| Author |
Message |
|---|
|
|
|
I'm assuming that the anecdotal observations of higher failure rates on SoB WUs is almost entirely due to the substantially increased length of the compute time required for the WUs.
That said, perhaps we could make a bit of a sport on betting on wingmen?
For example, I just finished a WU (http://www.primegrid.com/workunit.php?wuid=104623308) and see that although my wingman had a head start of about 9 days, he hasn't yet submitted any results, and has less than 24 hours before he times out. So, does anyone care to bet on when I might see credit on this WU? I'm giving about a 75% chance that my wingman times out, which will probably make it at least another 11-12 days before I get credit. Anyone else care to put out a guess?
-JMP |
|
|
|
|
|
I was in a similar situation with a SoB WU, only I started and finished it first and was then was waiting on my wingman. The second wingman did time out and the WU was sent to a third and fourth, they returned it as an error and it was then sent to number 5.
A day or two later wingman 2 completed it and returned with a successful result. All smiles here at last!
So it may be that your wingman will still complete it even after time is up.
Wingman 5 is still crunching it.
http://www.primegrid.com/workunit.php?wuid=103641113
Pete.
____________
35 x 2^3587843+1 is prime! |
|
|
|
|
|
Well, looking at who my wingman was, it seems that he has a total of zero PrimeGrid credit. I have no idea if he's participated in other BOINC projects, but there's the distinct possibility that he'll never finish a single WU at PrimeGrid.
I wonder how long it'll take for the server to send the WU out to someone else...
-JMP |
|
|
 Sysadm@Nbg Volunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 5 Feb 08
Posts: 1224
ID: 18646
Credit: 877,688,558
RAC: 320,800
                      
|
I wonder how long it'll take for the server to send the WU out to someone else...
There are two things, that are to mentioned:
a) The wingman get the chance to finish the WU until deadline (perhaps he will finish your SoB-WU as his first WU ?!)
b) if the wingman does not finish the WU, any other participant had to request for SoB-work (if there are many WU and only a few participants, the queue will get smaller very slow and the resend will last longer - if there are many participants and only a small queue of workunits, the resend will be very quick)
____________
Sysadm@Nbg
my current lucky number: 113856050^65536 + 1
PSA-PRPNet-Stats-URL: http://u-g-f.de/PRPNet/
 |
|
|
|
|
|
I've never had much faith in people's ability to do things properly. I'm anticipating at least a month wait for each SoB WU I turn in before I see credit. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
                            
|
|
You really need a lot of patience with SoB. The WUs are really long (approaching CPDN-long), and the deadlines are comparatively short. That means anyone trying to do this on an older computer will miss the deadlines. (The two slower computers that I still use for crunching would probably take about 2+ months to run one of the longer SoB WUs.)
So, timeouts are to be expected from slower computers. Errors are to be expected from faster computers, where 'fast' means overclocked just a little bit too much. Given both of those situations, if instant gratification is desired, this probably isn't the project you're looking for! :)
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
On the plus side, it's probably your best bet if you want to find a new record non-Mersenne prime. Any found now should slip in at 9th place (and knock the current SoB holdout on the top-10 list into #11). |
|
|
|
|
|
Yeah I figure the majority of people running primegrid avoid this project like the plague. I feel like I'm actually accomplishing something worthwhile by running it. |
|
|
JaySend message
Joined: 27 Feb 10
Posts: 132
ID: 56067
Credit: 64,433,752
RAC: 17,531
                   
|
|
I do wonder, but haven't investigated the stats enough to know if the answer is obvious, how many people actively run this project. I have SOB set as the only project in my preferences. I haven't been here long, but I think I'm in the top 10 for SOB work, which surprises me, and makes me think it's not a popular project.
I chose this project because it has a definite end.
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
I do wonder, but haven't investigated the stats enough to know if the answer is obvious, how many people actively run this project. I have SOB set as the only project in my preferences. I haven't been here long, but I think I'm in the top 10 for SOB work, which surprises me, and makes me think it's not a popular project.
I chose this project because it has a definite end.
You may not have been here long... but neither has SoB.
SoB only started in January or February, IIRC, so there's nobody who has been running it significantly longer than you have been running it.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
A lot of people who have ran the project for a while still crunch on the SoB server using the "new" client (Prime95/mprime) and there are still quite a few folks using the old client doing double-checks as well.
With no deadlines (as long as you check in once a month or so) and double-check tests available, working off the original server is much more forgiving on older hardware. (Although I wouldn't want to even do double-checks now with much less than a decent P4) |
|
|
|
|
I do wonder, but haven't investigated the stats enough to know if the answer is obvious, how many people actively run this project. I have SOB set as the only project in my preferences. I haven't been here long, but I think I'm in the top 10 for SOB work, which surprises me, and makes me think it's not a popular project.
Free-dc provides details of the subprojects including SoB.
There are about 1429 people with credit (The Riesel Problem (LLR) has a similar number). But most have only done 1 WU - some lucky ones get a bronze badge for it! Others have done more probably to get a badge. Only 80 people have at least a silver with you likely to the 10th person with gold.
Yes it is not that popular but these things do get in the way of the challenges (plus the last cleanup took ages) and that these are taking about 14 days of solid crunching. Also people have may many pending tasks (I have 6). |
|
|
|
|
I do wonder, but haven't investigated the stats enough to know if the answer is obvious, how many people actively run this project. I have SOB set as the only project in my preferences. I haven't been here long, but I think I'm in the top 10 for SOB work, which surprises me, and makes me think it's not a popular project.
I chose this project because it has a definite end.
It occurs to me that you have some weird statistics, with SOB LRR credits actually higher (479,944) than the total credits displayed on the bottom line (479,255). |
|
|
|
|
|
I knew this was going to happen: your stats have updated and now they seem to be all right, though the totals are still 1 short. |
|
|
|
|
|
Finished my very first SoB WU on June 1st, in just a couple hours over 7 days cpu time (611000+ seconds, on a 2.93 GHz core 2 duo Mac). Now waiting on wingman. My second SoB WU is at 92 hours, 53%. Cool!
--Gary |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
I think I have a winner here.
My wingman on this WU is running SoB on a netbook or laptop with an Intel Atom N270 processor.
By my guestimate, IF he ran it --plugged in-- 24/7, it will take anywhere from 60 to 120 days, depending on the 'n' of this WU.
He's crunching a huge WU on a processor that draws about 7 watts, give or take. :)
____________
My lucky number is 75898524288+1 |
|
|
|
|
I think I have a winner here.
It reminds me of the guy who asked about crunching on his phone...
|
|
|
|
|
|
*breaks out his HP48GX calculator*
Can someone tell me how to get SoB to run on this thing? |
|
|
|
|
I do wonder, but haven't investigated the stats enough to know if the answer is obvious, how many people actively run this project. I have SOB set as the only project in my preferences. I haven't been here long, but I think I'm in the top 10 for SOB work, which surprises me, and makes me think it's not a popular project.
I chose this project because it has a definite end.
Please consider
a) other users don't run only one subproject - maybe they want to have a full set of (available) badges
b) other users additionally run other projects beside PrimeGrid - really ;-)
c) SOB grants only a medium amount of BOINC credits with respect to the WU duration, compare eg sieve subprojects
d) a disadvantage of PrimeGrid, especially at SOB: there are no trickles like in ClimatePredition.net, ie if the WU errors out after 300h you won't get any credit - a very low incentive
e) some WUs are pending for a long time, at the moment I have a WU running which my wingman returned on Apr 29 (2x time out, 3x aborted, 3x error while computing, 1x error while downloading, 1x pending) |
|
|
|
|
|
Still waiting now on three SoBs and one PSP. *looks at the calendar* Yeah this is going to take awhile. |
|
|
|
|
|
*sigh* 86k in pending credit and its almost all SOB....
____________
 |
|
|
|
|
|
Have a look at this workunit
http://www.primegrid.com/workunit.php?wuid=118649797
At the time of writing:
- 1 person has completed it (which is me)
- 2 people experienced an error
- 2 people cancelled the task
- 1 person timed out
- 1 person is still working on it, but the deadline is getting closer (only 5 days left)
What a waste of computer cycles...
The system I used for this task is quite an old one, however it was working 24/7 on this single unit. Even then I didn't make the deadline, but I was lucky to have my work returned "only" three days too late.
This was my my first and definitely my last SoB task, I consider it one of the worst projects around. I did like the native project (contributed 167.12 T jEMs), but here at PrimeGrid it's just not working. At least not with the current test sizes, short deadlines and failure rates.
At the moment I'm working on PPS LLR only, but I'm afraid I'll get new SoB assignments once no work is available. I'd prefer to idle instead of doing one of these tasks again, is there a way I can block this subproject entirely?
Maxime. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
You have several options:
1) Turn off the checkbox that allows for sending other sub-projects if your selected sub-project(s) have no work available. If you do this, you'll go idle if you run out of work.
2) Select more than one sub-project. It's less likely that 2 or more sub-projects will run out of work simultaneously. If it's a system-wide problem that's not letting you get tasks from multiple sub-projects, then you probably won't be getting any SoB tasks either.
3) Manually abort any SoB tasks should they happen to get sent to you if you tun out of work on your selected sub-project(s).
I agree that the deadlines for SoB are too short, considering the length of the WUs. Generally speaking, anything much slower than a Core 2 class CPU has almost zero chance of making the deadline even running 24/7 at 100%. I'm guessing the admins set it that way because making the deadlines even longer (e.g., 6 or 8 weeks) would make missed deadlines due to abandoned WUs even more painful than it is now. The oldest active WU in the SoB database was first sent out February 6th.
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
I think I have a winner here.
My wingman on this WU is running SoB on a netbook or laptop with an Intel Atom N270 processor.
By my guestimate, IF he ran it --plugged in-- 24/7, it will take anywhere from 60 to 120 days, depending on the 'n' of this WU.
He's crunching a huge WU on a processor that draws about 7 watts, give or take. :)
Checking in THREE MONTHS and seven wingmen later, and that WU is still pending. :)
The Atom processor timed out two months ago, but it's possible it's still crunching and may finish any day now....
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
I think I have a winner here.
My wingman on this WU is running SoB on a netbook or laptop with an Intel Atom N270 processor.
By my guestimate, IF he ran it --plugged in-- 24/7, it will take anywhere from 60 to 120 days, depending on the 'n' of this WU.
He's crunching a huge WU on a processor that draws about 7 watts, give or take. :)
And, finally, 8 wingmen and three and a half months later, someone else completed this WU and I have my silver SoB badge.
The Atom will probably be crunching until Christmas Not necessarily Christmas of this year. :)
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Question. Is there a way to request only 1 WU? After reading some of the truly funny comments here on wing men, I was thinking of trying out 1 WU on one of my faster systems just for grins and giggles. It will have to be after the Leonids challenge and only if someone can tell me how to get just one unit. |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1957
ID: 352
Credit: 6,137,901,150
RAC: 2,261,300
                                  
|
|
You can try following micro-management.
Set a low work cache, i.e. just to fill a single WU per core, preferably short ones.
Make sure you have only SoB sub-project that will give you work.
Suspend one WU; BOINC - in order to keep your cores busy - should download extra WU. If no work is downloaded, try to make work cache a bit larger...until you get one.
____________
My stats |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
                 
|
You can try following micro-management.
Set a low work cache, i.e. just to fill a single WU per core, preferably short ones.
Make sure you have only SoB sub-project that will give you work.
Suspend one WU; BOINC - in order to keep your cores busy - should download extra WU. If no work is downloaded, try to make work cache a bit larger...until you get one.
1 - Set cache to 0.0 and set cores to 1
2 - Set Preferences to SoB
3 - Update
4 - Get your 1 SoB
5 - Reset Preferences to previous (or other)
6 - Return cache and cores to normal
7 - Update
____________
Murphy (AtP)
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
You can also just abort any extras if you get more than one. It will just get resent out to someone else.
BOINC isn't real smart about allowing users to choose what they want their computers to work on, so sometimes the only way to get it right is to kill the WUs you don't want to crunch.
Otherwise, you're playing a game of chess against the BOINC client, with each of you trying to out-think the other. Unfortunately, BOINC doesn't play rationally, so you end up with it not doing what you want, and aborting stuff anyway.
What I would do if I were you? In order to not have to "be at the computer at the right time" and all that fun stuff, this is what I'd do:
1) Go to PrimeGrid's preferences and select SoB as your only subproject
2) Increase your cache setting in the BOINC manager. This will cause it to download one or more WUs.
3) Put your cache settings back to what they normally are
4) Go back to the PrimeGrid preferences, and select the sub-projects you had been running before.
5) If you got more than 1 SoB WU (which is likely), abort them in the BOINC manager.
The whole thing should take you about a minute or two to do, and doesn't interrupt the work being currently done, and so forth. Very, very easy to do.
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
It will have to be after the Leonids challenge...
A word about this. (You may already know, but others may not.)
Definitely don't start until after the challenge. You can't suspend the SoB task during the challenge in order to run SGS -- if you do this, BOINC will refuse to download anything else from PrimeGrid. So your only options would be to run SoB during the challenge, abort the in-progress SoB and loose all that processing time, or set up a Virtual Machine to run BOINC for the challenge while the real BOINC is completely suspended.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Ok now comes the newbie question. Where and how do I change the cache settings or is it disguised as something else. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
Ok now comes the newbie question. Where and how do I change the cache settings or is it disguised as something else.
First of all, if your BOINC client isn't in "Advanced" mode, it needs to be. If it's in "simple view" (where it's yellow and green and shows a huge BOINC logo, and very little useful information), just click on "Advanced View" in the lower right corner.
Once you're in advanced view, where most of the window is occupied by a large table showing all the running tasks, do the following:
1) Click on "Advanced" in the menu on the top
2) On the menu that drops down, clock on "Preferences". This opens up a dialog box where you can set lots of options.
3) There are three tabs in this dialog box, select the middle one, "Network Usage"
4) Three lines down, on the right, is a setting labeled additional work buffer". It's probably 0.00. Set it higher to increase your cache size. That number is in days, so making it 1.00 will cause BOINC to download an extra days worth of work. Be carefull not to set it too high -- if you download, say, 5 days worth of work and the WUs have three day deadlines, very interesting things are going to happen.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Yep, I use advance mode and even saw the settings you referred to in the network tab. However, not knowing what might happen to BOINC I've made very few changes. I did however make the change you just explained on one of my pc's from .25 to .75 and it really loaded up some WUs.
Thanks Michael for that very useful information !!
ps - now I see how some of the power users work :) |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
I typically run with a 6 hour cache. That's usually long enough to keep the computers from running dry if there's a problem at PrimeGrid, or the BOINC client decides to do a one-hour backoff. At the same time, it's not so long that it takes forever to switch projects if I decide to do so.
On projects with notoriously bad downtime, and generally long deadlines, people will keep many days worth of WUs on hand. SETI is like this, in particular, and that was before they decided to only run a few days a week, not to mention taking a month off for renovations. It's got to be a real mess there now trying to keep your computers well-fed with WUs.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
142459984
142451396
142451382
142437245
Here's the SOB's I'm "trying" to run. It looks like the pc I have assigned is not fast enough to do them in time but we'll see. Wasn't sure if anyone was waiting on these or not.
(nothing like waiting for days on a task to complete) |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
142459984
142451396
142451382
142437245
Here's the SOB's I'm "trying" to run. It looks like the pc I have assigned is not fast enough to do them in time but we'll see. Wasn't sure if anyone was waiting on these or not.
(nothing like waiting for days on a task to complete)
Most of us who do SoBs are used to waiting months! Keep on running, even if over deadline (unless, of course, just started. or not yet started).
____________
Murphy (AtP)
|
|
|
|
|
|
I'm currently working on this SoB workuint:
http://www.primegrid.com/workunit.php?wuid=141379325
(I'm computer 160428.)
My computer is not overclocked and I've never had any problems with stability before, so I expect my half of the workunit to validate without any problems when the results come in. But I'm not so sure I can say the same for my wingman--despite having successfully completed two SoB workunits a month ago, he had a whole bunch of "completed, marked as invalid" SGS tasks during the recent rally. If he's getting failures often enough to invalidate that many instances of such small tasks, he must be seriously unstable. This is borne out further by his having failed validation on two other long-running WUs in October, on the Cullen and PSP subprojects respectively.
Anyone care to lay odds as to whether this guy will pass validation on my workunit? :-D |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
Is your wingman's computer a 32-bit Mac by any chance? Were those invalid WUs completing almost immediately after just a few seconds? Were they all SGS?
If I remember correctly, there was a bug in the SGS-LLR app for 32-bit Macs that caused all WUs to complete in a few seconds. It was a problem with the software, not the hardware. It shouldn't affect SoB.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Seems like I remember some of you mentioning checking some place for errors on jobs that are currently running. I was just wondering if I could check and see if I have very many errors on those SoB's I'm running and if so where would I look for that info.
TAP |
|
|
|
|
Is your wingman's computer a 32-bit Mac by any chance? Were those invalid WUs completing almost immediately after just a few seconds? Were they all SGS?
If I remember correctly, there was a bug in the SGS-LLR app for 32-bit Macs that caused all WUs to complete in a few seconds. It was a problem with the software, not the hardware. It shouldn't affect SoB.
No, the computer appears to be a 32-bit Linux box. All of his problem workunits finished in the usual amount of time, and as I mentioned earlier some of them were from the Cullen and PSP subprojects. |
|
|
|
|
|
I had to stop on of the SOB's but have 3 still running... boy these take a long time to run. I sure hope that when they finish they help someone!
208 hrs and 44 left
207 hrs and 129 left
207 hrs and 121 left |
|
|
|
|
I'm currently working on this SoB workuint:
http://www.primegrid.com/workunit.php?wuid=141379325
(I'm computer 160428.)
My computer is not overclocked and I've never had any problems with stability before, so I expect my half of the workunit to validate without any problems when the results come in. But I'm not so sure I can say the same for my wingman--despite having successfully completed two SoB workunits a month ago, he had a whole bunch of "completed, marked as invalid" SGS tasks during the recent rally. If he's getting failures often enough to invalidate that many instances of such small tasks, he must be seriously unstable. This is borne out further by his having failed validation on two other long-running WUs in October, on the Cullen and PSP subprojects respectively.
Anyone care to lay odds as to whether this guy will pass validation on my workunit? :-D
Well, the results are in: my wingman submitted his result yesterday, and I today. The upshot: "Validation inconclusive". Considering my wingman's track record with past workunits (compared to mine which is 100% stable so far) it seems likely that his result is the bad one.
Now the (first, possibly not the last) re-run begins... :-) |
|
|
|
|
|
I just noticed the available wu's dropped back to 2 from 4 earlier in the day. I checked my 3 running wu's and they all appear to be first run jobs (if that makes sense). I would've figured that incomplete or invalid jobs come first but apparently not.
So here's another question. Why are the users listed as anonymous ? Is that due to the use of the app_info file? |
|
|
|
|
I just noticed the available wu's dropped back to 2 from 4 earlier in the day. I checked my 3 running wu's and they all appear to be first run jobs (if that makes sense). I would've figured that incomplete or invalid jobs come first but apparently not.
The "Available: Seventeen or Bust (LLR)" counter represents how many tasks are currently available on the server to be handed out to clients. Resends (due to timeouts, errors, nonmatching results, etc.) are added to this count shortly after they are initially recognized as necessary to the validator; but since the demand for new workunits is greater than the number of resends typically available, the queue is topped off to 100 with fresh workunits (i.e., first run jobs) every hour. (Sometimes you'll see it greater than 100; that means that some resends became available recently after it was topped off.)
As I understand it, if there are resends in the queue at any given moment, those will be handed out to clients first before the fresh workunits are broken into. If you got a fresh workunit, then there were no resends available at that particular moment. (This is not uncommon; though SoB has a higher resend rate than the other subprojects due to its long workunits erroring out more frequently, there is still a fair amount of new work processed. If there wasn't, no progress would be made whatsoever. :-))
So here's another question. Why are the users listed as anonymous ? Is that due to the use of the app_info file?
Well, this can be a little confusing since BOINC uses the term "anonymous" in two different contexts to refer to two different things. The more common one, which I believe you mean to refer to here, is when a computer is listed as belonging to participant "Anonymous". There is an option in your PrimeGrid preferences (see Your Account), "Should PrimeGrid show your computers on its web site?" If this is set to no, your computers will show up as belonging to "Anonymous" to everyone but you, and if someone visits your account page it will say "hidden" where it would otherwise link to a list of your computers.
The other context is when BOINC refers to an "anonymous platform". This happens when someone uses an app_info.xml file to tell BOINC to use their own executables for the project applications. Some participants do this to allow them to use experimental versions of applications on their GPUs; others to use the latest version of LLR, which is a good 15-20% faster than the previous version but hasn't been rolled out into BOINC yet (except for SGS LLR on Linux and Mac). BOINC refers to this as "anonymous platform" since its default applications for the computer's recognized platform have been overridden. |
|
|
|
|
|
That all makes perfect sense. Thanks for the explanation. I unchecked my show computer flag just for fun.
Rick |
|
|
|
|
|
I finally finished my first sob after some 250+ hours. WU is 142451396 and it appears to have finished correctly. I have 2 more running and it looks like I have the same person running all 3 and the one I just finished is being run on a fast pc..
In order to keep my pc running the sob all the time I upped my time to switch between apps to 288000 minutes. That wouldn't hurt anything would it. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
In order to keep my pc running the sob all the time I upped my time to switch between apps to 288000 minutes. That wouldn't hurt anything would it.
Nope, it won't hurt anything. It should do exactly what I think you want it to do: keep running the tasks that are already running until they finish (or until a not-running task gets into deadline trouble.)
And if you're only running PG, it has no effect at all. It's only used for switching between projects.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Thanks Michael for the info ! |
|
|
|
|
|
I was looking at my 3 sob's and noticed that Lumiukko and I are doing the same 3 wu's (which I found to be very interesting). I've finished 1 and he's finished 1 (now waiting on each other) and we now have a race for the 3rd to see who's the initial. I'm taking bets on him winning based on his speed. |
|
|
|
|
|
Apparently SoB is not the only project where having a good wing man is nice. 321 is following a close 2nd.
____________
@AggieThePew
 |
|
|
|
|
Well, the results are in: my wingman submitted his result yesterday, and I today. The upshot: "Validation inconclusive". Considering my wingman's track record with past workunits (compared to mine which is 100% stable so far) it seems likely that his result is the bad one.
Now the (first, possibly not the last) re-run begins... :-)
The _2 run came in today and, as I expected, my result was validated and the original wingman's result was bad.
Now I just have to wait for the next SoB task I did to validate...then I'll have a bronze badge. :-) (That workunit, however, is not really worth analyzing here...my wingman has a pretty good track record.) |
|
|
mackerelVolunteer tester
Send message
Joined: 2 Oct 08
Posts: 2645
ID: 29980
Credit: 568,565,361
RAC: 266
                    
|
|
Ooh, looks like I ended up with a SOB unit from "send any work if none available" as I haven't selected SOB. At least I'll be helping someone out with credit, as it's a _4 unit...
Only 10 days to go... |
|
|
|
|
|
I guess I was EXTREMELY lucky with the wu's I got. I would start some more but since the challenge is this week and I need to use what CPU's I have to run the sieves I will wait again.
What's the difference between the _2 and the _4 units?
____________
@AggieThePew
|
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 850
ID: 18447
Credit: 712,577,373
RAC: 1,631,637
             
|
|
The _<number> is just the number of times the same WU has been sent out.
_0 and _1 will be the initial tasks.
_2 will be the first resend, _3 the second resend, _4 the third resend, etc.
____________
 |
|
|
|
|
|
Well, learn something new everyday! thanks much
____________
@AggieThePew
|
|
|
|
|
|
Back on the little SOB jobs... looks like I'm the wing man for Christoph and Lumiukko.
In about 175 hours (if no errors occur) they should be getting their credits.
____________
@AggieThePew
|
|
|
|
|
|
okay, I am waiting for my wingman to double check:
http://www.primegrid.com/workunit.php?wuid=142470456
although it has been awhile and I am somewhat beginning to think it might not happen, the WU tells it all. |
|
|
|
|
okay, I am waiting for my wingman to double check:
http://www.primegrid.com/workunit.php?wuid=142470456
although it has been awhile and I am somewhat beginning to think it might not happen, the WU tells it all.
Well from what I can see, it would appear you have a good wingman there.
Here's to you and your wing man :)
____________
@AggieThePew
|
|
|
|
|
Well from what I can see, it would appear you have a good wingman there.
Here's to you and your wing man :)
Thanks, just out of curiosity I estimate that the Wu should be done in ~8 days or less. It just my guess |
|
|
|
|
|
Since I have my gold in all the LLR except SOB, I decided to fill my 23 cores with them. It's going to be a great test on all the machines. I started them about the same time right after the challenge. So I am sure going to be helping out on a few, and some I was the first to download. I hope all turn out well and I get my gold.
It's interesting to watch the speed of these. I have some over 35% and some still under 20% after just over 2 days of crunching, and those speeds on the same machine.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 |
|
|
mackerelVolunteer tester
Send message
Joined: 2 Oct 08
Posts: 2645
ID: 29980
Credit: 568,565,361
RAC: 266
|
|
After 298 hours (12.4 days) I just completed my 3rd ever SoB unit in the primegrid era. So now it is my turn to wait on the wingman...
What's a typical credit for one of these now? Two units got me bronze last time but that was a while ago... SoB is the last active project I need to silver+ badge in (PSP has plenty pending). |
|
|
|
|
|
From about 8K to 13K. At least that what i received for my SoB.
____________
Polish National Team |
|
|
|
|
|
After being gone for a week and just getting back in, I just noticed my 2 SOB's are done. Took about 278 hours to run and one is verified.. got 12,248 points and am waiting on my other to be double checked. Looks like that might be a while since it's been bounced around some.
____________
@AggieThePew
|
|
|
|
|
|
http://www.primegrid.com/workunit.php?wuid=139990586
Looks like this poor wu of mine has been around the block and back. It's now on it's 7th trip.
Hey Pooh Bear how are all of your 23 wu's running? Now that's a lot of determination :)
____________
@AggieThePew
|
|
|
|
|
Hey Pooh Bear how are all of your 23 wu's running? Now that's a lot of determination :)
11 remaining to be reported in the next 3 days.
Of the 12 already reported, only 4 are pending. 1 has a inconclusive the other 3 are just waiting for wingmen to finish.
Now I am working on gold for 321 sieve and CW sieve.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 |
|
|
|
|
Of the 12 already reported, only 4 are pending. 1 has a inconclusive the other 3 are just waiting for wingmen to finish.
Now I am working on gold for 321 sieve and CW sieve.
That's impressive on many levels.
____________
@AggieThePew
|
|
|
|
|
|
Just had this workunit validate today. I was actually rather surprised that it came this earlier, as I expected my wingman to not make the deadline (by all appearances, he seemed to have abandoned the WU). But, as it turned out, he came through and--because of the way the claimed credit worked out--I ended up getting a nice fat 10,103 credits for this WU. That would have been enough to get a bronze badge in and of itself...ironically, the entire reason why I did this workunit was to finish off the badge which was not completed by my first SoB task (which only got 6,531 credits). If they'd been in reverse order I would have only needed to do one. :-) Oh well, SoB could definitely use the help anyway. |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
On to Silver (100K)!
I've been trying to talk myself into doing that since I had a similar experience on SoB last year.
____________
Murphy (AtP)
|
|
|
|
|
That's impressive on many levels.
1 more got a wingman to finish, so I have 3 pending of the first 12.
The other 11 will be reported between 16 and 38 hours from now. I cannot wait to see how well these stats go. I will be well over gold once approximately 9 more are validated.
I find it interesting that 12 got done in 12-14 days then the other 11 take 16-18 days. Many of the 12 that were done got the max 13K and I bet some of the longer ones will get less than the max.
Oh well, it was nice to see my machines are all stable to handle these massive LLR tasks all simultaneously.
Keep the dust bunnies at bay, and good ventilation around your devices. |
|
|
|
|
|
20 reported, 3 pending (1 still is the inconclusive). The other 3 should be reported in the next 12 hours or less. The next validated one should give me gold. Of the 3 left, I am a wingman on 1 and the other 2 I should be the primary awaiting wingmen.
What a thrill this has been. |
|
|
|
|
|
All 23 reported, no errors. 6 are awaiting wingmen (1 inconclusive) and I got all the LLR gold.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1957
ID: 352
Credit: 6,137,901,150
RAC: 2,261,300
|
|
What an accomplishement.
I need to still go down the road to get all gold+ from all LLRs... but LLRs will be there for a looong time.
____________
My stats |
|
|
|
|
|
http://www.primegrid.com/result.php?resultid=215335382
Wingman is Pasha on this one. Looks like a good wingman too.
Congrats Pooh Bear!!!
____________
@AggieThePew
|
|
|
|
|
|
Got my first 'Seventeen or Bust' work unit with an estimated run time of 167 hours (that remains to be seen).
I have only been crunching PrimeGrid since the 14th (1 Day) so only have 1 badge at the moment but I am working on them all.
It was first sent out on the 5/12/10 and has since been aborted once and 3 work units had memory issues.
The difference this time is I am the first Linux host as all the others are Windows so I hope when finished that the different operating systems will validate with each other.
Keen to see how it will run, these run times remind me of AQUA and Climate Prediction with 300 hour plus work units.
Keep crunching and we will be a 'Prime' example to all the others.
Conan
____________
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
|
There should not be any problems with results from different platforms validating.
____________
My lucky number is 75898524288+1 |
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
     
|
Got my first 'Seventeen or Bust' work unit with an estimated run time of 167 hours (that remains to be seen).
Conan
I've got 40 SOB Wu's running @ some Stage of Completion from 5% up to 60%, hope they all finish without incident ... :)
____________
 |
|
|
|
|
|
41 running at the moment. Most of them should finish during the weekend.
____________
There are only 10 kinds of people - those who understand binary and those who don't
 |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
I decided to crank up 4 SoBs, should take about 16 days on my machines.
____________
Murphy (AtP)
|
|
|
|
|
I decided to crank up 4 SoBs, should take about 16 days on my machines.
Hi....
how to get only 4 WU on this project ???
Would like to do the same...
Any special settings...
cheers |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I decided to crank up 4 SoBs, should take about 16 days on my machines.
Hi....
how to get only 4 WU on this project ???
Would like to do the same...
Any special settings...
cheers
Set No New Tasks ... allow all PG to flush ... set cache to something small (say 1 day) ... select SoB in Preferences ... set Allow New Tasks ... Update ... get 1 SoB per core ... set No New Tasks ... (repeat as needed per computer) ...
(run boinc 6.10.58 -- not one of the older ones and not one of the experimentals)
____________
Murphy (AtP)
|
|
|
|
|
I decided to crank up 4 SoBs, should take about 16 days on my machines.
Hi....
how to get only 4 WU on this project ???
Would like to do the same...
Any special settings...
cheers
Set No New Tasks ... allow all PG to flush ... set cache to something small (say 1 day) ... select SoB in Preferences ... set Allow New Tasks ... Update ... get 1 SoB per core ... set No New Tasks ... (repeat as needed per computer) ...
(run boinc 6.10.58 -- not one of the older ones and not one of the experimentals)
I find it important to set aside one of the PG locations for this so only the desired host will get the work but only when you want it to. It is also important to do one host at a time. Once done you then reset that location and hosts. If you do not want anymore SoBs, don't forget to change the location preferences as soon as you have get the WUs you need!
My approach is similar but I first run the your PG cache to near zero especially by using short subprojects. Then in network preferences I set the 'additional work buffer' to zero. This will force boinc to get new tasks only when it needs them. So a finished WU will be replaced by a new SoB WU. This works best when you have WUs at different stages of completion. You can also slowly increase the additional work buffer to get what you need - it helps to get the new SoB WUs running immediately by suspending other tasks.
Regardless, don't forget to reset the PG location before resetting your additional work buffer and reset your additional work buffer before resuming your regular projects etc.
|
|
|
|
|
http://www.primegrid.com/result.php?resultid=215335382
Wingman is Pasha on this one. Looks like a good wingman too.
Well it looks like this wu must be a tough one. Last wing man got it to complete but inconclusive so now it's back on the trend mill.
No more sob's until after the first challenge.
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I decided to crank up 4 SoBs, should take about 16 days on my machines.
Hi....
how to get only 4 WU on this project ???
Would like to do the same...
Any special settings...
cheers
Set No New Tasks ... allow all PG to flush ... set cache to something small (say 1 day) ... select SoB in Preferences ... set Allow New Tasks ... Update ... get 1 SoB per core ... set No New Tasks ... (repeat as needed per computer) ...
(run boinc 6.10.58 -- not one of the older ones and not one of the experimentals)
I find it important to set aside one of the PG locations for this so only the desired host will get the work but only when you want it to. It is also important to do one host at a time. Once done you then reset that location and hosts. If you do not want anymore SoBs, don't forget to change the location preferences as soon as you have get the WUs you need!
My approach is similar but I first run the your PG cache to near zero especially by using short subprojects. Then in network preferences I set the 'additional work buffer' to zero. This will force boinc to get new tasks only when it needs them. So a finished WU will be replaced by a new SoB WU. This works best when you have WUs at different stages of completion. You can also slowly increase the additional work buffer to get what you need - it helps to get the new SoB WUs running immediately by suspending other tasks.
Regardless, don't forget to reset the PG location before resetting your additional work buffer and reset your additional work buffer before resuming your regular projects etc.
You can also modify Preferences ... use at most x% processors to get WU for less that the cores available -- if 4 core cpu, set to 25% to just feed 1 core (assuming all of the other suggestions have been followed). Just be sure to set everything back to "normal" in the proper sequence BEFORE allowing any new tasks!
____________
Murphy (AtP)
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 21.75% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 26.44% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 26.26% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 30.76% done
Maybe this info will be useful/encouraging to my wing persons.
Slow, but steady (and I hope they all finish before Feb 3 ;-).
____________
Murphy (AtP)
|
|
|
|
|
I find it important to set aside one of the PG locations for this so only the desired host will get the work but only when you want it to.
Having said that, don't forget as I have just done (ya idiot), that there may be other hosts with the same location. Somehow, it got 4 WUs but I will let them burn and see if they prove me wrong about windows hosts. |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 24.95% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 30.51% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 30.23% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 35.11% done
Maybe this info will be useful/encouraging to my wing persons.
Slow, but steady (and I hope they all finish before Feb 3 ;-).
Updated: 01/26/2011 17:23 UTC
____________
Murphy (AtP)
|
|
|
|
|
|
Looks like it's gonna be a race to finish and Feb 3rd :)
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
Looks like it's gonna be a race to finish and Feb 3rd :)
Currently, Feb 3 looks to be winning! ;-)
____________
Murphy (AtP)
|
|
|
|
|
|
Looks like the SoBs I'm running will be done at or around Feb 5th. Luckily I've only got two so I'll have two cores open for (at least the beginning of) the challenge.
Dunno about the other people who've completed these tasks (since I am the wingman in this case - 4th and 5th resends) but if they're out there, they should know I'm running a very stable Darwin i5 system that runs 24/67...hence the guess of Feb. 5th. Currently they're at 15.666% and 14.678% after 40h45m of runtime. Since I'm in class most of the time, this will seem to pass much faster...!
____________
 |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 30.39% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 36.86% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 36.59% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 42.64% done
Updated: 01/27/2011 22:40 UTC
____________
Murphy (AtP)
|
|
|
|
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 30.39% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 36.86% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 36.59% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 42.64% done
Updated: 01/27/2011 22:40 UTC
Gonna go out on a limb here and say I put my money on Feb 3 for all 4 finishing. It looks like you might get 1, maybe 2 done before then. So I suppose you are gonna let them continue right?
____________
@AggieThePew
|
|
|
|
|
Looks like the SoBs I'm running will be done at or around Feb 5th. Luckily I've only got two so I'll have two cores open for (at least the beginning of) the challenge.
At least with Intel processors LLR's do not like hyperthreading due to design restrictions (not sure if sandybridge is better). So I find that running just LLR is done by setting boinc to have the number of physical cores as if there was no hyperthreads (but that is project and even PG subproject dependent).
If you have turboboost or equivalent then you can use it if and only if you need to finish a WU faster. It is hard to see with SoBs, but I have used it successfully to get a WU done before the challenge deadline. There are still enough days to go for this to work so just suspend all CPU tasks except one per physical processor for maximum benefit. Wikipedia's entry has an example of 4 core 2 GHz processor going up to 3.2GHz for one task and 3.06GHz for two tasks. So it could finish a SoB 2 to 3 days earlier than without it - of course the other SoBs will be sitting there.
If you do not care about finding primes (you will probably be double checker if you do this), then you apply some micromanagement to get all the cores for the 3 days of the challenge. First make sure you only can get challenge WUs then get a reasonable number of WUs and then suspend the SoBs. Some relatively time after the challenge starts the server load should be slightly lower. Unsuspend the SoBs, increase your cache up to three days and then suspend SoBs. There is likely to be limits on the number of WUs so you will probably have to do this many times. Also do not set boinc to return the WUs as soon as they are finished as it will slow the server and your hosts down. Once you are done with the challenge do not forget to first reset your additional work buffer and reset your PG setttings before resuming. |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 32.75% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 39.94% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 39.92% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 46.39% done
Updated: 01/28/2011 17:15 UTC
Gonna go out on a limb here and say I put my money on Feb 3 for all 4 finishing. It looks like you might get 1, maybe 2 done before then. So I suppose you are gonna let them continue right?
Yes, but when the challange starts I will micromanage to get WUs then suspend SoB for the duration.
____________
Murphy (AtP)
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 42.48% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 51.959% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 51.349% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 60.16% done
Updated: 01/30/2011 19:14 UTC
Apologies to my wingmen, but it looks like these will not be finished in time for the Challenge; hence, they will be delayed at least 3 days.
____________
Murphy (AtP)
|
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
 
|
Updated: 01/30/2011 19:14 UTC
Apologies to my wingmen, but it looks like these will not be finished in time for the Challenge; hence, they will be delayed at least 3 days.
Does this mean you discard four units???
____________
Best wishes. Knowledge is power. by jjwhalen
 |
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
|
I think he's just going to Suspend them until after the Feb 3 Challenge, I''l have some to Suspend myself if I decide to rn the Challenge, right now I only plan on running it minimally so I probably can let the SOB's I still have running continue to run ...
____________
|
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
|
|
Depends of your used client, suspend shouldn't work. My last try of suspending one WU on a single core host was a miss. I got no further units from the same project and my backup project got calculation time...
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
|
Jack your Contact time up, get the work you want and then Suspend the SOB Wu's. Rinse & Repeat if you need more, takes a little bit of effort but it can be done ... ;)
____________
|
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
|
Jack your Contact time up, get the work you want and then Suspend the SOB Wu's. Rinse & Repeat if you need more, takes a little bit of effort but it can be done ... ;)
Cool, thanks for the hint...
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
Jack your Contact time up, get the work you want and then Suspend the SOB Wu's. Rinse & Repeat if you need more, takes a little bit of effort but it can be done ... ;)
Cool, thanks for the hint...
Don't forget though that if you Jack your Contact time up it will propagate across all your Projects so you have to set them to NNW or you'll get a ton of work from them too.
____________
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
The above is exactly how I do it. If any of the SoBs are just a few hours from finishing I will let them run, otherwise, I set cache to 10 days, (PG is only project allowed to get tasks) grab some PPS, suspend SOBs, run PPS, resume SOBs, grab PPS, suspend SOBs, etc. I onlt have 2 machines in this situation, so micro-managing is acceptible.
____________
Murphy (AtP)
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1957
ID: 352
Credit: 6,137,901,150
RAC: 2,261,300
|
The above is exactly how I do it. If any of the SoBs are just a few hours from finishing I will let them run, otherwise, I set cache to 10 days, (PG is only project allowed to get tasks) grab some PPS, suspend SOBs, run PPS, resume SOBs, grab PPS, suspend SOBs, etc. I onlt have 2 machines in this situation, so micro-managing is acceptible.
Perhaps well known but...you can use eFMer BoincTasks to monitor/manage several hosts from one place. Good replacement for my favourity BOINCView.
____________
My stats |
|
|
|
|
|
^ That's a neat idea - I simply use TeamViewer to remotely access/control the entire computer, since half of my machines are a hundred miles away...
Somewhat more on-topic, it seems that I grossly miscalculated the date by which my two SoBs would be done, as I'm on a mobile processor. On 24/7 for 123h35m at this point, the tasks are at 47.289% and 44.262%.
So, it would look like I'll be doing something like Murphy here. Although I don't like suspending WUs, as I've heard it can lead to errors, something about that sounds a little weird. I've been doing about 150 PPS LLR tasks a day on top of the two SoBs, so the challenge won't be a problem.
____________
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
I have never had a problem due to suspending a task; however, I am careful to wait to suspend until just after a checkpoint (unless I have to do an emergency shut-down).
____________
Murphy (AtP)
|
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 850
ID: 18447
Credit: 712,577,373
RAC: 1,631,637
|
|
If your machine is on 24/7 (and maybe even if it isn't), and you have enough RAM, make sure you have "leave applications in memory when suspended" checked in the boinc client. This avoids many potential errors around shutdown/restart/checkpoint/etc.
____________
|
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
|
Had 1 error overnight, 800,000 Seconds down the drain, oh well I still have 9 Pending, 11 old ones still running & I started up 24 more new ones.
____________
|
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
The above is exactly how I do it. If any of the SoBs are just a few hours from finishing I will let them run, otherwise, I set cache to 10 days, (PG is only project allowed to get tasks) grab some PPS, suspend SOBs, run PPS, resume SOBs, grab PPS, suspend SOBs, etc. I onlt have 2 machines in this situation, so micro-managing is acceptible.
Perhaps well known but...you can use eFMer BoincTasks to monitor/manage several hosts from one place. Good replacement for my favourity BOINCView.
I use them Both, each has it's own good Features that the other may not have so they can both be used at times depending on what you want to do ...
____________
|
|
|
|
|
The above is exactly how I do it. If any of the SoBs are just a few hours from finishing I will let them run, otherwise, I set cache to 10 days, (PG is only project allowed to get tasks) grab some PPS, suspend SOBs, run PPS, resume SOBs, grab PPS, suspend SOBs, etc. I onlt have 2 machines in this situation, so micro-managing is acceptible.
Perhaps well known but...you can use eFMer BoincTasks to monitor/manage several hosts from one place. Good replacement for my favourity BOINCView.
I use them Both, each has it's own good Features that the other may not have so they can both be used at times depending on what you want to do ...
Don't suppose either you or Honza know of the same program or like program that runs under Windows 2k? The eFMer is xp or later only but it sure did look interesting.
____________
@AggieThePew
|
|
|
mackerelVolunteer tester
Send message
Joined: 2 Oct 08
Posts: 2645
ID: 29980
Credit: 568,565,361
RAC: 266
|
|
It took a while but 6 units have finally gone through on the X6. It looks like two validated instantly as they were high number resends. Now I have 4 others to wait on before I (hopefully) get SoB silver badge. |
|
|
STE\/EVolunteer tester
Send message
Joined: 10 Aug 05
Posts: 573
ID: 103
Credit: 3,659,101,651
RAC: 422,966
|
|
Nope, not me, never used 2k, I have an Illegal Copy with Key of it that somebody sent me but never used it & don't see using it now anymore since I have Legal Copy's of XP Pro 32 & 64-Bit & Win7 ... Sorry
____________
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1957
ID: 352
Credit: 6,137,901,150
RAC: 2,261,300
|
Don't suppose either you or Honza know of the same program or like program that runs under Windows 2k? The eFMer is xp or later only but it sure did look interesting.
Not using W2K for years (since Win XP and W2K3) but I would try BOINCView.
You can download latest build from http://www.boinc.cz/download/boincview.zip
UCB has older build, which may come as benefit for older OSes. You may also try other add-ons there - http://boinc.berkeley.edu/alpha/download_network.php
____________
My stats |
|
|
|
|
Don't suppose either you or Honza know of the same program or like program that runs under Windows 2k? The eFMer is xp or later only but it sure did look interesting.
Not using W2K for years (since Win XP and W2K3) but I would try BOINCView.
You can download latest build from http://www.boinc.cz/download/boincview.zip
UCB has older build, which may come as benefit for older OSes. You may also try other add-ons there - http://boinc.berkeley.edu/alpha/download_network.php
Great will try them and thanks for the links
____________
@AggieThePew
|
|
|
|
|
|
Well hopefully the volunteer Randall will be happy in the next few days as I hope to return a Seventeen or Bust WU that has been kicking around since the 15th October 2010. He returned it in late October and has been waiting for credit ever since.
Number of WU's sent out is now 14.
____________
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
I am working on the following SoB WUs:
http://www.primegrid.com/workunit.php?wuid=146165663 = 52.64% done
http://www.primegrid.com/workunit.php?wuid=137238626 = 65.01% done
http://www.primegrid.com/workunit.php?wuid=146165665 = 64.02% done
http://www.primegrid.com/workunit.php?wuid=137513397 = 74.32% done
Updated: 02/01/2011 23:22 UTC for the benefit(?) of my wingmen.
____________
Murphy (AtP)
|
|
|
|
|
http://www.primegrid.com/workunit.php?wuid=146165663 = 52.64% done
Updated: 02/01/2011 23:22 UTC for the benefit(?) of my wingmen.
Seems the first one has already validated--looks like an over-deadline task that came in anyway. Now the question: abort or finish? |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
http://www.primegrid.com/workunit.php?wuid=146165663 = 52.64% done
Updated: 02/01/2011 23:22 UTC for the benefit(?) of my wingmen.
Seems the first one has already validated--looks like an over-deadline task that came in anyway. Now the question: abort or finish?
I could use the points, if I was sure that I would still get them. Still 4 days to finish and I do not know how long the DB stays "open". Painful as it is I maybe should abort...
____________
Murphy (AtP)
|
|
|
|
|
|
Due date is 2 March then it times out, so at least that long. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
http://www.primegrid.com/workunit.php?wuid=146165663 = 52.64% done
Updated: 02/01/2011 23:22 UTC for the benefit(?) of my wingmen.
Seems the first one has already validated--looks like an over-deadline task that came in anyway. Now the question: abort or finish?
I could use the points, if I was sure that I would still get them. Still 4 days to finish and I do not know how long the DB stays "open". Painful as it is I maybe should abort...
Barring some sort of catastrophic server crash, or a bug in the server software, the WU will never close while there are still tasks outstanding. Your result isn't due until March 2nd, so that's the earliest that this WU can be purged. It may take longer than that. You never have to worry about not getting credit as long as you get your result in by the deadline.
Depending on circumstances, you may be able to miss the deadline (sometimes by months!) and still get credit, but if you're in before the deadline, there's never anything to worry about.
The only question you have to consider is whether you want to continue processing a result that has no real value anymore, and move on to something else. But in terms of credit, you're fine.
____________
My lucky number is 75898524288+1 |
|
|
|
|
http://www.primegrid.com/workunit.php?wuid=146165663 = 52.64% done
Updated: 02/01/2011 23:22 UTC for the benefit(?) of my wingmen.
Seems the first one has already validated--looks like an over-deadline task that came in anyway. Now the question: abort or finish?
I could use the points, if I was sure that I would still get them. Still 4 days to finish and I do not know how long the DB stays "open". Painful as it is I maybe should abort...
HA now you have both a moral and ethical dilemma staring you in the face. Since YOU are the wingman on all 4 wu's and at least one of the wu's is a couple of months old. I'm just glad it's not me having to decide since I'm pretty sure that your system(s) are stable and you'd finish the wu's no problem.
LOL am I pouring it on thick or what :)
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
Thanks for the feedback guys. I think I will let it keep crunching (3 more days) for the 8000 points. The other 3 were never a consideration to abort.
____________
Murphy (AtP)
|
|
|
|
|
Thanks for the feedback guys. I think I will let it keep crunching (3 more days) for the 8000 points. The other 3 were never a consideration to abort.
If you suspend them for the 3 day challenge, that won't hurt will it? I've never done it but was just wondering.
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
Thanks for the feedback guys. I think I will let it keep crunching (3 more days) for the 8000 points. The other 3 were never a consideration to abort.
If you suspend them for the 3 day challenge, that won't hurt will it? I've never done it but was just wondering.
I have never had a problem with suspending. Although all PG must be running in order to download new WUs, I suspend SoB and let the PPS run.
____________
Murphy (AtP)
|
|
|
|
|
|
So, for those of you who had to ride out the little storm of pps's how did your sob's survive?
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
|
All 4 crunching away:
63.7% ~168 hrs to go
77.5% ~84 hrs to go
78.5% ~80hrs to go
89.5% ~36hrs to go (? Boinc estimates usually suspect)
____________
Murphy (AtP)
|
|
|
|
|
|
I had mine crunch through the challenge and they both validated pretty fast, considering they were at 5 and 6 replications respectively.
One took over a million CPU seconds...I suppose that's a sign my system is stable.
Now I'll test the Celeron! (seriously, would anyone bet on that?)
____________
|
|
|
ich_eben*Volunteer tester
Send message
Joined: 14 Oct 10
Posts: 101
ID: 70063
Credit: 392,240,946
RAC: 101,167
      
|
|
Still waiting for validation of this task: http://www.primegrid.com/workunit.php?wuid=142547875
Other Wingman timed out and the resend to this machine looks not that good either.
Also started two new SoBs on my Laptop running a 64Bit Ubuntu 10.10
Lets see how it handles these two tasks:
http://www.primegrid.com/workunit.php?wuid=162197677
http://www.primegrid.com/workunit.php?wuid=147577344 |
|
|
|
|
|
All the ones I did, I still have 1 left that needs the wingman to finish. Everything has validated.
I am getting close to my 321 Sieve Gold. Working on CW Sieve and then TRP Sieve next. |
|
|
|
|
Still waiting for validation of this task: http://www.primegrid.com/workunit.php?wuid=142547875
Other Wingman timed out and the resend to this machine looks not that good either.
hmmmmmm no sticks or stones but that looks like a really slow pc running your wingman :(
____________
@AggieThePew
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 20,784,268
RAC: 984
|
All 4 crunching away:
63.7% ~168 hrs to go
77.5% ~84 hrs to go
78.5% ~80hrs to go
89.5% ~36hrs to go (? Boinc estimates usually suspect)
http://www.primegrid.com/workunit.php?wuid=137513397 72.0% ~ 140 hrs to go
http://www.primegrid.com/workunit.php?wuid=137238626 89.4% ~ 39.5 hrs to go
http://www.primegrid.com/workunit.php?wuid=146165665 89.1% ~ 41 hrs to go
http://www.primegrid.com/workunit.php?wuid=146165663 (done & reported)
Doing my part to help out the Validator!
;-)
____________
Murphy (AtP)
|
|
|
|
|
[quote]
Doing my part to help out the Validator!
;-)
Yup, me too, 25 sob's wu in progress. :D
____________
Polish National Team |
|
|
|
|
|
I was thinking about running SoBs on my Pentium 4...after all, it's on 24/7 and I barely, if ever, use it. It's my GPU cruncher so it's fairly stable...thing is, it's running a P4. :P
Ah well, I've got my little badge. Time to complete the set and then I'll see about doing more of these. Last ones took over a week on the i5. With only one real computer here, I don't know if I wanna give up those cores.
Estimates on a 3.0GHz P4, WinXP 32-bit SP3, 1024MB RAM? Or same processor, but Fedora 14 OS instead...
____________
|
|
|
|
|
|
Another 13 wu's added. :D
____________
Polish National Team |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,157,676
RAC: 1,017,804
|
I was thinking about running SoBs on my Pentium 4...after all, it's on 24/7 and I barely, if ever, use it. It's my GPU cruncher so it's fairly stable...thing is, it's running a P4. :P
Ah well, I've got my little badge. Time to complete the set and then I'll see about doing more of these. Last ones took over a week on the i5. With only one real computer here, I don't know if I wanna give up those cores.
Estimates on a 3.0GHz P4, WinXP 32-bit SP3, 1024MB RAM? Or same processor, but Fedora 14 OS instead...
I'm not sure what the deadline these days is, but the last time I checked I think it was 6 weeks, if I remember correctly. That's good, because you'll need it on a P4. Rough estimate is that you'll need about 4 or 5 weeks per WU, crunching 24/7.
On the other hand, we've got new souped-up LLR code, which can be up to 20% faster than the old code. That could cut a week off your run times. The optimizations are, possibly, very depending on the CPU you have, so it's possible a P4 would see a better, or a worse, speed up than other CPUs.
As for Windows vs. Fedora, I'd stick with whatever is currently running your CUDA apps. Why open up a potential can of worms getting it to run correctly on another OS? You'll see slightly better performance on Fedora because of the bloat in Windows, but at least there's no longer a penalty on Windows because the new optimized app hadn't been ported.
____________
My lucky number is 75898524288+1 |
|
|
ich_eben*Volunteer tester
Send message
Joined: 14 Oct 10
Posts: 101
ID: 70063
Credit: 392,240,946
RAC: 101,167
|
Also started two new SoBs on my Laptop running a 64Bit Ubuntu 10.10
Lets see how it handles these two tasks:
http://www.primegrid.com/workunit.php?wuid=162197677
http://www.primegrid.com/workunit.php?wuid=147577344
First Update on my two SoB tasks on my Ubuntu machine.
Both units are at ~10% running for 32h.
So they are finished in about 320 hours which would be 13,3 days.
As i dont run my Laptop 24/7 they should finish in about 30 days.
I just hope they dont error out with some weird error because i shutdown and boot my machine roughly about once per day ...
____________
|
|
|
|
|
As i dont run my Laptop 24/7 they should finish in about 30 days.
I just hope they dont error out with some weird error because i shutdown and boot my machine roughly about once per day ...
Wow that's a lot of rebooting :(
____________
@AggieThePew
|
|
|
|
|
As for Windows vs. Fedora, I'd stick with whatever is currently running your CUDA apps. Why open up a potential can of worms getting it to run correctly on another OS? You'll see slightly better performance on Fedora because of the bloat in Windows, but at least there's no longer a penalty on Windows because the new optimized app hadn't been ported.
Yeah...that and I am having a supremely hard time getting CUDA drivers (not to mention GPU drivers) installed on Fedora. I know it can be done but apparently I need to install the R dev environment and lots of CUDA SDKs from source as root...which is a lot of lost computation time, considering both OSs are on the same machine!
Ah well. A software engineer (boyfriend's dad) told me that Linux on older machines is the way to go, and I believe it...PPS LLRs on this old thing take between 35 and 50 minutes (!) which is absurd. Also I've eliminated almost all the background crap with which XP is bloated, but to no avail.
I need a real CPU farm, I guess...
____________
|
|
|
|
|
Ah well. A software engineer (boyfriend's dad) told me that Linux on older machines is the way to go, and I believe it...PPS LLRs on this old thing take between 35 and 50 minutes (!) which is absurd. Also I've eliminated almost all the background crap with which XP is bloated, but to no avail.
This sounds about right with the larger PPS LLRs that are out there. PPS LLR has been going up in n (and I believe k) sizes so they will get longer. |
|
|
|
|
