NVIDIA B100 vs NVIDIA GeForce RTX 3070 — GPU-Vergleich (Apr 2026)
NVIDIA B100 (192GB HBM3e, 1,750 TFLOPS FP16, Blackwell) vs NVIDIA GeForce RTX 3070 (8GB GDDR6, 20.3 TFLOPS FP16, Ampere)
|
NVIDIA B100
192GB HBM3e · Blackwell
|
NVIDIA GeForce RTX 3070
8GB GDDR6 · Ampere
|
||
|---|---|---|---|
| Spezifikationen | |||
| Hersteller | NVIDIA | NVIDIA | |
| Architektur | Blackwell | Ampere | |
| VRAM | 192 GB HBM3e | 8 GB GDDR6 | |
| Bandbreite | 8,000 GB/s | 448 GB/s | |
| FP16 (Tensor) | 1750.0 TFLOPS | 20.3 TFLOPS | |
| FP32 | 60.0 TFLOPS | 10.2 TFLOPS | |
| TDP | 700W | 220W | |
| Erscheinungsjahr | 2024 | 2020 | |
| Segment | Data center | Consumer | |
| Am besten geeignet für | AI training large-scale inference | Budget inference gaming | |
| Cloud-Preise | |||
| Günstigste On-Demand | — | — | |
| Günstigste Spot | — | — | |
| Anbieter | 0 | 0 | |
NVIDIA B100
NVIDIA GeForce RTX 3070