Overview
The NVIDIA Quadro RTX 8000 is the Turing-generation professional flagship with 48GB GDDR6 ECC, 4,608 CUDA cores, and 576 Tensor Cores at 295W. NVLink support allows two cards to pool 96GB of memory. With 130.5 TFLOPS Tensor performance and proven Quadro enterprise certification, it provides strong large-model AI capacity at a competitive price vs current-generation Ada cards.
Key Features
- 48GB GDDR6 ECC with NVLink for 96GB dual-card configuration
- Turing architecture: 4,608 CUDA + 576 Tensor Cores
- 130.5 TFLOPS Tensor and 16.3 TFLOPS FP32 performance
- 672 GB/s memory bandwidth at 295W TDP
- Proven Quadro enterprise driver stability and ISV certification
Ideal For
AI researchers and enterprise teams needing cost-effective 48GB VRAM (or NVLink-scaled 96GB) for large model inference and training on a mature professional GPU platform.
Feature
4,608 NVIDIA® CUDA® Cores
Feature
576 NVIDIA® Tensor Cores
Feature
72 NVIDIA® RT Cores
Feature
Up to 672GB/s Memory Bandwidth
Feature
10 Giga Rays/s Rays Cast
Feature
16.3 TFLOPS FP32 Performance
Feature
32.6 TFLOPS FP16 Performance
Feature
65.8 TOPS INT8 Performance
Feature
130.5 TFLOPS of Tensor Operation
Max. Power Consumption
295W
Warranty
1 year manufacturer replacement or repair
$4,915.00
Prices may vary. Verify on vendor site.
Quick Specs
- Feature
- 130.5 TFLOPS of Tensor Operation
- Max. Power Consumption
- 295W
- Warranty
- 1 year manufacturer replacement or repair
Tags
llm-inferencellm-traininggenerative-aicomputer-visionsimulation