World Community Grid

Leaving it is for the best I think. IIRC I had trouble running Atlas at load on my Windows 4p during the Pentathalon so ran SixTrack which managed to fully load the cpu.

Just a quick thing to check if you have the time. On the LHC website, under your account and then computing preferences are you set to use 100% of the cpus even when the computer is in use?
 
MGMCCALLEY, assuming you have virtual box already installed and CPU virtualization is enabled, check "ATLAS Simulation" in your preference menu.

I added this app_config.xml file and it works fine on my Windows 7 rig. I wanted to run 5 concurrent tasks but I think my memory or CPU is limiting this since I'm also running SixTrack.

<app_config>
<app>
<name>ATLAS</name>
<max_concurrent>5</max_concurrent>
<fraction_done_exact/>
</app>
<app_version>
<app_name>ATLAS</app_name>
<plan_class>vbox64_mt_mcore_atlas</plan_class>
<avg_ncpus>4.0</avg_ncpus>
<cmdline>--memory_size_mb 5800</cmdline>
</app_version>
</app_config>

I just finished 3 tasks and the points are good but somehow during the Pentathlon, I noticed that the credit score keeps going down. Might want to keep an eye on this. BTW ATLAS does not need a wingman.

upload_2017-6-22_22-14-45.png
 
Since this is a 4P, also make sure you have enough RAM being that you have 48cores at your disposal. The vbox applications typically take more than most projects.
 
Yes, I have 100 percent usage. I also have 64 GB of DDR3 ECC, so I believe that's enough for anything required. I'm guessing I don't need virtualization or Vbox in general, given that someone said Six Track doesn't use Vbox. Also, magically the CPU is now about 90% loaded. And it seems I got 2k points yesterday? Odd that I had 1300 or so when I had only been running the project for a couple of hours. I have no idea how the stats are figured or if they give some kind of bonus to new users. Either way, I'm happy to help for the weekend.

Have fun, good luck.
 
It appears that the FAH has no work at the moment. Well, I had a very good run with this project. Hopefully something will come up in the next few months.
 
We have identified the issue and are working on the clean up process. Once we have everything cleaned up we will restart sending out work omitting the batches that have the problem.

Thanks,
armstrdj
 
We are getting ready to restart sending of MIP1 work. The batch numbers that had the bad input files were in the range of 4783 - 4872. We have done what we can to prevent any more of these from going out. Thanks for everyone's patience while we cleaned this up.

Thanks,
armstrdj
 
can it be that HSTB is really that slow or so monster WU ? around a day each WU on a Ryzen 1700x ?

Screen Shot 2017-10-09 at 19.28.09.png

A bit unexpected ... other WU before was around 3 hours ... first I complain I don't get WU's ... now I complain they take too much time ... please someone make my mind :rolleyes:


Update: thats from an Intel i7 6700 driving also GTX 1080; much faster ... Intel in favor for AMD ?
Screen Shot 2017-10-09 at 19.45.28.png

Update 2: seems to be ok'ish

found some post ( [Link] ) that WU with T300,T325,T350 or T400 in their names are in deed monster WUs. At least it seems not trouble with OS or CPU.
 
Last edited:
ChristianVirtual, I think the long run times may be caused by running so many at the same time. I've seen similar behaviour on my 4p. A normal run time for 2 or so concurrent HSTB monster tasks is normally under 12 hours on a Ryzen.

These are some results from a Windows 1800X @ 3.7GHz.

Ryzen HSTB Run Times.png
 
ChristianVirtual, I think the long run times may be caused by running so many at the same time. I've seen similar behaviour on my 4p. A normal run time for 2 or so concurrent HSTB monster tasks is normally under 12 hours on a Ryzen.

These are some results from a Windows 1800X @ 3.7GHz.
Interesting observation; but how can I influence the mix off assignments ? Sure, I can open up again to other WCG sub projects (I just did as the current method not getting new HSTB WU in at all). So let see if a mix of MIP and HSTB improve the runtime.
But still would not understand. Memory was not a bottleneck. No swapping. Maybe process-Lasso style of assignment ?
 
I'm not sure mate. I was equally scratching my head when it happened a couple of times. It's usually not an issue as HSTB tends to be in the minority on a particular rig and I think today is actually the first time I haven't received any at all.

It could just be co-incidence and the true cause is an occasional release of uber wtf monster tasks.
 
Something wrong with WCG ? Somehow I don't get any project today; already opened up to all projects and "updated" all clients. Nothing really comes.

Is there a server page we can see the queue ?
 
I just grabbed 29 new units

World Community Grid 10/22/2017 10:07:00 AM Scheduler request completed: got 29 new tasks


Current time for me right now is 10:08. So they're sending work. What does your log say CV?
 
eventually I found the log said: my preferences preventing downloads... funny: I can't remember that I changed any except reassign profiles to certain computer and updated.

I now tried to reset all profiles and assign once more; and download started again ... still don't understand why.
 
So maybe this is common knowledge, and I'm just in the dark after being offline for so long, but since when does WCG have a "per machine" limit for jobs? Some of my systems have a lot of cores sitting idle due to the 66 job limit. At least I think that's the limit as it's the same number of max running jobs on a couple different boxes. Any way to deal with this?
 
You know... I hope the above problem isn't some how due to my using a common Windows system image on different machines. I do change the system name and reboot. Hopefully it's not that. Do know I've spent WAY too much time this afternoon / evening getting a bunch of systems online for the weekend and I'm going to be a bit ticked off if it was all for nothing. :(
 
Something else is up. That's not a limit. Check the projects you're signed up for.

One of my 64 core rigs got 1000+ tasks in a 2 day cache setup.
 
I was signed up for certain ones; but did have the option enabled to accept other work. Just changed it a few minutes ago to be opted in to all tasks. Just checked the event log and it's saying no tasks are available for all projects... which can't be right.
 
Yet another box I just put online pulled down a bunch of WUs. FFS. Oh well, screw this, I'm heading home. Brought ~8 systems online today for a total of 13; the weakest of which is a 4P E5-4607. ;) This issue might be stopping them from running at 100%... but it's still a shit ton of WUs being cranked out.
 
Any idea where I can maintain the names of my computer for boinc / WCG ?

B6B88E21-ECBB-490C-B424-703E8C8A07BB.jpeg

They’d are not very descriptive the way they are ...
 
It pulls the name directly from your computer. I renamed a computer and it updated on the site a short time later. I am not too familiar with anything but Windows so I don't know where to rename it in other OSs.
 
Ouch, thanks ...

I have nice DNS names defined but not aligned the hostnames ... calling myself an idiot I hope is not violating any rules ?
 
Nah, we can call ourselves idiots all we like. ;) In fact, am sure I've called myself an idiot in various threads here numerous times! :ROFLMAO:

Am 99% sure you know this, but on the very off chance you don't... you're the only one that gets to see those computer names in the stats. :)
 
Oh, and in other news, this here idiot just noticed that MIP has ticked over to Sapphire. :)
mip1_5.jpg


OpenZika crossed 78 years, too.
 
free-dc.org quit updating my stats 11 days ago, my stats still update on the WCG site, and other people are getting updated at free-dc...weird
 
free-dc.org quit updating my stats 11 days ago, my stats still update on the WCG site, and other people are getting updated at free-dc...weird

This has been happening for me as well but its random. sometimes when i check it i see my stats with the rest of the team and if i click my username i can see recent work ,but a few hours later im not on the team and when i search my username it says i havn't submitted work for a week. check back 4 hours later and im back again.
 
Back
Top