Welcome to MilkyWay@home

So what's the project status??

Message boards : Number crunching : So what's the project status??
Message board moderation

To post messages, you must log in.

AuthorMessage
JAMC

Send message
Joined: 9 Sep 08
Posts: 96
Credit: 336,443,946
RAC: 0
Message 15883 - Posted: 17 Mar 2009, 23:57:44 UTC

It's been many days since the 'great fix' and subsequent slide into 'got 0 new tasks' hell...
Any update from the project staff as to what is being done to get us back to work aplenty?
Thanks
ID: 15883 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 21 Aug 08
Posts: 625
Credit: 558,425
RAC: 0
Message 15885 - Posted: 18 Mar 2009, 0:23:06 UTC - in response to Message 15883.  
Last modified: 18 Mar 2009, 0:24:25 UTC

It's been many days since the 'great fix' and subsequent slide into 'got 0 new tasks' hell...
Any update from the project staff as to what is being done to get us back to work aplenty?
Thanks


Obviously YMMV, but my system has been idle for a grand total of about 38-40 minutes today, which means that out of a 24 hour day, my system had work from this project (I'm only processing for this project right now) for at least 97.22% of the day. I appreciate that your system is faster and/or you have more systems and/or have a higher drive to see your rankings increase, but no project can guarantee 100% uptime or 100% availability of work.
ID: 15885 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JAMC

Send message
Joined: 9 Sep 08
Posts: 96
Credit: 336,443,946
RAC: 0
Message 15887 - Posted: 18 Mar 2009, 0:34:59 UTC

Mine are idle more than that, but the point is things were working nearly perfectly for a week or so and then began slipping again so the potential is there for better... my credit today dropped ~ 25% from yesterday... the project appears in a holding pattern with Travis being incommunicado...
ID: 15887 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 15890 - Posted: 18 Mar 2009, 0:42:07 UTC

New work seems to be cyclic, when I look at my machines.

Those running normally always have full caches, even if these are limited to 6 WU per core. The one GPU cruncher seems to cycle from a full cache (6 per core) to nothing, and the server back off working. I cannot really say how much idle time was spent during the day, but impressions seem to make out the GPU was not doing nothing for too long a period of time (I hope).
ID: 15890 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 21 Aug 08
Posts: 625
Credit: 558,425
RAC: 0
Message 15892 - Posted: 18 Mar 2009, 0:49:58 UTC - in response to Message 15887.  

Mine are idle more than that, but the point is things were working nearly perfectly for a week or so and then began slipping again so the potential is there for better... my credit today dropped ~ 25% from yesterday... the project appears in a holding pattern with Travis being incommunicado...


You're probably going to see good days and bad days, depending on how heavily people are banging on the door with the GPU apps. People running update scripts somewhat irritate me, because while I could do it, I feel it is akin to attempting to "butt in line". So far, I've just let BOINC do its' thing and am at 97-98% utilization... Dunno what to tell ya...
ID: 15892 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 41
Message 15896 - Posted: 18 Mar 2009, 1:40:28 UTC - in response to Message 15890.  
Last modified: 18 Mar 2009, 1:42:12 UTC

I've noticed that getting CPU tasks has gotten a more tedious (0 new tasks) of late -- as it was with the larger queues. I find often I have to pulse the server with manual update requests 4 to 8 times and sometimes more to refill the cache on workstations.

The thing is, with the relatively short work unit times of 15 to 25 minutes, the small cache readily can run dry. Originally the 6 unit per core wasn't too bad as I wasn't seeing the 0 new task monster and so it offset having the small cache size. But now, it is becoming a bit of a lose-lose scenario -- small cache AND the dreaded 0 new task problem. It isn't quite as bad as it was with the 8 and 12 task per CPU configurations of a couple of months ago, but it is definitely not as clean as it was 3 or 4 weeks ago.
ID: 15896 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 21 Aug 08
Posts: 625
Credit: 558,425
RAC: 0
Message 15898 - Posted: 18 Mar 2009, 2:13:39 UTC - in response to Message 15896.  
Last modified: 18 Mar 2009, 2:26:24 UTC

It isn't quite as bad as it was with the 8 and 12 task per CPU configurations of a couple of months ago, but it is definitely not as clean as it was 3 or 4 weeks ago.


YMMV... As indicated, my system has only been idle for 2-3% of the day today... My next work fetch attempt should be in about 6 minutes. I'll edit with the results...then I'm planning on some "mind over mattress" time...

Edit:

The task completed then started uploading. A request for work went out while the uploading was still in process. I was assigned 5 new tasks, which is not the expected 6, but who knows if there might've not been anything else ready at that point...OR (DUH!) it could've been that the server still thought I had 1, so they've got the max in cache (or something like that) turned on and 1 + 5 = 6 (?)

Time for bed...
ID: 15898 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile GalaxyIce
Avatar

Send message
Joined: 6 Apr 08
Posts: 2018
Credit: 100,142,856
RAC: 0
Message 15917 - Posted: 18 Mar 2009, 4:51:04 UTC - in response to Message 15892.  
Last modified: 18 Mar 2009, 4:52:51 UTC

People running update scripts somewhat irritate me, because while I could do it, I feel it is akin to attempting to "butt in line".

It's irritating that someone should have publicised this feature in this thread - Update Script. And then to go on to encourage with;

My update script appears to be working well. It's been able to update and get work when my 4850 has run out a few times now. It's not hammering the schedular, just every so often. Still, it's a little clunky and should really look for the dreaded "Got 0 new tasks" message - not just update every so often.

What I find even more irritating is that he didn't even share his findings, just ask for help, brag about having found it, and then brag that it's "not hammering the schedular" as if he knew squat about what he was talking about.

ID: 15917 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Thamir Ghaslan

Send message
Joined: 31 Mar 08
Posts: 61
Credit: 18,325,284
RAC: 0
Message 15923 - Posted: 18 Mar 2009, 5:32:10 UTC - in response to Message 15896.  
Last modified: 18 Mar 2009, 5:34:28 UTC

I've noticed that getting CPU tasks has gotten a more tedious (0 new tasks) of late -- as it was with the larger queues. I find often I have to pulse the server with manual update requests 4 to 8 times and sometimes more to refill the cache on workstations.

The thing is, with the relatively short work unit times of 15 to 25 minutes, the small cache readily can run dry. Originally the 6 unit per core wasn't too bad as I wasn't seeing the 0 new task monster and so it offset having the small cache size. But now, it is becoming a bit of a lose-lose scenario -- small cache AND the dreaded 0 new task problem. It isn't quite as bad as it was with the 8 and 12 task per CPU configurations of a couple of months ago, but it is definitely not as clean as it was 3 or 4 weeks ago.



My tasks host history was reset today, but I've noticed that the number of times client contacted server > number of tasks by at least a margin of x2!
ID: 15923 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 15937 - Posted: 18 Mar 2009, 10:27:11 UTC

Part of the problem is that for those of us running the GPU application is that the BONC Manager knows nothing of the GPU and does not properly account for it in the resource planning. Even sadder, and more long lasting is that the developers have (to this point) refused to acknowledge that the Resource Share model is broken by the addition of GPUs to the computing mix (using the theory that if the problem is ignored long enough it will go away).

These issues even affect those of us running the CUDA applications as the work fetch calculations for all the versions of BOINC do not make proper distinctions between CPU and GPU work loads. So, to keep the systems fed we must bash our system settings to make it work.

The upside to this is that the earning power of a good GPU is pretty fantastic ... each GPU, on average, has earned me an additional 20K CS per day (5 GPUs) and I don't even have the top of the line GPUs in all slots yet ... so there is room for growth yet ... :)
ID: 15937 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile James Sotherden
Avatar

Send message
Joined: 3 Jan 09
Posts: 139
Credit: 50,066,562
RAC: 0
Message 15979 - Posted: 18 Mar 2009, 17:15:50 UTC

I run seti and milkyway on this mac, running stock. And some times I have a hard time getting WU's. I figure it will only get worse seeing how every one cant wait to get a GPU. I can see the handwriting on the wall, someday i will be forced to get o GPU or get out of both projects.
ID: 15979 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 29 Aug 07
Posts: 115
Credit: 501,599,202
RAC: 12,070
Message 15985 - Posted: 18 Mar 2009, 17:49:28 UTC

Any chance the project could just make the tasks run longer, or bundle multiple tasks per WU, or...?

ID: 15985 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JAMC

Send message
Joined: 9 Sep 08
Posts: 96
Credit: 336,443,946
RAC: 0
Message 16815 - Posted: 25 Mar 2009, 8:24:52 UTC

The lights are on, but is there anybody home??
Offers of equipment and even money to help improve the project haven't enticed any response ... there are flashes of new work availability but still a whole lot of 'got 0 new tasks' and machines with no WU's to crunch.
Any developments to report?
Thanks
ID: 16815 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile The Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
Message 16816 - Posted: 25 Mar 2009, 9:08:48 UTC

Q. So what's the project status??
A. Dead Quiet!
ID: 16816 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JAMC

Send message
Joined: 9 Sep 08
Posts: 96
Credit: 336,443,946
RAC: 0
Message 16824 - Posted: 25 Mar 2009, 16:54:30 UTC
Last modified: 25 Mar 2009, 16:55:12 UTC

Well I guess if enough people leave for lack of work the demand will drop enough to meet the supply curve and all will be right with the world. The project is getting all the work it can generate processed and very quickly so I guess they can call it a roaring success, but I have switched a 4870 over to F@H so at least it is doing something... when the little brown truck shows up with a package I am configuring a dual 4870 machine and it may follow the first to F@H if there continues to be no work for it...but hey, more work available for everyone else :)
ID: 16824 · 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 16825 - Posted: 25 Mar 2009, 17:09:25 UTC

Travis informed me that he's working on a project status update and it should be ready by the end of the day today.
Dave Przybylo
MilkyWay@home Developer
Department of Computer Science
Rensselaer Polytechnic Institute
ID: 16825 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Bruce
Avatar

Send message
Joined: 28 Apr 08
Posts: 1415
Credit: 2,716,428
RAC: 0
Message 16827 - Posted: 25 Mar 2009, 17:30:55 UTC - in response to Message 16825.  
Last modified: 25 Mar 2009, 17:31:35 UTC

Thanks for the update Dave, was starting to think ALIENS had kidnapped Travis and were reprogramming his mind! Tee-hee ;-P
ID: 16827 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Al*
Avatar

Send message
Joined: 8 Nov 07
Posts: 323
Credit: 1,362,120
RAC: 0
Message 16832 - Posted: 25 Mar 2009, 18:10:23 UTC

Major coolness.

ID: 16832 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Arion
Avatar

Send message
Joined: 10 Aug 08
Posts: 218
Credit: 41,846,854
RAC: 0
Message 16833 - Posted: 25 Mar 2009, 18:21:30 UTC - in response to Message 16825.  

Travis informed me that he's working on a project status update and it should be ready by the end of the day today.


I sure hope its good news about having work available and larger caches. Being out of work more than having it seem to defeat the purpose of processing.



ID: 16833 · 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 16840 - Posted: 25 Mar 2009, 19:16:22 UTC - in response to Message 16825.  

Travis informed me that he's working on a project status update and it should be ready by the end of the day today.


Looking forward to it. (hopefully) :p
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 16840 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : So what's the project status??

©2024 Astroinformatics Group