Llama 2 7B › AMD Instinct MI300X
Llama 2 7B on the AMD Instinct MI300X
Best measured: 233 tokens/s generation and 11,476 tokens/s prompt processing (Q4_0, ROCm, FA off).
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_0 | ROCm FA off | 233 | 11,476 | tg128 | ee3a9fc @yeahdongcn | llama.cpp ROCm scoreboard (#15021) |
| Q4_0 | ROCm FA on | 219 | 11,946 | tg128 | ee3a9fc @yeahdongcn | llama.cpp ROCm scoreboard (#15021) |
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 AMD Instinct MI300X · Llama 2 7B on other hardware