# Wed Oct 28 02:59:16 PM +03 2026 mode=single server=127.0.0.1:18020
ROW cohort C1 real prompts T=default | e2e=214.81 tok/s | decode(C/meanTPOT)=221.7 | tok/step=3.20 | meanTTFT=87.03 ms
ROW cohort C2 real prompts T=default | e2e=364.15 tok/s | decode(C/meanTPOT)=418.4 | tok/step=3.18 | meanTTFT=114.14 ms
ROW cohort C4 real prompts T=default | e2e=650.59 tok/s | decode(C/meanTPOT)=747.7 | tok/step=3.39 | meanTTFT=146.06 ms
ROW cohort C8 real prompts T=default | e2e=509.42 tok/s | decode(C/meanTPOT)=809.7 | tok/step=3.25 | meanTTFT=1434.55 ms
ROW cohort C1 real prompts T=0 | e2e=224.07 tok/s | decode(C/meanTPOT)=231.5 | tok/step=3.28 | meanTTFT=81.54 ms
ROW cohort C2 real prompts T=0 | e2e=386.70 tok/s | decode(C/meanTPOT)=434.8 | tok/step=3.25 | meanTTFT=116.14 ms
ROW cohort C4 real prompts T=0 | e2e=607.36 tok/s | decode(C/meanTPOT)=736.6 | tok/step=3.24 | meanTTFT=139.00 ms
ROW cohort C8 real prompts T=0 | e2e=527.17 tok/s | decode(C/meanTPOT)=797.6 | tok/step=3.29 | meanTTFT=1511.52 ms
# raw logs in /home/zeka/ai/benchmarks/hyperqwen-5090-2026-10-28/results/single-dflash2/run1
