Invalid Results-way too many

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5,870
Credit: 115,919,450,818
RAC: 35,383,015

RE: Thank-you Gary for

Message 16399 in response to message 16396

Quote:

Thank-you Gary for your input it has been appreciated and also if I could get a link to the new application it may help .....

You're welcome. There is a thread in the Cafe entitled "NEW: WINDOWS TEST APPLICATION FOR EINSTEIN@HOME" which may be the new app that Stick was referring to. I have no knowledge of it other than to say that I thought it was something to do with fixing things for people who were suffering from graphics related problems. I don't use graphics and I don't have any problems with the old app so I've never bothered to investigate.

Good luck!!

EDIT: We've probably made enough noise in this thread to grab the attention of someone like BA or BM who would certainly be able to look at your problem WU and perhaps give us a rundown on really why your result was invalidated. Here's hoping anyway!!

Cheers,
Gary.

Jayargh
Jayargh
Joined: 9 Feb 05
Posts: 64
Credit: 1,205,159
RAC: 0

Seems like this thread is

Seems like this thread is describing my prob but no one is screaming about the invalid results that should be valid....http://einsteinathome.org/node/189795
Perhaps this whole thing may be related to the interreaction between applications 4.79 and 4.82 in validation ...seems suspicious

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 33,018,056
RAC: 16,626

RE: Stick- My Invalids

Message 16401 in response to message 16390

Quote:
Stick- My Invalids started wiyh Boinc4.19....and continued....4.25...4.45...and now 4.72.
It seems platform made little difference...perhaps application IS the key. Can you link me( and others obviously) the Beta application?

to the

The link to the Beta App is near the bottom left corner of the EAH home page under "More Information"

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 33,018,056
RAC: 16,626

RE: RE: Thank-you Gary

Message 16402 in response to message 16399

Quote:
Quote:

Thank-you Gary for your input it has been appreciated and also if I could get a link to the new application it may help .....

You're welcome. There is a thread in the Cafe entitled "NEW: WINDOWS TEST APPLICATION FOR EINSTEIN@HOME" which may be the new app that Stick was referring to. I have no knowledge of it other than to say that I thought it was something to do with fixing things for people who were suffering from graphics related problems. I don't use graphics and I don't have any problems with the old app so I've never bothered to investigate.

Good luck!!

EDIT: We've probably made enough noise in this thread to grab the attention of someone like BA or BM who would certainly be able to look at your problem WU and perhaps give us a rundown on really why your result was invalidated. Here's hoping anyway!!

Just for the record, the thread that got me to try the Beta App was: Message boards : Problems and Bug Reports : "Client Errors". It is a bit old. I was having the problems in early August and the thread had already "died out" when I found it.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5,870
Credit: 115,919,450,818
RAC: 35,383,015

RE: Just for the record,

Message 16403 in response to message 16402

Quote:
Just for the record, the thread that got me to try the Beta App was: Message boards : Problems and Bug Reports : "Client Errors". It is a bit old. I was having the problems in early August and the thread had already "died out" when I found it.

A keyword search on thread titles reveals an amazing number of threads that have something to do with "Client Error" in their title :). However your use of quotes narrowed it down to the one you referred to. I reread quite a few of these threads (fairly quickly) and got the impression they were all about cases of the science app terminating abnormally rather than successfully completed results that then had validation problems.

Anyway, thanks for the reference. It was interesting to check out problems that people had been reporting.

Cheers,
Gary.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5,870
Credit: 115,919,450,818
RAC: 35,383,015

RE: Seems like this thread

Message 16404 in response to message 16400

Quote:
Seems like this thread is describing my prob but no one is screaming about the invalid results that should be valid....http://einsteinathome.org/node/189795
Perhaps this whole thing may be related to the interreaction between applications 4.79 and 4.82 in validation ...seems suspicious

No Devs have yet responded to your pleas for an investigation so lets up the "noise level" a bit more to see if we can get someone to take notice :).

I now have an actual example of my own of a "validation problem" of the type you describe. The WUID is 2005591 and the 4 ResultIDs are 8454998 to 8455001.

The first result in was 5000 from a Intel/WinXP box, followed by 5001 from an Opteron Dual Core/Linux, and then 4999 from a PowerMac/Darwin box which is also a 2 cpu box. With these three results in, the validator was called and the WinXP box was declared invalid and credit was awarded to the other two only.

Then my result 4998 came along (AthlonXP/WinXP) and was also declared invalid. This is the first "failed to validate" result I've seen in thousands of successful results. I've been producing around 50 per day although this has dropped a bit as I've diverted significant resources into LHC. The results list of the box in question has no other problems of any type so this one is a bit of a mystery. However, the presence of two "failed to validate" results for this WUID does indeed seem a bit suspicious. The only other point of significance that I saw when looking into this was the fact that both the Linux and Darwin boxes had *large* numbers of "ghost" WUs, some of which had started to expire. I used to see quite a few "ghosts" in my own lists previously but have not seen *any* for at least a month or two. I thought the problem had been cured.

Anyway, a response from BA or BM would be much appreciated.

Cheers,
Gary.

Jayargh
Jayargh
Joined: 9 Feb 05
Posts: 64
Credit: 1,205,159
RAC: 0

RE: No Devs have yet

Message 16405 in response to message 16404

Quote:

No Devs have yet responded to your pleas for an investigation so lets up the "noise level" a bit more to see if we can get someone to take notice :).

I now have an actual example of my own of a "validation problem" of the type you describe. The WUID is 2005591 and the 4 ResultIDs are 8454998 to 8455001.


Anyway, a response from BA or BM would be much appreciated.

Thanks Gary,(for making noise) :) I have brought this up to devs back in feb/mar area and never got a response. So gave up. An invalid result when it should be valid represents approx 10-15% of my daily output ....quite significant...It really quells any excitement about a project when you see a big fat 0 sittin there after 12 hours of crunching.

I am also trying the beta application in hopes to see no more invalids. Thank-you for the replies as it gives me something to try

Tern
Tern
Joined: 27 Jul 05
Posts: 309
Credit: 99,440,614
RAC: 15

I can now come in from the

I can now come in from the other side - I'm on a Mac, and since Mac Version 4.82 was downloaded towards the end of last month, I've had 2 of my 16 verified results get 0 credit as "invalid". Both were cases where the other three computers were all on Windows. The V4.82 is the Altivec-enhanced version, and 15 of these 16 (including both invalid ones) were on a G4, the other was on a G3.

I think we're back to where we were some time ago; Linux and Mac agree, Windows doesn't, whoever gets there first gets credit - it's just not every time, only sometimes. (I've found cases where I _did_ get credit, and the others were all Windows, and I was not the first...)

WUs flagged "invalid": 1988628 and 1980307

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 33,018,056
RAC: 16,626

RE: RE: Just for the

Message 16407 in response to message 16403

Quote:
Quote:
Just for the record, the thread that got me to try the Beta App was: Message boards : Problems and Bug Reports : "Client Errors". It is a bit old. I was having the problems in early August and the thread had already "died out" when I found it.

A keyword search on thread titles reveals an amazing number of threads that have something to do with "Client Error" in their title :). However your use of quotes narrowed it down to the one you referred to. I reread quite a few of these threads (fairly quickly) and got the impression they were all about cases of the science app terminating abnormally rather than successfully completed results that then had validation problems.

Anyway, thanks for the reference. It was interesting to check out problems that people had been reporting.



Sounds like my problems were a little different from yours. Although my units appeared to be running all the way to completion, they were rejected as "computation errors". I again reviewed the thread I referred to earlier and found the post that prompted me to try the Beta App. It is shown below. Note the "failing address" reference. Anyway the Beta App worked for me. Hope you have the some of the same luck.

----------------------------
from Message 16070 - Posted 27 Jul 2005 16:24:57 UTC

I would suggest to try the App that is currently in our Public Beta Test. Seems that some people having problems with the 4.45 client and our "official" App, which was built with a rather old version of the BOINC library.

The failing address is deep down in the system libs, might be somewhat got broken down there on your installation. A hardware problem I would consider to be rather unlikely (though not impossible), as the error happens always in the same region of the process' memory space.

BM
-------------------------------

Bruce Allen
Bruce Allen
Moderator
Joined: 15 Oct 04
Posts: 1,119
Credit: 172,127,663
RAC: 0

RE: Anyway, a response

Message 16408 in response to message 16404

Quote:

Anyway, a response from BA or BM would be much appreciated.

Gary, you spend so much time solving other peoples problems, that I thought I had a duty to reply!

Currently the validator marks about 0.3% of results INVALID. Most of these are probably valid, it's just that the validator has a difficult job to do and is sometimes too conservative. We've done our best to tune it, but at the 0.3% level we don't know how to improve it.

The different numerical results returned by different CPUs, different OSes and different OS versions all mean that sometimes results are correct, but not recognized as such.

I looked to see what was going on with jrenkar's results, but all the ones currently in the database are VALID. No sign of trouble.

Cheers,
Bruce

Director, Einstein@Home

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.