Behold, the jankiest setup ever (Tesla P40 + Delta server fan)

gpu4.pics
via T-VIRUS999
NVIDIA Tesla P40
1× NVIDIA Tesla P40 24GB · 24 GB total VRAM
24 GB VRAM NVIDIA 250 W TDP
running Qwen3 32B· Q4
5.9
tokens / sec
Generation
5.9tok/s
higher is better
Prompt / prefill
tok/s
higher is better
Time to first token
lower is better
Efficiency
0.01 tok/s/W
Electricity cost
$8/Mtok
Power draw
1000 W
gpu4.pics/r/curated-1nxjhnj

All benchmarks on this rig

ModelQuantGenPromptTTFTPowerPerf/W$/MtokScore
Qwen3 32BQ45.91,000 W0.01 tok/s/W$8/Mtok