Issue 25669 - OOo takes too much CPU power
Summary: OOo takes too much CPU power
Status: CONFIRMED
Alias: None
Product: Calc
Classification: Application
Component: code (show other issues)
Version: OOo 1.1.1b
Hardware: All All
: P3 Trivial (vote)
Target Milestone: ---
Assignee: AOO issues mailing list
QA Contact:
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2004-02-19 09:23 UTC by tuharsky
Modified: 2013-08-07 15:12 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description tuharsky 2004-02-19 09:23:20 UTC
I had to try to open one very big (dumb) .XLS document and since OOo1.1 under
WXP crashed, I opened it with OOo111b under Debian. Opening took 10 minutes,
however, and half of the time I was unable to use any other app (Mozilla,MC,..).
I'm not pleased to use Celeron 2.4 but I have it on desk, and with 256MB RAM it
is not completely crap, usually I can work on it with GNOME 1.4 well. I use
well-tuned 2.6.3 kernel with preemtibility enabled, so I was surprised that OOo
blocked out everything else.

Even after saving the file in OOo format and opening it, any operation utilizes
the CPU in unwanted manner.

I noticed a difference between opening the file and doing something with the
file. Opening -the power was used mostly (over 95%) by system and 5% by user,
and nearly nothing was possible to do.
When making operations on document (select or so), the CPU was used over 65% by
user, and 35% by system, and I was able to do other things, limited but able. 

Maybe the opening needs the system to do some hardware (memory? swap?) tasks, so
the system utilization of CPU. The swap was used by 150M, and maybe Celeron is
too weak to solve I/O and user-apps simultaneously. Thus it may not be fully
fault of OOo. I'll try with AMD processors and let You know.

It makes only smal difference for me however. I consider using OOo on larger
(20+, 50+,..) thin-client environment and now I'm afraid that even with
powerfull enough server machine and the excellent 2.6 class kernel, single OOo
user could block out all other users if he opens too complex or too large file.
If it is possible to make some parts of OOo code more responsive..

The dumb document can be found at
http://www.education.gov.sk/fin/norm/vypocet_a_data_RS_po_TK_V14_WEB.rar
Comment 1 jack.warchold 2004-02-25 15:04:52 UTC
i think this is more an spreadsheet or framework issue
changed component and reassigned to the defaultowner of this component.
Comment 2 frank 2004-03-01 15:36:25 UTC
Hi Sascha,

could you have a look on this please ?

Direct this Issue into the correct direction if you're not responsible.

Thanks.
Frank
Comment 3 sascha.ballach 2004-03-11 13:25:45 UTC
Target?
Comment 4 frank 2004-03-11 13:29:00 UTC
targeted
Comment 5 frank 2004-03-11 13:30:27 UTC
and prio
Comment 6 frank 2004-05-04 15:37:57 UTC
Due to limited time ressources I re-Target this one to OOo Later
Comment 7 tuharsky 2004-10-27 09:06:30 UTC
I have promised that I will test the behavior with AMD system.

I thought then that I'll get an AMD system soon after the post, it was however
designed for somebody else.

Now, in some 1-2 weeks I could possibly finally have the AMD system. So the test
could be possible in near future.
Comment 8 tuharsky 2004-10-27 09:10:30 UTC
The link is not functional, half year after report.

It is just a huge dumb file; packed 5MB, unpacked 40MB XLS. Luckily, I have
found it locally on my computer, but don't know, how to post it (probably it'll
be too large).

Maybe I'll create smaller one. Some 1MB sxc should be enough for testing I think.
Comment 9 tuharsky 2005-12-07 14:34:11 UTC
Well, today I have tested with Athlon64, 512MB RAM, Debian Sarge, OpenOffice.org
2.0, GNOME.

The 40M XLS file is opened in less than 2 minutes (haven't used stopwatch), and
responsiveness of PC is alright. I assume, that the problem with my previous
machine has been caused by extraordinary memory consupmtion of OOo opening the
huge file, so that inevitable _swapping_ overloaded the bus and CPU. The
document itself has consumed about 155M of memory!

If we take in account the not very starring multitasking performance of Celeron,
the result (PC completely unresponsible) shouldn't surprise anyone.

I think Issue could be closed for now. Memory consumption is quite different issue.

Said that, the document's memory consumption may need some future improvements.
Comment 10 Rob Weir 2013-07-30 02:16:29 UTC
Reset assignee on issues not touched by assignee in more than 2000 days.