Hey Everyone,

I released a new version of Separation Modfit today (v1.36). If you have any issues with it please post them here.


Thank you for your continued support,

Jake W.

Sorry Jake.

I am getting exactly the same result as with V1.34. Immediate (0.00 sec CPU time) fail as a computational error, reporting:

(unknown error) - exit code -1073741515

My computer is a Dell XP machine with a 2-core Intel CPU, and no GPU. So far, no ModFit task under 1.34 or 1.36 has even looked as though it might run.

Jake,

I have been running (modified-fit) off and on for the past 30 days or so stopping because of high error rates, and restarting each time new versions appeared. Unfortunately, v1.36 doesn't seem any better than the previous versions (v1.32 & v1.34). Currently I'm running ~6% VALIDATE ERROR rate over the past 20 hours. I checked the top 100 Hosts to see if anyone else was having a problem. Most of the top 100 hosts use the HD 7970 card with the driver identified by BOINC as 1.4.1848 as do I. I found that the only ones that were not realizing a significant VALIDATE ERROR % rate were those not running MOD-FIT. Without exception it seems that everyone is having a problem with VALIDATE ERRORs. I also made another observation that may or may not be related to another recently identified problem. The top Hosts running MOD-FIT also have a large number of COMPLETED, CAN'T VALIDATE tasks. A lot of these will ultimately validate, but some of these probably won't. Here we find that a number of these tasks have been downloaded to other Hosts only to be ABORTED BY USER. Further research of these hosts shows that in some cases, literally thousands of tasks have been manually aborted by the user. And its not just one or two Hosts that show such behavior - with very little effort I've identified at least a dozen. Since I can't for the life of me posit a sane motive for such behavior, I'm guessing it may be frustration over the errors volunteers are currently experiencing. (O.K., maybe not.) At any rate, could you please fix this. Last month at this time I had an RAC rating of ~325000. Now its barely 200,000 and falling. At 325000 I felt I was contributing to the effort. Now I'm starting to feel like an outsider.

We had a similar problem to this last year when I tried to release the Win32 Version. I will take a look into which platforms are having the highest abort rates and see if killing those platforms will help. Might have to deprecate Win32 again.

Jake W.

Jake, sorry, but I just lost 7.5 hours of computer time working a ModFit 1.36. That was one of 2 that self aborted. I aborted 12 more because I didn't want them eating up my processor's time.
Keep up the good work. I'm confident you'll find the answer.
Mike

Looks like the issue is contained to Win32 so I cut them off. Sorry for the inconvenience everyone. I will continue to look into the issue but for now we are dropping Win32 support so we can get everyone else's runs validating.

Thank you for your continued support,

Jake W.

Jake,
I don't know if you need this sort of feedback or not.

Lost 3h 12m on one of the Modfit ver 1.36.
I aborted the remaining 3 units of work.

It's every day/night. I can wind up losing hours of processing time due to these units that bomb out. The project is losing a lot of computer time, as well.

Mike

It yet another session of failed tasks and me aborting the rest.
Exit 0 again!
Gawd....what is with this Exit 0 crud anyway?
Blocked Modfit yet again, can't run it.

Jake,
5 Modfit 1.36 suffered the dreaded Computation error.
1 Modfit 1.36 was aborted by me.
Mike

I checked my firewall, modift was being trapped by it. Trying to change some settings and try again.

It yet another session of failed tasks and me aborting the rest.
Exit 0 again!
Gawd....what is with this Exit 0 crud anyway?
Blocked Modfit yet again, can't run it.

Well what do you know....it was the firewall.
But now I have a few tasks that validation was inconclusive and one that could not validate. What's with that?

I think (hope) I have finally gotten through the buggy tasks and that my firewall can behave now.

Hi, 7 of my Milkyway@Home Separation (Modified Fit) v1.36 (opencl_ati_101) WU's complate but waiting. this is one of them.
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=849329177

and also 8 of them problems like this http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=849307241

Hey Guys,

While the Win32 version was running the error rate was a little high. I took it down so it should now be running more smoothly and validating more frequently and quickly.

Jake W.

Hey Guys,

While the Win32 version was running the error rate was a little high. I took it down so it should now be running more smoothly and validating more frequently and quickly.

Jake W.

i so your message and resume the milkyway again lets see how it go.

Jake, Yep, it's running smoother. I only had to delete 10 wu's, last night, due to the one that processed and errored out after taking 5 hours of computer time. Not a good show. I've not seen any difference on my side of the modem.
All the MilkyWay@Home 1.02 (opend_nvidia)work great. It's 100%
All the ModFit 1.36 are a total bust. 0% completion.
Thanks,
Mike

Michael Bennett,

Looks like you are running Win32 so you should not be seeing any new work units for Modfit. I had to pull the Win32 application for now to keep everything stable. I will be looking into a fix for the future but no guarantees on how long it will take. Until then you should still be able to run nbody and our stable version of MilkyWay@home just like before.

Sorry for all of the issues.

Jake W.

Jake

I am also running a Win32 system, and just had 3 more ModFit 1.36 tasks downloaded. So far there have been no 1.34 or 1.36 tasks which have run on my system.

I am just about to put a block on receiving more until you can find what is wrong with it and correct the problem.[/u]

Hey Martin,

Feel free to block them, you should stop getting them soon anyway. You will still get nbody and normal separation runs. It is just the new separation that is effected.

Sorry,

Jake W.

I've got 3 computers and all have had problems with Modfit v1.36 - but the symptoms differ. Two computers are 32 bit Intel processors and run Win XP. Modfit v1.36 units fail immediately with computation errors on these 2 computers. The third computer has a 64 bit AMD processor and runs Win 7. Modfit v1.36 units run to completion on this computer but immediately go to "validation inconclusive".

Stick,

Your Win32 machines will no longer be receiving Modfit work units so that won't be a problem any more. As for your 64 bit computer that is completely normal. We require ~3 computers to return the same result for every work unit we send out. Validation inconclusive just means we are waiting for others to return their results for the work unit before we award credit. This ensure people aren't trying to game the system just for credits and it ensures we are getting reliable results for our optimizations.

Sorry for the issues with win32 and confusion,

Jake W.