Workstation GPUs
NVIDIA GeForce RTX 5080
Memory
16 GB
Bandwidth
960 GB/s
Platform
1,801 TOPS
FP16
225.1 TFLOPS
- Maker
- NVIDIA
- Model
- GeForce RTX 5080
- Type
- Workstation GPUs
- Memory
- 16 GB
- Memory type
- GDDR7
- Bandwidth
- 960 GB/s
- Platform
- 1,801 TOPS
- FP16
- 225.1 TFLOPS
- FP8
- 450.2 TFLOPS
- FP4
- 900.4 TFLOPS
- Power
- 360 W
- Form factor
- Founders Edition PCIe 2-slot, 304 x 137 mm (partner cards vary)
- Interface
- PCIe 5.0
- Released
- 2025
- Maker's list price
- $999
Notes on these figures
AI TOPS 1801 = FP4 with sparsity. TFLOPS are dense, from NVIDIA RTX Blackwell architecture whitepaper (source_url): FP4 900.4/1801 sparse, FP8 450.2/900.4 (FP16 accumulate), FP16 225.1/450.2 (FP16 accumulate); with FP32 accumulate FP8 and FP16 are half (225.1 and 112.6 dense). TGP 360 W. ECC not listed. $999, available 2025-01-30 per NVIDIA press release https://nvidianews.nvidia.com/news/nvidia-blackwell-geforce-rtx-50-series-opens-new-world-of-ai-computer-graphics. Specs page: https://www.nvidia.com/en-us/geforce/graphics-cards/50-series/rtx-5080/
Source: the maker's published specifications at images.nvidia.com. A missing figure means the maker does not publish it. Confirm current specifications with the maker or a dealer before you buy.
Comparable workstation gpus
| Model | Memory | Memory type | Bandwidth | Platform |
|---|---|---|---|---|
| NVIDIA RTX PRO 4000 Blackwell | 24 GB | GDDR7 ECC | 672 GB/s | 1,290 TOPS |
| AMD Radeon AI PRO R9700 | 32 GB | GDDR6 (ECC on Linux only) | 640 GB/s | - |
| AMD Radeon PRO W7800 | 32 GB | GDDR6 ECC | 576 GB/s | - |
| NVIDIA GeForce RTX 5090 | 32 GB | GDDR7 | 1,792 GB/s | 3,352 TOPS |
| NVIDIA RTX PRO 4500 Blackwell | 32 GB | GDDR7 ECC | 896 GB/s | 1,617 TOPS |
| AMD Radeon PRO W7900 | 48 GB | GDDR6 ECC | 864 GB/s | - |