NVIDIA B200 180GB vs AMD Strix Halo 128GB for local LLMs
0 reports on the NVIDIA B200 180GB and 82 on the AMD Strix Halo 128GB
Which is faster for local LLMs?
Nobody has reported the same model at the same quant on both cards yet, so there's no measured head-to-head. Memory bandwidth, which caps how fast a card can write, is 31.3x higher on the NVIDIA B200 180GB (8,000 GB/s on the NVIDIA B200 180GB, 256 on the AMD Strix Halo 128GB). The estimates below compare them on the models people run most.
- Memory
- 180GB
- Memory bandwidth
- 8,000 GB/s
- FP16 compute
- 2250 TFLOPS
- Reports
- 0
- Memory
- 128GB unified
- Memory bandwidth
- 256 GB/s
- FP16 compute
- 59 TFLOPS
- Reports
- 82
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape.
| Model | NVIDIA B200 180GB | AMD Strix Halo 128GB |
|---|---|---|
| Qwen3.8 27B | 81.5 t/sFP16 | 3.5 t/sFP16 |
| Qwen3.8 125B · 6B active | 266.2 t/sQ8_0 | 12.6 t/sQ6_K |
| DeepSeek V4 Flash 284B · 13B active | 435.6 t/sQ4_K_S | 33.5 t/sQ2_K |
| Qwen3.6 35B · 3B active | 293.3 t/sFP16 | 12.6 t/sFP16 |
| Qwen3.6 27B | 81.5 t/sFP16 | 3.5 t/sFP16 |
| Gemma 4 26B · 4B active | 220.0 t/sFP16 | 9.5 t/sFP16 |
Change the context, RAM or model in the calculator: NVIDIA B200 180GB · AMD Strix Halo 128GB
Every report on each card, with its source: NVIDIA B200 180GB · AMD Strix Halo 128GB