ap_Graphics_5_05_windows_intelx86 Memory Leak?

Questions and Answers : Windows : ap_Graphics_5_05_windows_intelx86 Memory Leak?
Message board moderation

To post messages, you must log in.

AuthorMessage
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1073875 - Posted: 3 Feb 2011, 19:10:45 UTC

Hiya,

I don't know if this issue has been addressed before, but I seem to be experiencing a "memory leak" problem with the ap_graphics_5_05_windows_intelx86 program. Its the process that is started with the "Show Graphics" option under the Tasks tab in the Boinc Manager.

So 3 questions.

1. Does anyone else experience this problem?

2. Have I caused this myself with wrong settings in the manager?

3. Is it a problem specific to my setup?

I am running Bionc on an Athlon II X3 440, Asus M4A78LT-M LE motherboard, 1GB DDR3 RAM, Windows Vista Home Premium Service Pack 2.

The astropulse computation processes seem stable, but the ap_graphics Private working set Memory gradually grows, till the process has to close due to lack of memory. It's disappointing as I like watching them run on my second monitor. :-(

Any ideas/help appreciated! :-)

Kind regards

Lesley
ID: 1073875 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 1073995 - Posted: 4 Feb 2011, 0:54:46 UTC - in response to Message 1073875.  

Hiya,

I don't know if this issue has been addressed before, but I seem to be experiencing a "memory leak" problem with the ap_graphics_5_05_windows_intelx86 program. Its the process that is started with the "Show Graphics" option under the Tasks tab in the Boinc Manager.

So 3 questions.

1. Does anyone else experience this problem?

2. Have I caused this myself with wrong settings in the manager?

3. Is it a problem specific to my setup?

I am running Bionc on an Athlon II X3 440, Asus M4A78LT-M LE motherboard, 1GB DDR3 RAM, Windows Vista Home Premium Service Pack 2.

The astropulse computation processes seem stable, but the ap_graphics Private working set Memory gradually grows, till the process has to close due to lack of memory. It's disappointing as I like watching them run on my second monitor. :-(

Any ideas/help appreciated! :-)

Kind regards

Lesley

It sounds like you have found a bug. I am going to contact the dev team to get in contact with you.


BOINC WIKI
ID: 1073995 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1074022 - Posted: 4 Feb 2011, 2:48:25 UTC - in response to Message 1073995.  

oops! I didn't think it was that serious! :-O

Thanks for the quick response. :-)

I'll update once I hear from the Dev Team.

L x
ID: 1074022 · Report as offensive
Eric Korpela Project Donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 3 Apr 99
Posts: 1382
Credit: 54,506,847
RAC: 60
United States
Message 1084706 - Posted: 7 Mar 2011, 18:57:36 UTC - in response to Message 1074022.  

Hi Lesley,

I'll try to replicate the problem here. I may ask some questions about your system.

How big does the working set get before it dies?

Eric
@SETIEric@qoto.org (Mastodon)

ID: 1084706 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1084828 - Posted: 8 Mar 2011, 2:00:06 UTC - in response to Message 1084706.  

HIya Eric,

Really it can vary, I've saw it as large as 400MB, and remembering I have 3 threads running, it get big! It does take a bit of time, but after a while it basically fills the memory, and then windows tries to start shuffling things around, fails, then wants to close programs to clear memory.

How I discovered it, was I ran task manager, then started the graphics modules (3 in my case, as I have a tri core athlon), and if you watch the memory, you will notice over time a slow creep up of memory, and if you watch the setigraphics or ap_graphics threads, you will see the private working set continually grow over time. This will keep going till windows has problems with memory management, and pop up to close threads. I havent been running them recently as I was away a few days, and it does eventually crash windows.

My current set up is an Athlon 440 3ghz tri core, asus M4A78LT-M LE motherboard, i've upped the memory to 2GB, though some is pinched for graphics (onboard AMD 760G) which leaves 1789MB physical mem according to task manager. OS is Vista Home Premium SP2 .

I'm running the threads just now, I'll post again the rough size when windows complains. If you need more information, please feel free to ask! And remember, i'm no programmer, i'm a hardware monkey. :-)


At time of writing, they are at about 240MB each thread, two ap_graphics, and one setigraphics.

Kind regards

Lesley x


Hi Lesley,

I'll try to replicate the problem here. I may ask some questions about your system.

How big does the working set get before it dies?

Eric


ID: 1084828 · Report as offensive
Eric Korpela Project Donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 3 Apr 99
Posts: 1382
Credit: 54,506,847
RAC: 60
United States
Message 1084832 - Posted: 8 Mar 2011, 2:08:41 UTC - in response to Message 1084828.  

Thanks. I'm not seeing it on this machine, but it could have something to do with specific hardware (i.e. I might need to track down an ATI card) or specific workunits. It could also be that its related to CPU speed, in which case this machine might not be fast enough for me to see it. But I'm on the case, for what good that does.
@SETIEric@qoto.org (Mastodon)

ID: 1084832 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1084835 - Posted: 8 Mar 2011, 2:25:15 UTC - in response to Message 1084832.  

Thanks Eric,

As I'm typing the private working set is at 420,000KB and counting, the commit size (just turned this option on) is 505,000KB and counting, memory usage about 94% and stuff getting shuffled into the page file 3831MB in size, 1393MB free and decreasing. :-)

I'm sure once the page file is full, it'll start playing up. Gotta love windoze. :-)


Lesley x
ID: 1084835 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1084852 - Posted: 8 Mar 2011, 3:24:50 UTC - in response to Message 1084835.  

A wee update.

The commit size is getting WAY higher than the private working set.

I've got up 1GB per thread!

I took a wee printscreen, also my pagefile's increasing to cope.

Print Screen Link

Lesley
ID: 1084852 · Report as offensive
Eric Korpela Project Donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 3 Apr 99
Posts: 1382
Credit: 54,506,847
RAC: 60
United States
Message 1084878 - Posted: 8 Mar 2011, 4:22:16 UTC - in response to Message 1084852.  

Thanks, I'll capture the workunits that are currently assigned to you so I can take a look at them.


@SETIEric@qoto.org (Mastodon)

ID: 1084878 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1084900 - Posted: 8 Mar 2011, 5:09:28 UTC - in response to Message 1084878.  

Heres the sorry tale of the memory buildup.

Come capture 8, the graphics froze, machine sluggish, but functioning. The Commit size makes interesting reading. I don't think its a work unit problem, but a problem with the graphics processes, as you'll notice just below the crunching threads are running fine, with no memory problems.

Screen 1
Screen 2
Screen 3
Screen 4
Screen 5
Screen 6
Screen 7
Screen 8

Lesley x

ID: 1084900 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1091576 - Posted: 30 Mar 2011, 8:08:26 UTC

Another wee update.

It seems to be specific to my Athlon 440, as another machine I have (Athlon 3200+, ASUS A7V600 M/B, Nvidia Geforce 6600LE Graphics) doesn't have the problem. I'm wondering if it could be a graphics driver problem in the Athlon 440 machine?

Lesley
ID: 1091576 · Report as offensive
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9541
Credit: 50,759,529
RAC: 60
Yemen
Message 1091637 - Posted: 30 Mar 2011, 15:08:15 UTC - in response to Message 1091576.  

I recall reading somewhere in the forums that the 2XXX series ATI cards are now considered legacy cards. You could try an update but I'm not sure it will help.


In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 1091637 · Report as offensive
Eric Korpela Project Donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 3 Apr 99
Posts: 1382
Credit: 54,506,847
RAC: 60
United States
Message 1091651 - Posted: 30 Mar 2011, 15:55:22 UTC - in response to Message 1091576.  

I haven't been able to reproduce it on the ATI card I have (also a legacy card). But it's possible that it is a driver issue. Let me know if an update to the latest driver helps.

Eric

Another wee update.

It seems to be specific to my Athlon 440, as another machine I have (Athlon 3200+, ASUS A7V600 M/B, Nvidia Geforce 6600LE Graphics) doesn't have the problem. I'm wondering if it could be a graphics driver problem in the Athlon 440 machine?

Lesley


@SETIEric@qoto.org (Mastodon)

ID: 1091651 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 1091699 - Posted: 30 Mar 2011, 18:23:21 UTC

The Radeon HD 2000 series may be legacy to us, but AMD is still supporting them in the latest Catalyst drivers.
ID: 1091699 · Report as offensive
Eric Korpela Project Donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 3 Apr 99
Posts: 1382
Credit: 54,506,847
RAC: 60
United States
Message 1091717 - Posted: 30 Mar 2011, 20:17:26 UTC - in response to Message 1091699.  

Good. I hate when GPU manufacturers drop old cards prematurely.
@SETIEric@qoto.org (Mastodon)

ID: 1091717 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1094343 - Posted: 7 Apr 2011, 16:42:00 UTC


I don't think it is a legacy card issue, its the onboard graphics on a relatively new AM3 board. :-S

Still, I'll see if I can find a new driver, although I'm pretty sure I put the most recent one in, the machine only got pieced together in the past 3 months. Hopefully AMD has an update out. I'll let you know how that goes.....:-)

Kind regards

Lesley x


ID: 1094343 · Report as offensive
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9541
Credit: 50,759,529
RAC: 60
Yemen
Message 1094525 - Posted: 8 Apr 2011, 0:08:48 UTC - in response to Message 1094343.  
Last modified: 8 Apr 2011, 0:09:28 UTC

The integrated graphics might be the problem since it usually shares memory on your RAM.
You wouldnt happen to have an extra GPU lying around to test it.


In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 1094525 · Report as offensive
LateLesley

Send message
Joined: 20 Mar 04
Posts: 9
Credit: 616,250
RAC: 0
United Kingdom
Message 1098880 - Posted: 19 Apr 2011, 4:56:43 UTC - in response to Message 1094525.  

Well, I just shelled out £45 on a Geforce 8800 GTX, and it seems to have solved the problem. It also has the added advantage of running CUDA tasks, which has upped my crunching ability. I must say I'm impressed with the speed the graphics card can crunch numbers at. :-)

The issue seems to be related to the particular onboard graphics setup, on the AMD chipset. :-S

All sorted now.

Thanks to all for their help and suggestions, Especially Eric who spent some time investigating this issue.

Lesley x
ID: 1098880 · Report as offensive

Questions and Answers : Windows : ap_Graphics_5_05_windows_intelx86 Memory Leak?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.