June 13th, 2003
botttlenecked scripts
nontechnical post: if you want to laugh at pictures of me photoshopped onto various bodies, go here.
technical post regarding audiomatch and clearing of script bottlenecks:
Man, I need to get more work done on tabulas. I swear I'll be done with Audiomatch soon ... but I swear that I cannot *not* work on audiomatch. the bottlenecks within the script were *killing* the server.
so i spent some time analyzing the script that receives the data and inputs it into mysql. after making a few hypotheses, i input a bunch of benchmarking functions to time the execution time of certain portions.
in any case, i did fix up a _HUGE_ bottleneck and i'm pretty sure how to improve performance even more. i reduced the average runtime of a script 40% (wow!).
i've put up a page that has the data analyses as a side-by-side comparison . go here.
i should note that with this data, it's been 'corrected.' basically when i was running the old script, te bottlenecks would cause the server to slow down significantly, so the data would get slightly skewed.
the percentages you see are 'standardized' to the image runtimes. since the image runtimes haven't really been changed, they should match. everything else is referencing the image runtimes. i have an inkling that the greater data set for the right column as well as the increased stability of the server is contributing to the reduced runtime.
cool!
technical post regarding audiomatch and clearing of script bottlenecks:
Man, I need to get more work done on tabulas. I swear I'll be done with Audiomatch soon ... but I swear that I cannot *not* work on audiomatch. the bottlenecks within the script were *killing* the server.
so i spent some time analyzing the script that receives the data and inputs it into mysql. after making a few hypotheses, i input a bunch of benchmarking functions to time the execution time of certain portions.
in any case, i did fix up a _HUGE_ bottleneck and i'm pretty sure how to improve performance even more. i reduced the average runtime of a script 40% (wow!).
i've put up a page that has the data analyses as a side-by-side comparison . go here.
i should note that with this data, it's been 'corrected.' basically when i was running the old script, te bottlenecks would cause the server to slow down significantly, so the data would get slightly skewed.
the percentages you see are 'standardized' to the image runtimes. since the image runtimes haven't really been changed, they should match. everything else is referencing the image runtimes. i have an inkling that the greater data set for the right column as well as the increased stability of the server is contributing to the reduced runtime.
cool!
SuperSunJ
