Machine
Quadro K620 · Ryzen 7 PRO 3700 #e6767809
17 runs of 4 models since Nov 16, 2026. Last run Nov 21, 2026.
17
runs
4
models
5
configs
4
configs run 2+ times
Hardware
- GPU
- Quadro K620 2 GB
- CPU
- AMD Ryzen 7 PRO 3700 8-Core Processor 8 cores
- RAM
- 30 GB
- OS
- Ubuntu 26.04.1 LTS
- Tools
- llama.cpp 17
- Client
- 0.4.99
Best config per model
sort
Runs with the same model, quant, tool, context, thinking and KV cache count as one config. Numbers are the median of its runs, and each dot is one run on the 0–100 quality scale. ★ marks the best config with at least 2 rated runs.
configrunsqualityeach runspeed
r5060
5 runs · 1 config
★ bestllama.cpp64k ctxthinking on
×5 runs
84.672.4–87.2
varies1 slow run
39.8 tok/s
17.9–43.7
r3060
9 runs · 2 configs
★ bestllama.cpp64k ctxthinking on
×7 runs
77.070.9–87.7
varies1 slow run3 fast runs
31.8 tok/s
18.5–100
ox253
2 runs · 1 config
★ bestllama.cpp64k ctxthinking on
×2 runs
64.656.1–73.1
varies
37.7 tok/s
18.4–56.9
frigo2
1 run · 1 config
Compare with another machine
Run history
Nov 21, 2026
23:42r3060llama.cpp31.8 tok/s87.723:09ox253llama.cpp18.4 tok/s73.119:53r3060llama.cpp28.7 tok/s86.113:15ox253llama.cpp56.9 tok/s56.1Nov 20, 2026
18:56r5060×2llama.cpp39.0 tok/s83.7Nov 19, 2026
18:50r5060llama.cpp17.9 tok/s72.4Nov 17, 2026
23:39r3060×3llama.cpp24.2 tok/s73.3