gpu4.pics
data via LocalScore
NVIDIA RTX 3090 Ti
24 GB VRAM NVIDIA 450 W TDP
running Meta Llama 3.1 8B Instruct· Q4_K_M
110
tokens / sec
1,113 · Excellent #9 of 370
Generation
110tok/s
higher is better
Prompt / prefill
4,024tok/s
higher is better
Time to first token
320 ms
lower is better
Composite score
1,113
Model size
8B
Runtime
gpu4.pics/r/ls-nvidia-rtx-3090-ti-meta-llama-31-8b-instruct-q4-k-m

All benchmarks on this rig

ModelQuantGenPromptTTFTPowerPerf/W$/MtokScore
Meta Llama 3.1 8B InstructQ4_K_M1104,024320 ms1,113

How it stacks up — Meta Llama 3.1 8B Instruct

NVIDIA RTX 3090 Ti
110
NVIDIA H100
120
NVIDIA A100 80GB
110
NVIDIA RTX PRO 6000 Blackwell Workstation Edition
103
NVIDIA RTX 3090
103
NVIDIA RTX A6000
90.5