Llama 2 7B › AMD Radeon RX 9070 XT
Llama 2 7B on the AMD Radeon RX 9070 XT
Best measured: 137 tokens/s generation and 5,036 tokens/s prompt processing (Q4_0, Vulkan, FA off).
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_0 | Vulkan FA off | 137 | 5,036 | tg128 | e9fd8dc | llama.cpp Vulkan scoreboard (#10879) |
| Q4_0 | Vulkan FA on | 132 | 5,048 | tg128 | e9fd8dc | llama.cpp Vulkan scoreboard (#10879) |
| Q4_0 | ROCm FA off | 101 | 5,055 | tg128 | 583cb83 @Hadrianneue | llama.cpp ROCm scoreboard (#15021) |
| Q4_0 | ROCm FA on | 97.3 | 4,904 | tg128 | 583cb83 @Hadrianneue | 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.
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 AMD Radeon RX 9070 XT · Llama 2 7B on other hardware