Llama 2 7B on the NVIDIA RTX 3090 Ti

Best measured: 172 tokens/s generation and 6,924 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	172	6,924	tg128	9c35706 @slaren	llama.cpp CUDA scoreboard (#15013)
Q4_0	CUDA FA off	171	6,567	tg128	9c35706 @slaren	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.

Check prices

Amazon Newegg

As an Amazon Associate I earn from qualifying purchases.

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 RTX 3090 Ti · Llama 2 7B on other hardware