Skip to contentSkip to conversation

2.5.38{-mm1} contest results

BENCHMARK Con Kolivas 1 participant 1 message Sep 22, 2002

lore
Share & tools
lore
#0 lore raw

Here are the latest contest results including 2.5.38 and 2.5.38-mm1

NoLoad:

Kernel                  Time            CPU
2.4.19                  66.56           99%
2.4.19-ck7              65.77           99%
2.4.19-ac4              66.31           99%
2.5.36                  67.45           99%
2.5.36-mm1              67.39           99%
2.5.37                  67.25           99%
2.5.38                  68.25           99%
2.5.38-mm1              67.17           99%

Process Load:

Kernel                  Time            CPU
2.4.19                  81.29           80%
2.4.19-ck7              70.14           93%
2.4.19-ac4              71.10           92%
2.5.36                  71.04           94%
2.5.36-mm1              70.68           95%
2.5.37                  70.51           95%
2.5.38                  71.60           95%
2.5.38-mm1              70.49           95%

IO Half Load:

Kernel                  Time            CPU
2.4.19                  101.39          69%
2.4.19-ck7              75.96           88%
2.4.19-ac4              97.56           73%
2.5.36                  79.30           91%
2.5.36-mm1              100.05          74%
2.5.37                  77.69           93%
2.5.38                  81.26           90%
2.5.38-mm1              82.52           87%

IO Full Load:

Kernel                  Time            CPU
2.4.19                  170.70          41%
2.4.19-ck7              90.95           74%
2.4.19-ac4              105.53          68%
2.5.36                  197.08          36%
2.5.36-mm1              220.14          33%
2.5.37                  209.75          33%
2.5.38                  170.21          42%
2.5.38-mm1              434.41          16%

Mem Load:

Kernel                  Time            CPU
2.4.19                  93.33           77%
2.4.19-ck7              123.15          57%
2.4.19-ac4              117.09          61%
2.5.36                  121.02          59%
2.5.36-mm1              100.47          73%
2.5.37                  104.75          70%
2.5.38                  104.22          70%
2.5.38-mm1              92.97           77%

As you can see, 2.5.38 is improving in most areas. 2.5.38-mm1 has excellent performance under mem load now, but exhibits significant slow down under full IO load.

To reproduce full IO load, continuously copy /dev/zero in Ram sized chunks to the same disk and conduct a kernel compile on that disk and time the kernel compile.

http://contest.kolivas.net

Comments?
Con.