Welcome to MilkyWay@home

Please post app 1.17/1.18/1.19 memory leaks/errors

Message boards : Number crunching : Please post app 1.17/1.18/1.19 memory leaks/errors
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · Next

AuthorMessage
Profile Cori
Avatar

Send message
Joined: 27 Aug 07
Posts: 647
Credit: 27,592,547
RAC: 0
Message 1951 - Posted: 5 Mar 2008, 22:46:09 UTC - in response to Message 1950.  
Last modified: 5 Mar 2008, 22:46:42 UTC

Funny enough the "wrong" 1.19 app (32-bit) was running faster than the "correct" 1.17 app (64-bit).
Same thing here
64bit Fedora7 with 32bit app here
64bit Fedora7 with 64bit app here


Ok guys/gals... i can offer a quick fix for that one... Travis ? you don't mind ?
I'm compiling a proper x64 bit app for linux now ;)

Fine, fine... Go ahead! :-)))
Lovely greetings, Cori
ID: 1951 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1953 - Posted: 5 Mar 2008, 22:47:20 UTC - in response to Message 1944.  

No those are 32bit machines apparently trying to load a 64bit app of 1.19.

Can't try anymore unless you reset the quotas for hosts 7406 and 3296.Says I have reached max and defering communications for 7 hrs



ahhh ok. we're going to need to recompile the 32 bit linux then. However, it looks like the error you got was more of a linking issue than a problem with the binary...



Ok I am going to try to detach 1 of those hosts and reattach...that should take care of the quota thing and see if it links me to the 32bit app instead of the 64 bit...will let you know.


Nope the server picked up it was the same host and won't d/l new work because I exceeded my daily quota of 286(reduced due to 18 hrs worth of errors)Have to wait 6 1/2 hours unless quota's are reset :(
ID: 1953 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Odysseus

Send message
Joined: 10 Nov 07
Posts: 96
Credit: 29,931,027
RAC: 0
Message 1954 - Posted: 5 Mar 2008, 22:50:47 UTC

My G4 Macs that were having problems with v1.15 and vv1.17–1.18 respectively are now happily crunching with v1.19. Good work, team!

ID: 1954 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Stick

Send message
Joined: 8 Oct 07
Posts: 52
Credit: 5,630,511
RAC: 223
Message 1957 - Posted: 5 Mar 2008, 23:06:19 UTC - in response to Message 1930.  
Last modified: 5 Mar 2008, 23:08:20 UTC

EDIT: I watched my second unit a little more carefully. The first 10% took about 9 minutes and the last 90% took about 2 minutes. In other words, the "Progress" meter works OK - it's just not linear.


yeah crunch3r had it right about this. the first 10% is doing an integral calculation (which is pretty computationally intensive), and the last 90% is comparing the star values to this integral calculation. there's really no good way to calculate the progress in a linear way, since on some architectures the integral takes a lot less time than on others (probably due to optimizations).


I didn't mean to imply that being non-linear was a problem. As long as we know what to expect, we can deal with it.
ID: 1957 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crunch3r
Volunteer developer
Avatar

Send message
Joined: 17 Feb 08
Posts: 363
Credit: 258,227,990
RAC: 0
Message 1958 - Posted: 5 Mar 2008, 23:10:33 UTC - in response to Message 1951.  

Funny enough the "wrong" 1.19 app (32-bit) was running faster than the "correct" 1.17 app (64-bit).
Same thing here
64bit Fedora7 with 32bit app here
64bit Fedora7 with 64bit app here


Ok guys/gals... i can offer a quick fix for that one... Travis ? you don't mind ?
I'm compiling a proper x64 bit app for linux now ;)

Fine, fine... Go ahead! :-)))

Well, i know what wen't wrong ... though taken a look at the time now...well not today any more BUT tomorrow that will be sure !

However, we've all had some success after all today ! all windows apps are working properly now !

So let's keep on focusin on liunx now... i'll compile some new apps tomorrow/today ...depends on you timezone...

Jist don't loose your faith, today we accompilshed at least one goal... windows apps are working properly and we will fix that linux issue very soon that i'm pretty sure of !

Together we'll get it working !




Join Support science! Joinc Team BOINC United now!
ID: 1958 · Rating: -1 · rate: Rate + / Rate - Report as offensive     Reply Quote
Hefto99

Send message
Joined: 29 Dec 07
Posts: 9
Credit: 100,140,182
RAC: 0
Message 1959 - Posted: 5 Mar 2008, 23:13:34 UTC

1.19 works well so far on my 32-bit Linux box :-)
ID: 1959 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Cori
Avatar

Send message
Joined: 27 Aug 07
Posts: 647
Credit: 27,592,547
RAC: 0
Message 1960 - Posted: 5 Mar 2008, 23:13:46 UTC
Last modified: 5 Mar 2008, 23:14:52 UTC

Ok, so I can go to bed now and leave my compi crunching under Win XP.
And tomorrow I'll boot into Linux 64-bit again. *grin* Looking forward to the shiny new Linux 64-bit app, Crunch3r! :-)

Nighty night everyone! *YAWN*
Lovely greetings, Cori
ID: 1960 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Temujin

Send message
Joined: 12 Oct 07
Posts: 77
Credit: 404,471,187
RAC: 0
Message 1961 - Posted: 5 Mar 2008, 23:18:03 UTC

Good work Team Milkyway
ID: 1961 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 1962 - Posted: 5 Mar 2008, 23:20:54 UTC - in response to Message 1945.  

1.19, but I'm back to getting the units that freeze again (seems the same prob I had before). 2 stopped at 2 secs, the next at 4:15, out of the 4 that have ran. Another at 4:54.

I had no probs with 1.18, I ran it since it was put out.


Anything on this?

whatever was changed between 1.18 and 1.19 seems to have brought back this problem for me. So looking into that may fix this problem.


Progress meter counts slow until the last min then it rapidly counts up to 100%. Also the 1.18 took around 9.5 min now the 1.19 take 10.5 min.
ID: 1962 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crunch3r
Volunteer developer
Avatar

Send message
Joined: 17 Feb 08
Posts: 363
Credit: 258,227,990
RAC: 0
Message 1964 - Posted: 5 Mar 2008, 23:30:57 UTC - in response to Message 1962.  


whatever was changed between 1.18 and 1.19 seems to have brought back this problem for me. So looking into that may fix this problem.


Progress meter counts slow until the last min then it rapidly counts up to 100%. Also the 1.18 took around 9.5 min now the 1.19 take 10.5 min.



Well we're all well arare of those issue and tomorrow we will fix it ...
Thanks.




Join Support science! Joinc Team BOINC United now!
ID: 1964 · Rating: -1 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1968 - Posted: 5 Mar 2008, 23:54:43 UTC - in response to Message 1962.  

1.19, but I'm back to getting the units that freeze again (seems the same prob I had before). 2 stopped at 2 secs, the next at 4:15, out of the 4 that have ran. Another at 4:54.

I had no probs with 1.18, I ran it since it was put out.


Anything on this?

whatever was changed between 1.18 and 1.19 seems to have brought back this problem for me. So looking into that may fix this problem.


Progress meter counts slow until the last min then it rapidly counts up to 100%. Also the 1.18 took around 9.5 min now the 1.19 take 10.5 min.


1.18 wasn't doing checkpointing or updating the progress at all, which probably accounts for the 1.19 being slower.

The way i did the progress was (at least on the machine i tested on), the integral calculation took about 10% of the time as the likelihood calculation, so i took the progress of the first part, multiplied it by 0.1 and then took the progress of the 2nd part, multiplied it by 0.9 and added 0.1.

If the numbers are a bit off, i could change it to 0.5 and 0.5 or something like that.
ID: 1968 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1969 - Posted: 5 Mar 2008, 23:56:02 UTC - in response to Message 1954.  

My G4 Macs that were having problems with v1.15 and vv1.17–1.18 respectively are now happily crunching with v1.19. Good work, team!


unfortunatley 1.19 is the same as 1.13 for ppc macs, but then again, macs weren't having much problems with 1.13, except for the small memory leak. i'll be working on updating the ppc binary tomorrow and hopefully it'll be the new (memory leak free) version along with everything else.
ID: 1969 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1971 - Posted: 6 Mar 2008, 0:04:43 UTC
Last modified: 6 Mar 2008, 0:35:45 UTC

The Milkway team has worked very hard today and looks like the problems with windows is about solved.Congratulations.Thanks for the communication as well.

I am assuming the Linux app will be changed tommorrow so the 2 linux machines(identicals) I have that previously worked work again?It has been 2 days of errors now.Is it due to different hardware architecture that is causing some 32bit to work ok and others not?(all exact same O/S from same burned CD) The 64 bit machines never had a problem.
ID: 1971 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Webmaster Yoda
Avatar

Send message
Joined: 21 Dec 07
Posts: 69
Credit: 7,048,412
RAC: 0
Message 1972 - Posted: 6 Mar 2008, 2:34:02 UTC

Seems to be working better on Windows (32bit) now than it does on 64 bit Linux. Quite a difference. I understand new Linux apps (including 64 bit) are coming but just thought I'd share.

I am running Windows SBS 2003 in a virtual box on my quad (at 3.45GHz). Work units under Linux are taking approx. 7 minutes. The Windows app running in the virtual box is taking only 5.5 minutes!
Join the #1 Aussie Alliance on MilkyWay!
ID: 1972 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 2 Jan 08
Posts: 122
Credit: 69,479,776
RAC: 1,437
Message 1973 - Posted: 6 Mar 2008, 2:56:07 UTC - in response to Message 1885.  

Have gotten this error on the last 60 WU's as soon as they start, both 1.17 and 1.18. Using Linux Fedora Core 6, none have been successful,

<core_client_version>5.10.21</core_client_version>
<![CDATA[
<message>
process exited with code 22 (0x16, -234)
</message>
<stderr_txt>
2008-03-05 06:10:03 [Milkyway@home] Starting gs_281_1204711021_258_0
execv: No such file or directory
2008-03-05 06:10:03 [Milkyway@home] [error] Process creation (../../projects/milkyway.cs.rpi.edu_milkyway/astronomy_1.15_i686-pc-linux-gnu) failed: Error -1, errno=8

</stderr_txt>

EDIT:: Have just checked and you better make that 80 crashed WU's as all 20 that were on my computer have errored out as well.


Well done all now working with 1.19 on Linux.
Also well done on the Windows 1.17 application, very nice speed up, it is now as fast as Linux on the same hardware instead of half the speed.
ID: 1973 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Dave Przybylo
Avatar

Send message
Joined: 5 Feb 08
Posts: 236
Credit: 49,648
RAC: 0
Message 1974 - Posted: 6 Mar 2008, 3:02:16 UTC

The windows builds got some VERY high level optimization thanks to Cruncher. They should be just as fast if not much faster than any other builds
Dave Przybylo
MilkyWay@home Developer
Department of Computer Science
Rensselaer Polytechnic Institute
ID: 1974 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ebahapo
Avatar

Send message
Joined: 6 Sep 07
Posts: 66
Credit: 636,861
RAC: 0
Message 1976 - Posted: 6 Mar 2008, 17:37:39 UTC - in response to Message 1897.  

1.19 is already listed on the apps page :) so it should get to you after your last set of wus is done.

Yet it seems to be a tad slower than both 1.17 and 1.18...

Thanks.
ID: 1976 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1977 - Posted: 6 Mar 2008, 17:41:19 UTC

A detach/reattach worked on the 2 linux 32 bit hosts...they are back to crunching mw now....do we still have a server problem? It has crashed once a day for 4 days now...
ID: 1977 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[PST]Howard
Avatar

Send message
Joined: 31 Aug 07
Posts: 21
Credit: 21,004,179
RAC: 0
Message 1985 - Posted: 6 Mar 2008, 20:33:02 UTC

Getting Output file absent message on 32bit Linux, all wus errored on 2 boxes

Detach/reattach, didn't work, still getting Output file absent


ID: 1985 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1987 - Posted: 6 Mar 2008, 20:51:44 UTC - in response to Message 1985.  

Getting Output file absent message on 32bit Linux, all wus errored on 2 boxes

Detach/reattach, didn't work, still getting Output file absent




I went to synaptic package manager and d/l libstd 5.0+++ as well as detach/reattach to get mine to work as the stderr of my work showed that library as being dependent with the new app....so maybe thats why mine worked
ID: 1987 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Number crunching : Please post app 1.17/1.18/1.19 memory leaks/errors

©2024 Astroinformatics Group