phi-3 Microsoft, Phi3, Mixtral, DBRX, Grok, Qwen, qwen3, deepseek-r1, StableLM, Olmo, T5, BERT, XLM-RoBERTa, DistilBERT, ALBERT, BLOOM, GPT-Neo, GPT-J, DeepSpeed, Megatron-LM, LLaVA, Vicuna, Alpaca, llava, NVIDIA CUDA, NVIDIA TensorRT, Rack Server, NVIDIA PCIE GPU, AnSys, 3D, Rendering, CFD , CAD / CAM , VMWARE ESXi 8


Zustandsbericht:

Condition: Used in data center. Removed after project completion and now sold after storage. No original packaging. No further details on the features or functionality.