Welcome to MilkyWay@home

Won't finish in time

Message boards : Number crunching : Won't finish in time
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Bill

Send message
Joined: 3 Oct 07
Posts: 21
Credit: 49,862
RAC: 0
Message 26745 - Posted: 30 Jun 2009, 12:33:45 UTC

I run Boinc with 11 projects, (Milkyway gets 2%.) But I can't get any work, server returns: (won't finish in time) BOINC runs 100.0% of time, computation enabled 99...

I suggest increasing the deadline/task size ratio to get more participation.
ID: 26745 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John R. @ SETI.USA

Send message
Joined: 1 Jan 09
Posts: 15
Credit: 85,816,654
RAC: 0
Message 26747 - Posted: 30 Jun 2009, 13:47:10 UTC - in response to Message 26745.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

ID: 26747 · 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 26749 - Posted: 30 Jun 2009, 14:15:27 UTC - in response to Message 26747.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

It seems to me that I've had to spend a lot of time tweaking to get things working right. Then it seems to decay into further problems. Right now I've gone back to 5.10.45 on all my heaters (as in PCs) to see if that gives more consistency.


ID: 26749 · 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 26753 - Posted: 30 Jun 2009, 15:39:36 UTC - in response to Message 26749.  
Last modified: 30 Jun 2009, 15:41:01 UTC

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

It seems to me that I've had to spend a lot of time tweaking to get things working right. Then it seems to decay into further problems. Right now I've gone back to 5.10.45 on all my heaters (as in PCs) to see if that gives more consistency.


I'd encourage 5.8.16... Unless a project requires otherwise, you need CUDA support, or you have Vista / 7, then it's all you really need... The only drawback is 5.8.16 does not support server-side aborts, but since this is a quorum of 1 project...not much missing by not having the support for it...
ID: 26753 · 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 26765 - Posted: 30 Jun 2009, 18:38:05 UTC - in response to Message 26753.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

It seems to me that I've had to spend a lot of time tweaking to get things working right. Then it seems to decay into further problems. Right now I've gone back to 5.10.45 on all my heaters (as in PCs) to see if that gives more consistency.


I'd encourage 5.8.16... Unless a project requires otherwise, you need CUDA support, or you have Vista / 7, then it's all you really need... The only drawback is 5.8.16 does not support server-side aborts, but since this is a quorum of 1 project...not much missing by not having the support for it...

Thanks for that Brian, I'll give 5.8.16 a try.


ID: 26765 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Bill

Send message
Joined: 3 Oct 07
Posts: 21
Credit: 49,862
RAC: 0
Message 26852 - Posted: 1 Jul 2009, 18:20:44 UTC - in response to Message 26747.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

The message comes from the server, so the ratio does not need to be changed. Something else does. The Boinc version shouldn't affect WU allocation.
ID: 26852 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
STE\/E

Send message
Joined: 29 Aug 07
Posts: 486
Credit: 576,514,284
RAC: 37,074
Message 26862 - Posted: 1 Jul 2009, 20:14:50 UTC - in response to Message 26852.  
Last modified: 1 Jul 2009, 20:15:24 UTC

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

The message comes from the server, so the ratio does not need to be changed. Something else does. The Boinc version shouldn't affect WU allocation.


That something else may be you just decreasing the amount of Projects/Work Units your trying to do on your Single Core 1.86Ghz Over Worked Laptop.
ID: 26862 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Bill

Send message
Joined: 3 Oct 07
Posts: 21
Credit: 49,862
RAC: 0
Message 26917 - Posted: 2 Jul 2009, 14:16:43 UTC - in response to Message 26862.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

The message comes from the server, so the ratio does not need to be changed. Something else does. The Boinc version shouldn't affect WU allocation.


That something else may be you just decreasing the amount of Projects/Work Units your trying to do on your Single Core 1.86Ghz Over Worked Laptop.


If boinc is trying to get WUs and is returning all of them on time, then it is not overworked.

On previous versions I would get a Milky, process it and return it, just like the other "minor" projects.

If the Boinc version affects the project then there is a problem with the project.

Your arrogant attitude is another. My lowly laptop has 5K credits on Milky and 10K on Einstein (and we know how big those WUs are) Perhaps you think I should drop Milky and contribute to Einstein instead. Was that your purpose in your post?
ID: 26917 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
STE\/E

Send message
Joined: 29 Aug 07
Posts: 486
Credit: 576,514,284
RAC: 37,074
Message 26927 - Posted: 2 Jul 2009, 16:33:06 UTC - in response to Message 26917.  

I started to see that message when I upgraded BOINC to 6.6.36

I went back to 6.6.20 and all is well.

The message comes from the server, so the ratio does not need to be changed. Something else does. The Boinc version shouldn't affect WU allocation.


That something else may be you just decreasing the amount of Projects/Work Units your trying to do on your Single Core 1.86Ghz Over Worked Laptop.


If boinc is trying to get WUs and is returning all of them on time, then it is not overworked.

On previous versions I would get a Milky, process it and return it, just like the other "minor" projects.

If the Boinc version affects the project then there is a problem with the project.

Your arrogant attitude is another. My lowly laptop has 5K credits on Milky and 10K on Einstein (and we know how big those WUs are) Perhaps you think I should drop Milky and contribute to Einstein instead. Was that your purpose in your post?


You give Milkway a 2% Resource Share along with 11 other Projects and then complain you can't get work, you figure it out. I stand by my statement that mayby your Laptop is just over worked. I'm sure it can do the work from any Project but your not going to get work when you want it with a Resource share like that when your runing 11 other Projects too.
ID: 26927 · 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 26944 - Posted: 2 Jul 2009, 20:43:55 UTC - in response to Message 26745.  

I run Boinc with 11 projects, (Milkyway gets 2%.) But I can't get any work, server returns: (won't finish in time) BOINC runs 100.0% of time, computation enabled 99...

I suggest increasing the deadline/task size ratio to get more participation.

Have you installed the optimised application?

You also have to understand that this project has short deadlines for a reason and isn't going to change it.

If you can't complete a wu in the time you have your machine running with the resource share you have set up, then maybe you need to either 1) increase the resource share, 2) install the optimised application or 3) consider not doing this project.
ID: 26944 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Bill

Send message
Joined: 3 Oct 07
Posts: 21
Credit: 49,862
RAC: 0
Message 26988 - Posted: 3 Jul 2009, 14:16:39 UTC - in response to Message 26944.  

http://boinc.berkeley.edu/dev/forum_thread.php?id=4070 reports:
But with 6.6.36, CPDN gets ALL FOUR CORES assigned (four tasks) and I get no new Leiden or WCG tasks. The messages from the server state that the tasks "won't finish in time".
so this problem is with 3.3.36.

A 2% has nothing to do with getting work. It should only affect how much and how often. That user has 4 cores and experiences the same problem on another project.

Have you installed the optimised application?
No

You also have to understand that this project has short deadlines for a reason and isn't going to change it.
I suggested changing the WU size/deadline ratio. Keep the deadline, use smaller WU. But that was before finding that the problem is Boinc/project related and not because of a recent project change.

If you can't complete a wu in the time you have your machine running with the resource share you have set up, then maybe you need to either 1) increase the resource share, 2) install the optimised application or 3) consider not doing this project.
I already stated that I have 5000 credits on this project and no problems returning WU on time.

your not going to get work when you want it with a Resource share like that
When I want it? Boinc has racked up enough long term debt and wants it.

All you guys want to do is argue (the definition of a Troll!) I'm just reporting the the current version of Boinc has problems getting work from projects including this one, previous version didn't.
ID: 26988 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
Message 26993 - Posted: 3 Jul 2009, 16:44:49 UTC - in response to Message 26988.  
Last modified: 3 Jul 2009, 16:46:39 UTC

All you guys want to do is argue (the definition of a Troll!) I'm just reporting the the current version of Boinc has problems getting work from projects including this one, previous version didn't.

Every new BOINC version introduces new errors, there is nothing to argue ;)

All these guys you libel as trolls just wanted to help you with your problem (or the BOINC one). No need to get personal. BOINC may be designed that way you finally get some work even with a 1% share by building up debts. It may even manage to return them within deadline by going into the high priority panic mode. But frankly, you are begging for problems if you rely solely on these kind of emergency measures.

But back to topic, I really recommend running one of the optimized apps as this reduces the runtime and will alleviate the problem.
ID: 26993 · 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 26998 - Posted: 3 Jul 2009, 18:52:47 UTC - in response to Message 26988.  


Have you installed the optimised application?
No


Why not? It completes a wu in a tenth of the time on a cpu....
ID: 26998 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
STE\/E

Send message
Joined: 29 Aug 07
Posts: 486
Credit: 576,514,284
RAC: 37,074
Message 27002 - Posted: 3 Jul 2009, 19:55:45 UTC - in response to Message 26998.  
Last modified: 3 Jul 2009, 19:56:34 UTC


Have you installed the optimised application?
No


Why not? It completes a wu in a tenth of the time on a cpu....


Have you seen the Comcast Commercial with Bill & Karolyn Slowsky yet ??? Maybe it's Bill from the Commercial & you know how he hates Speed ... hhahahaha ... j/k
ID: 27002 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Nightfire73

Send message
Joined: 4 Apr 08
Posts: 10
Credit: 32,120,887
RAC: 0
Message 27128 - Posted: 5 Jul 2009, 21:14:31 UTC

I have the same problem myself...
Boinc 6.6.36, Dual Core 2.66 GHz, 4 GB Ram, Asus ENGTS250 DK TOP.
I run 12 to 14 projects, including MW with an optimized app. When I get some MW WUs they all complete in time. Of course if I reset the project, I get like 15 WUs and after they are completed (going into high priority mode), I only get 2 to 4 WUs at a time, which is easily manageable. But then all of a sudden the "won't finish in time" message appears, and I don't get any WUs for some time.
ID: 27128 · 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 27141 - Posted: 6 Jul 2009, 0:05:54 UTC

Running 12 to 14 projects is the problem then and will always be when running projects that have 3 to 4 day deadlines. BOINC will ensure it won't try to even get wu's it thinks it won't be able to complete before the deadline.

When running projects that have 3 to 4 day deadlines you also can't have a cache more than 1 day otherwise BOINC will go kafluey.
ID: 27141 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Nightfire73

Send message
Joined: 4 Apr 08
Posts: 10
Credit: 32,120,887
RAC: 0
Message 27156 - Posted: 6 Jul 2009, 11:47:43 UTC - in response to Message 27141.  
Last modified: 6 Jul 2009, 11:50:52 UTC

I'm sorry if I do insist, but I really don't believe the problem resides in running 12 or 14 projects, or at least not the ones I run. MW is the only project with short deadlines, and when I get work, as I stated before, I get 2 to 4 WUs only, and I believe this is because Boinc "knows" that getting more would not allow them to finish in time. And so far that's ok. Of course, If I'd run less projects, I would get more MW WUs.
The real problem is that after completing the daily WUs for some days in a row, the project starts sending the "won't finish in time" message, although all the WUs completed in time (normally the same day, sometimes the next, but always way before the deadline). And I don't have any projects running in high priority either. After a few days I will eventually get new WUs, which will be crunched in time, only to see the ominous message appear again a few days later.
Resource allocation for all my projects is 7.69%, but i run FreeHal, which always runs as a non cpu intensive app, then I run GPUGRID, and that is handled by the GPU as well as SETI. So in the end I have only 10 projects running on the 2 CPUs, which means more than 4 hours crunching for each project every day. On top of that I use an optimized MW app, and WU times go from 30 minutes to 1.5 hours, which IMHO means that even if MW had a 1-day deadline I still should be able to get at least 2 WUs a day. So I don't really believe that the comp is overworked, and if it really was, then I would at least see the apps running in high priority mode, but they don't.
As John R. states, the problem appeared after installing Boinc 6.6.36. Everything worked fine before, and I used to run up to 16 projects at a time.
I believe the problem resides somewhere in the new Boinc version, or in its ability to communicate with the MW project. Now I am not one to go back to the old version and call it a day, because if everybody did so, the problems would never get solved.
ID: 27156 · 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 27182 - Posted: 6 Jul 2009, 22:09:51 UTC - in response to Message 27156.  
Last modified: 6 Jul 2009, 22:19:51 UTC

I'm sorry if I do insist, but I really don't believe the problem resides in running 12 or 14 projects, or at least not the ones I run. MW is the only project with short deadlines, and when I get work, as I stated before, I get 2 to 4 WUs only, and I believe this is because Boinc "knows" that getting more would not allow them to finish in time. And so far that's ok. Of course, If I'd run less projects, I would get more MW WUs.
The real problem is that after completing the daily WUs for some days in a row, the project starts sending the "won't finish in time" message, although all the WUs completed in time (normally the same day, sometimes the next, but always way before the deadline). And I don't have any projects running in high priority either. After a few days I will eventually get new WUs, which will be crunched in time, only to see the ominous message appear again a few days later.


The "problem" is indeed the high number of projects... What you don't seem to be understanding about the scheduling process is that it will look across all non-suspended projects that have work on your computer and then make calculations to the effect of making sure the work can get completed by 90% of the deadline time. It takes into consideration both system uptime as well as the amount of time that BOINC is running. While I won't disagree that I have seen complaints about scheduling in newer versions, the problems you are having might be either severely exacerbated by the number of projects that your system is trying to work on, or is completely due to the high number of projects attached...IMO...

Edit: Oh, and JohnR is running tasks with a GPU. That is a problem for the current BOINC versions because they don't know about the GPU time for ATI devices. He has a different problem than what you have. Just recently the runtime of the tasks here about doubled. That's what pushed your systems over the cutoff, not the new BOINC version...

IMO, YMMV, etc, etc, etc...
ID: 27182 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Beau

Send message
Joined: 3 Jan 09
Posts: 270
Credit: 124,346
RAC: 0
Message 27191 - Posted: 7 Jul 2009, 1:21:23 UTC - in response to Message 27182.  

I only have 2 active projects (one if which is orbit@home, so does that really count?) and have started seeing this with MW since the workunits were extended without extending the deadline as well. I have also had workunits that finished thrown out because they "weren't needed" even though they were resent to someone else to be crunched. MW is definitely sliding downhill in recent months, such a shame...
ID: 27191 · 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 27198 - Posted: 7 Jul 2009, 2:43:25 UTC - in response to Message 27191.  
Last modified: 7 Jul 2009, 2:43:40 UTC

I only have 2 active projects (one if which is orbit@home, so does that really count?) and have started seeing this with MW since the workunits were extended without extending the deadline as well. I have also had workunits that finished thrown out because they "weren't needed" even though they were resent to someone else to be crunched. MW is definitely sliding downhill in recent months, such a shame...

Have you installed the optimised MW application? It will help overcome some of the issues you have mentioned.
ID: 27198 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · Next

Message boards : Number crunching : Won't finish in time

©2024 Astroinformatics Group