1) Forums : Technical Support : Problem getting work` (failed downloads) (Message 8069)
Posted 10 Apr 2009 by Skip Da Shu
Post:
If it matters same problem in x64 Linux:

Thu 09 Apr 2009 10:53:54 PM CDT|Cosmology@Home|Finished download of camb_2.15_x86_64-pc-linux-gnu
Thu 09 Apr 2009 10:53:54 PM CDT|Cosmology@Home|[error] Signature verification failed for camb_2.15_x86_64-pc-linux-gnu
Thu 09 Apr 2009 10:53:54 PM CDT|Cosmology@Home|[error] Checksum or signature error for camb_2.15_x86_64-pc-linux-gnu
2) Forums : Technical Support : Sleepy workunit (Message 6313)
Posted 12 Jun 2008 by Skip Da Shu
Post:
Oh and Al I recommend 400-500 mb ram per core on this project.


Ran 3 quads today for a couple of hours with each machine forced to only execute two projects (suspended all else). After a few combos it seemed the ones with Cosmo running were eating up memory. Now down to two quad machines each running only Cosmo (4 WUs at a time)... logging mem and swap usage. One is a 1GB RAM machine and the other is a 2GB RAM, both have 1.9GB swap files. Results soon but... So far I think all I\'m doing is proving the quoted line above.


The 1GB machine maxed out at 1132MB of memory + swap being used while running 4 Cosmo WUs or averaging the system memory usage over the workunits, just over 283MB per WU (Xubuntu, v8.04, 64b, x11vnc server active, gnome sysmon, gkrellm and boinc gui running). However the machine with 2GB installed maxed out at 996.1GB or just under 250GB per WU.

1GB (971.3MB available, swap 1.9GB) machine running 4 WUs(entries after restart):

time - memory - swap
3:53pm 513.0MB 0 bytes
4:03pm 567.5MB 20.0KB
4:13pm 512.3MB 20.0KB
4:23pm 515.8MB 20.0KB
4:29pm 830.8MB 33.6MB
4:34pm 914.9MB 217.7MB (Max)
4:39pm 818.8MB 122.6MB
4:44pm 427.2MB 69.4MB
4:54pm 449.8MB 69.2MB
5:06pm 597.8MB 68.5MB
5:46pm 391.3MB 157.7MB

2GB (2.0GB available, swap 1.9GB) machine running 4 WUs (entries after restart):

time - memory - swap
4:06pm 673.0MB 0 bytes
4:11pm 820.8MB 0 bytes
4:21pm 708.8MB 0 bytes
4:30pm 530.3MB 0 bytes
4:35pm 660.9MB 0 bytes
4:45pm 661.8MB 0 bytes
4:55pm 683.7MB 0 bytes
5:05pm 969.1MB 0 bytes (Max)
5:47pm 461.2MB 0 bytes

So it looks to me like I might have been able to avoid \"swap land\" on the 1GB machine if the full GB had been available (will check but suspect this machine has 32 or 64MB tied up for onboard video). But for sure someplace around 1.2GB would\'ve keep things happier.

Bottom line is I don\'t see a memory leak just some true swap file usage due to WU memory sizes. This is compounded for me as I currently run with the \"leave apps in memory\" option enabled.
3) Forums : Technical Support : Sleepy workunit (Message 6312)
Posted 12 Jun 2008 by Skip Da Shu
Post:
Oh and Al I recommend 400-500 mb ram per core on this project.


Ran 3 quads today for a couple of hours with each machine forced to only execute two projects (suspended all else). After a few combos it seemed the ones with Cosmo running were eating up memory. Now down to two quad machines each running only Cosmo (4 WUs at a time)... logging mem and swap usage. One is a 1GB RAM machine and the other is a 2GB RAM, both have 1.9GB swap files. Results soon but... So far I think all I\'m doing is proving the quoted line above.
4) Forums : Technical Support : Sleepy workunit (Message 6309)
Posted 12 Jun 2008 by Skip Da Shu
Post:
Hmmmm... was just browsing the forum here looking to see if anyone has reported any memory leak type problems.

I have NOT isolated down which project it is but a couple (maybe all) of my Linux machines (Xubuntu v8.04, 64b) are experiencing a growing memory usage that I don\'t I recall having seen in the past. Right now I\'m just going from a graphical memory and swap usage bar in gkrellm. I saw around 75% memory usage and 50% swap usage today. I am more used to seeing 33~50% and 0~1% respectively. The machines become sluggish logging into them. A restart of BOINC resolves it for a day or two.

Hopefully tomorrow I will get some more time to investigate and see if I can really narrow down what WUs on what project is causing this. If somebody else finds something first please post it.

Very little help yet but ... Projects running on Xubuntu machines that I\'ve been seeing this on (top 3 projects get 45% of resource share):

ABC
CPDN
Cosmo
Einstein
Leiden
Malaria
MilkyWay (was down last I checked, no work)
Orbit (no work last 2+ days)
POEM
QMC
RCN
RieselS
Rosetta
Simap
SpinH
WCG
5) Forums : Wish list : Feedback on planned Cosmology@Home supporting material and C@H prize idea (Message 5920)
Posted 13 Apr 2008 by Skip Da Shu
Post:
Will crunch for coffee mugs... 126GHz available :-)