| Summary: | Slow File OPEN of 477 pages ODT with 282 comments and track changes | ||
|---|---|---|---|
| Product: | LibreOffice | Reporter: | Telesto <telesto> |
| Component: | Writer | Assignee: | Not Assigned <libreoffice-bugs> |
| Status: | NEW --- | ||
| Severity: | major | CC: | aron.budea, buzea.bogdan, ilmari.lauhakangas |
| Priority: | medium | Keywords: | perf |
| Version: | 7.3.0.0 alpha0+ | ||
| Hardware: | All | ||
| OS: | All | ||
| See Also: |
https://bugs.documentfoundation.org/show_bug.cgi?id=148013 https://bugs.documentfoundation.org/show_bug.cgi?id=140443 |
||
| Whiteboard: | target:24.8.0 | ||
| Crash report or crash signature: | Regression By: | ||
| Bug Depends on: | |||
| Bug Blocks: | 83946, 107431, 136524 | ||
|
Description
Telesto
2021-08-31 09:21:49 UTC
Hi, Telesto, not sure if you are aware: in Linux, you may start LO with 'time OOO_EXIT_POST_STARTUP=1 instdir/program/soffice' and it will measure and exit, you don't have to measure manually start time. in Windows, I guess it can be done with power-shell or ptime or timemem utility. So, while what you found is useful, you had better find exact branches of slowdown. I hope a good example is bug 144395. OO 3.3 Linux real 1m26,533s user 1m70,703s sys 0m1,059s 4.3all oldest real 2m37,436s user 2m35,723s sys 0m0,362s 4.1max master real 2m56,234s user 2m47,434s sys 0m0,878s 4.2max master real 3m23,532s user 3m19,044s sys 0m0,634s 4.3max master real 4m08,712s user 4m04,007s sys 0m0,619s 5.2 gtk3 real 3m25,973s user 3m20,107s sys 0m0,594s 6.0 gtk3 real 4m20,577s user 4m15,235s sys 0m1,050s 7.0 gtk3 real 4m25,326s user 4m14,292s sys 0m1,292s 7.3 gtk3 real 3m54,018s user 3m51,589s sys 0m1,241s While LO 4.x was faster, difference is not that big with those Lin numbers. Real difference is compared to OO. Should be checked for content, though. But if track changes are accepted than 7.3+ is: real 0m8,283s user 0m7,024s sys 0m0,544s So this is a duplicate of some track changes perf bug. *** Bug 144533 has been marked as a duplicate of this bug. *** Here are times from different machine with GEN. I don't see it faster in 5.2 (I think that's wrong in duplicate bug which is the same file by the same reporter), only in OO time, so preBibisect. OO 3.3 074,71s user 1,03s system 87% cpu 1:26,25 total 43all oldest 161,90s user 0,73s system 94% cpu 2:51,95 total 41max master 162,83s user 0,83s system 95% cpu 2:50,58 total 50max 182,22s user 0,66s system 93% cpu 3:15,11 total 5.2 174,28s user 0,82s system 92% cpu 3:09,48 total 5.4 200,23s user 1,07s system 92% cpu 3:38,08 total 7.3+ 168,09s user 0,89s system 98% cpu 2:51,27 total OO opened 477 pages and 7.3+ opens 509 pages and only later changes to 477. That shouldn't be reported without finding version where it started and searching first, I remember seeing it. I didn't confirm because I guess it's already reported, should be found, and raised priority. LO sucks here. Similar results here. Let's set to new anyway. Version: 7.5.0.0.alpha0+ (X86_64) / LibreOffice Community Build ID: 9cd0f4c2d25462feba0ffcbd906c199273821243 CPU threads: 8; OS: Linux 6.0; UI render: default; VCL: kf5 (cairo+xcb) Locale: fi-FI (fi_FI.UTF-8); UI: en-US Calc: threaded Noel Grandin committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/a06946271e7139352501e0840256402ac43f4c65 tdf#144208 speedup doc with lots of redline It will be available in 24.8.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback. Noel Grandin committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/1291edd7ed369ee97177cab2910f2396302ebacf tdf#144208 speedup doc with lots of redline(2) It will be available in 24.8.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback. I built with the three patches on Linux. Time with OOO_EXIT_POST_STARTUP=1 real 0m26,429s user 0m26,304s sys 0m0,392s With 24.2 the time is real 2m59,777s user 2m58,274s sys 0m0,910s Time to document/UI responsiveness is greater in reality. Close to 2 minutes even with the patches. Noel Grandin committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/bcc30f482b0833460e38a48c07295a8e86b23478 tdf#144208 speedup doc with lots of redline(3) It will be available in 24.8.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback. |