Skip to content

Instantly share code, notes, and snippets.

@adamo1139

adamo1139/llama-benchy-8x-3090ti-config1.md

Created May 2, 2026 15:34
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save adamo1139/372de9f6cdfd38155d0dbea0b2bb3878 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save adamo1139/372de9f6cdfd38155d0dbea0b2bb3878 to your computer and use it in GitHub Desktop.
Download ZIP
llama-benchy-8x-3090ti-config1
Raw
llama-benchy-8x-3090ti-config1.md
model test t/s (total) t/s (req) peak t/s peak t/s (req) ttfr (ms) est_ppt (ms) e2e_ttft (ms)
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 (c1) 1710.77 ± 23.16 1710.77 ± 23.16 1412.71 ± 16.35 1197.34 ± 16.35 1412.71 ± 16.35
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 (c1) 17.62 ± 0.20 17.62 ± 0.20 19.67 ± 0.47 19.67 ± 0.47
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 (c2) 1610.24 ± 54.20 904.96 ± 43.48 2483.65 ± 108.10 2268.28 ± 108.10 2483.65 ± 108.10
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 (c2) 32.86 ± 1.99 17.21 ± 0.09 38.00 ± 0.00 19.00 ± 0.00
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d4096 (c1) 3720.29 ± 64.43 3720.29 ± 64.43 1867.35 ± 28.89 1651.99 ± 28.89 1867.35 ± 28.89
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d4096 (c1) 17.36 ± 0.03 17.36 ± 0.03 20.00 ± 0.00 20.00 ± 0.00
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d4096 (c2) 2335.91 ± 30.35 1866.22 ± 648.68 3960.08 ± 1302.17 3744.71 ± 1302.17 3960.08 ± 1302.17
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d4096 (c2) 30.28 ± 1.82 16.58 ± 0.64 38.67 ± 0.94 19.50 ± 0.50
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d8192 (c1) 4737.77 ± 219.05 4737.77 ± 219.05 2381.47 ± 102.65 2166.10 ± 102.65 2381.47 ± 102.65
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d8192 (c1) 17.31 ± 0.00 17.31 ± 0.00 19.67 ± 0.47 19.67 ± 0.47
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d8192 (c2) 2392.64 ± 21.22 2316.21 ± 1091.25 5892.54 ± 2668.71 5677.18 ± 2668.71 5892.54 ± 2668.71
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d8192 (c2) 25.63 ± 0.74 15.68 ± 1.45 38.00 ± 0.00 19.00 ± 0.00
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d16384 (c1) 5463.03 ± 101.19 5463.03 ± 101.19 3590.46 ± 61.79 3375.10 ± 61.79 3590.46 ± 61.79
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d16384 (c1) 17.24 ± 0.10 17.24 ± 0.10 19.33 ± 0.47 19.33 ± 0.47
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d16384 (c2) 2229.33 ± 6.83 2764.97 ± 1636.59 10470.67 ± 6065.69 10255.30 ± 6065.69 10470.67 ± 6065.69
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d16384 (c2) 22.13 ± 2.20 14.33 ± 2.91 37.33 ± 0.94 18.83 ± 0.37
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d32768 (c1) 5452.14 ± 55.02 5452.14 ± 55.02 6601.77 ± 64.56 6386.40 ± 64.56 6601.77 ± 64.56
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d32768 (c1) 17.37 ± 0.13 17.37 ± 0.13 19.33 ± 0.47 19.33 ± 0.47
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d32768 (c2) 1815.27 ± 4.22 2849.59 ± 1936.95 22924.32 ± 15434.87 22708.96 ± 15434.87 22924.32 ± 15434.87
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d32768 (c2) 16.46 ± 0.54 12.79 ± 4.44 38.33 ± 0.47 19.17 ± 0.37
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d65536 (c1) 4613.71 ± 11.76 4613.71 ± 11.76 14863.97 ± 37.29 14648.60 ± 37.29 14863.97 ± 37.29
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d65536 (c1) 17.51 ± 0.10 17.51 ± 0.10 20.00 ± 0.00 20.00 ± 0.00
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d65536 (c2) 1287.10 ± 3.41 2457.44 ± 1812.84 60531.34 ± 44487.20 60315.98 ± 44487.20 60531.34 ± 44487.20
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d65536 (c2) 8.26 ± 0.08 9.65 ± 5.43 33.33 ± 0.94 17.33 ± 1.25
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d131072 (c1) 3212.19 ± 8.61 3212.19 ± 8.61 41657.76 ± 111.09 41442.39 ± 111.09 41657.76 ± 111.09
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d131072 (c1) 16.31 ± 0.18 16.31 ± 0.18 19.33 ± 0.94 19.33 ± 0.94
cyankiwi/MiniMax-M2.7-AWQ-4bit pp2048 @ d131072 (c2) 2749.55 ± 420.35 2520.75 ± 957.42 65998.35 ± 34157.33 65782.98 ± 34157.33 65998.35 ± 34157.33
cyankiwi/MiniMax-M2.7-AWQ-4bit tg512 @ d131072 (c2) 12.91 ± 3.58 16.43 ± 0.02 19.00 ± 0.00 18.67 ± 0.47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment