Llama 2 7B › NVIDIA Quadro RTX 4000
Llama 2 7B on the NVIDIA Quadro RTX 4000
Best measured: 67.6 tokens/s generation and 1,663 tokens/s prompt processing (Q4_0, CUDA, FA on).
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_0 | CUDA FA on | 67.6 | 1,663 | tg128 | 7d77f07 @Hedede | llama.cpp CUDA scoreboard (#15013) |
| Q4_0 | CUDA FA off | 65.6 | 1,537 | tg128 | 7d77f07 @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.
No card yet?
Try it on a rented GPU (RunPod)Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.
All results for the NVIDIA Quadro RTX 4000 · Llama 2 7B on other hardware