NVIDIA GB10 Grace Blackwell vs NVIDIA GeForce RTX 5090

Side-by-side comparison of specifications, performance, and efficiency.

Specifications

NVIDIA GB10 Grace Blackwell NVIDIA GeForce RTX 5090
Architecture
Blackwell Blackwell
Launch Year
2025 2025
Form Factor
Integrated PCIe
VRAM
128 GB32 GB
Memory Bandwidth
273 GB/s1.8 TB/s
TDP
180 W575 W
Process Node
4nm

Performance (TFLOPS)

NVIDIA GB10 Grace Blackwell NVIDIA GeForce RTX 5090
FP64
1.6 TFlops
FP32
31.0 TFlops 104.8 TFlops
FP16
419.1 TFlops
sparse not published
BF16
419.1 TFlops
sparse not published
FP8
838.2 TFlops
sparse not published
INT8
838.2 TFlops
sparse not published

FLOPS by Precision

Efficiency — TDP vs FP32 Performance

Analysis

Most Efficient
NVIDIA GeForce RTX 5090
0.18 TFLOPS/W

Where to Rent

NVIDIA GeForce RTX 5090

ProviderInstancePrice/GPU-hrChecked
Vast.ai
$0.34 just now View →
Excess Supply
n3-RTX-5090x1 $0.65 just now View →
RunPod
RTX 5090 (Community Cloud) $0.69 just now View →
RunPod
RTX 5090 (Secure Cloud) $0.99 just now View →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.