Anyone else experiencing WU / memory / sluggish response problems?
I noticed that over the last 2 weeks that the memory requirement per task started increasing for this project.
I have also noticed that the CPU interrupts are not working very well -> the tasks are hogging the CPU and not handing back very well at all causing desktop lockups in Windows 7 x64.
The main issue of limited memory represents as WU-hang - work units that never finish - which has meant some baby sitting.
Check any of your machines that have around 1.5GB per core. 2GB per core should still be OK at time of writing.
The upshot is that after isolating my other projects from interruption and restricting work units to a suitable RAM-per-core I am suddenly down to 2 cores (2 separate dual core machines) on this project!
Hopefully the trend will not continue.
. . . . . ___
. . . . . . .\___/\______
. . . . . . . \__AMD___\\__
---------------------------------------------
Yes agree with the memory hogging in this project Dave. On the official site there is an app_config.xml file posted that I use now on my 4GB RAM Win7 systems. It limits running tasls to 2 or 3 instead of 4. On a Xeon system of mine with 8GB when it was trying to run 8 threads the tasks were taking 12+ hours. After limiting the system to 7 tasks they now finish in 1-2 hours. This machine is still suffering slow downs so I might limit it to 6 or even 5 concurrent tasks.
<app_config>
<app>
<name>example_app</name>
<max_concurrent>2</max_concurrent>
</app>
</app_config>
Just a heads-up team that I will be slowing my participation in 123 when I reach 1M. I cannot reach the boxen in Canberra as my VPN is down. The boxes continue to crunch. I'll be back there in a week.
I had the same problems with FiND until today, when my A8-3850 started to crunch them as a rabbit on steroids -to great dismay of John C. MacAlister I guess
I have one machine running 123numbers with 8Gb of RAM (4 cores) and another with 13Gb (6 cores), so far both seem to be ok but I hope they don't increase the memory requirements again or I may have to lower the number of cores for it, or switch to something else.
2009 MULTI CORES CONTEST STATS
http://neogen.amdusers.com/contest2009/contestoverall.htm
CHRISTMAS QUEST CONTEST STATS
http://neogen.amdusers.com/contest2008/xmascontest.htm
http://neogen.amdusers.com/contest2008/stats.png (snapshot)
FEBRUARY '08 RACE STATS
http://neogen.amdusers.com/contest2008/racefeb08b.htm
http://neogen.amdusers.com/contest2008/racefeb08nb.htm
Thanks guys. I'll try the app config when I get time.
. . . . . ___
. . . . . . .\___/\______
. . . . . . . \__AMD___\\__
---------------------------------------------
My project stats are showing 100% invalid since 2am 6th April on all machines.
Aborting and backing off for a while.
. . . . . ___
. . . . . . .\___/\______
. . . . . . . \__AMD___\\__
---------------------------------------------
After a sabbatical from this project, I gave it another run.
100% invalid work unit results on Windows client and linux client.
As someone already pointed out elsewhere this project has ground to an almost complete halt with zero successful results from all users only days away and the admins are not doing anything about it.
http://stats.free-dc.org/stats.php?page=proj&proj=vdw
. . . . . ___
. . . . . . .\___/\______
. . . . . . . \__AMD___\\__
---------------------------------------------
UPDATE on 123Numbers.ord : Van Der Waerden Numbers
The project admin recently broadcast a message out to existing members that a number of issues were fixed.
Testing so far reveals that the results are positive.
For the most part the validator itself was not the problem, it was memory allocation bugs in the clients causing mismatches in the result validation.
The good news is that the project has new clients and this time it is working as intended.
There are still some errors coming back from a few windoze machines but that is being monitored and assessed.
Otherwise all stations are GO!
. . . . . ___
. . . . . . .\___/\______
. . . . . . . \__AMD___\\__
---------------------------------------------
I wait a bit after getting zilch credits for
Invalid (7150) · Error (3913),
which may very well have been caused by crappy programming of the client.