Llama 2 7B on Tesla P100
66.5best gen tok/s
787best 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 | 66.5 | 686 | llama.cpp (Vulkan) ↗ | 512+128 | — | llama.cpp Vulkan @eec1e33 |
| Q4_0 | 63.2 | 678 | llama.cpp (Vulkan) ↗ | 512+128 | — | llama.cpp Vulkan @eec1e33 |
| Q4_0 | 62.0 | 787 | llama.cpp (CUDA) ↗ | 512+128 | — | llama.cpp CUDA @b8372ee |
| Q4_0 | 58.4 | 761 | llama.cpp (CUDA) ↗ | 512+128 | — | llama.cpp CUDA @b8372ee |
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.