Here is a csv file based on markus script (for some reason plotting doesnt work for me) You can see the resuls are pretty good basically unitl batch size 1000 the hot cache is always faster, then it is just like the noncached version in trunk
--- ;50;500;950;1000;1050;2000;5000;10000 zeitgeist_C;0,002746;0,024526;0,046756;0,048572;0,051662;0,098936;0,245838;0,518534 zeitgeist_H;0,0023;0,023295;0,045963;0,04813;0,052496;0,097886;0,252786;0,493975 add_cache_for_get_events_C;0,003201;0,029415;0,054442;0,060075;0,05168;0,098671;0,24538;0,509424 add_cache_for_get_events_H;0,000868;0,00736;0,014856;0,013968;0,051253;0,09653;0,252904;0,485511 ---
« Back to merge proposal
Here is a csv file based on markus script (for some reason plotting doesnt work for me)
You can see the resuls are pretty good
basically unitl batch size 1000 the hot cache is always faster, then it is just like the noncached version in trunk
--- 950;1000; 1050;2000; 5000;10000 C;0,002746; 0,024526; 0,046756; 0,048572; 0,051662; 0,098936; 0,245838; 0,518534 H;0,0023; 0,023295; 0,045963; 0,04813; 0,052496; 0,097886; 0,252786; 0,493975 for_get_ events_ C;0,003201; 0,029415; 0,054442; 0,060075; 0,05168; 0,098671; 0,24538; 0,509424 for_get_ events_ H;0,000868; 0,00736; 0,014856; 0,013968; 0,051253; 0,09653; 0,252904; 0,485511
;50;500;
zeitgeist_
zeitgeist_
add_cache_
add_cache_
---