Llama 2 7B on RTX 5080
192best gen tok/s
9,488best prompt tok/s
640max context tested
4results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_0 | 192 | 8,055 | llama.cpp (Vulkan) ↗ | 512+128 | — | llama.cpp Vulkan @f6b533d — coopmat2 |
| Q4_0 | 185 | 7,445 | llama.cpp (Vulkan) ↗ | 512+128 | — | llama.cpp Vulkan @f6b533d — coopmat2 |
| Q4_0 | 185 | 9,488 | llama.cpp (CUDA) ↗ | 512+128 | — | llama.cpp CUDA @8a4280c |
| Q4_0 | 182 | 8,297 | llama.cpp (CUDA) ↗ | 512+128 | — | llama.cpp CUDA @8a4280c |
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.