Regressions galore?

Posted Feb 9, 2012 22:58 UTC (Thu) by blujay (guest, #39961)
Parent article: What happened to disk performance in 2.6.39

Is it just me or is Linux suffering more needless regressions than ever before? I'm getting the impression that some devs are writing code that's too smart for its own good. I honestly think that 10 years ago my Debian system running the then-current kernel had better interactive performance, especially while swapping, than 2.6 or 3.0 have today. Yeah, I know a bunch of interactivity-related patches have been added during this time--but in the end, I remember using systems with less than half as much memory as I have now and swapping taking less time and apps blocking less. I don't remember my cursor movement lagging back then--now it happens whenever swapping happens. Is the kernel's desktop suitability on the decline? :(

to post comments

Regressions galore?

Posted Feb 10, 2012 12:59 UTC (Fri) by jospoortvliet (guest, #33164) [Link]

Kernel 3.2 is certainly noticeably better than 3.1 and 3.0 in the interactivity regard, esp in low memory/swap situations and on high IO. But you're right in that older linux versions had better behavior in many situations - worse in others, however. Like how often it happened that one high-cpu task would hog your system and make your mouse cursor or even music skip. That rarely happens these days, with the exception of cases with high swapping.

Regressions galore?

Posted Apr 5, 2012 15:55 UTC (Thu) by Andrew_Cady (guest, #83993) [Link]

Maybe your memory usage has grown faster than the speed of your swap disk.