Llama 2 7B on the NVIDIA Tesla M40
Best measured: 41.7 tokens/s generation and 93 tokens/s prompt processing (Q4_0, Vulkan, FA on).
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_0 | Vulkan FA on | 41.7 | 93 | tg128 | b8372ee | llama.cpp Vulkan scoreboard (#10879) |
| Q4_0 | CUDA FA on | 40.0 | 290 | tg128 | 97d5117 @Hedede | llama.cpp CUDA scoreboard (#15013) |
| Q4_0 | Vulkan FA off | 39.4 | 92 | tg128 | b8372ee | llama.cpp Vulkan scoreboard (#10879) |
| Q4_0 | CUDA FA off | 38.0 | 283 | tg128 | 97d5117 @Hedede | llama.cpp CUDA scoreboard (#15013) |
tg128 = generating 128 tokens; pp512 = reading a 512-token prompt (llama-bench). LocalScore averages several prompt and output lengths, so its numbers run lower than tg128.
Rent one by the hour
Rent a GPU on RunPodAffiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.
All results for the NVIDIA Tesla M40 · Llama 2 7B on other hardware