Message345654
Raymond, please read my very recent comment one above yours. A (overall) quadratic-time algorithm (O(A**2) where A is the number of arenas) in obmalloc.c is (to my eyes) probably the _primary_ cause of the sloth here. That's been fixed for 3.8, but I don't have enough RAM even to run Terry's test code to confirm it.
I can confirm that there's no quadratic-time behavior anymore deleting large sets of strings, but only until I run out of RAM. Neither is the behavior linear, but it's much closer to linear than quadratic now.
If someone with more RAM can confirm this for larger sets, then fine by me if this is closed again. |
|
Date |
User |
Action |
Args |
2019-06-15 02:12:43 | tim.peters | set | recipients:
+ tim.peters, lemburg, twouters, rhettinger, terry.reedy, vstinner, luis@luispedro.org, methane, serhiy.storchaka |
2019-06-15 02:12:43 | tim.peters | set | messageid: <1560564763.29.0.882089488345.issue32846@roundup.psfhosted.org> |
2019-06-15 02:12:43 | tim.peters | link | issue32846 messages |
2019-06-15 02:12:43 | tim.peters | create | |
|