AMD Radeon RX 6900 XT × Meta Llama 3.1 8B Instruct
Meta Llama 3.1 8B Instruct on AMD Radeon RX 6900 XT
50.3best gen tok/s
1,600best prompt tok/s
640max context tested
9results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 50.3 | 1,170 | LocalScore ↗ | 512+128 | 1,130 ms | llamafile 0.9.2 |
| Q6_K | 44.5 | 1,201 | LocalScore ↗ | 512+128 | 1,122 ms | llamafile 0.9.2 |
| Q4_K - Medium | 40.3 | 1,028 | LocalScore ↗ | 512+128 | 1,429 ms | llamafile 0.9.3 |
| Q8_0 | 38.3 | 1,600 | LocalScore ↗ | 512+128 | 874 ms | llamafile 0.9.2 |
| Q4_K - Medium | 36.7 | 1,050 | LocalScore ↗ | 512+128 | 1,366 ms | llamafile 0.9.2 |
| Q5_K - Medium | 34.9 | 1,157 | LocalScore ↗ | 512+128 | 1,266 ms | llamafile 0.9.2 |
| Q6_K | 33.1 | 1,016 | LocalScore ↗ | 512+128 | 1,399 ms | llamafile 0.9.2 |
| Q8_0 | 29.5 | 1,389 | LocalScore ↗ | 512+128 | 1,110 ms | llamafile 0.9.2 |
| F16 | 17.7 | 1,243 | LocalScore ↗ | 512+128 | 1,048 ms | llamafile 0.9.2 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
What else can AMD Radeon RX 6900 XT run? →
Includes results from LocalScore, a Mozilla Builders project by cjpais, mirrored in the cn0303/local-llm-speed-benchmarks dataset. LocalScore runs use llamafile, which is built on llama.cpp.