Hello!

Ops... what is going here?

3 results but no Credits?
The fourth result is missing still...

All 3 results are with optimized apps.
And the status: "Checked, but no consensus yet"


http://setiathome.berkeley.edu/workunit.php?wuid=99171644


What's wrong?
It's not good to let run optimized?


Greetings!

---

Your computer is hidden so I can't look at the results for myself, so a guess will have to do. Everytime a WU is processed there is a small variation in the result, even if recrunched on the same machine. Seti looks for close matchs to validate results. In this case the results may be close to the same but still just outside the parameters for validation therefore validation is waiting for the fourth result to see if it will fill the gap between the other WU.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

Interestingly, they all ran on different app's and none of them agree on the number of gaussians found.

Not too surprising there's no concensus yet.

Alinator

Interestingly, they all ran on different app's and none of them agree on the number of gaussians found.

Not too surprising there's no concensus yet.

Alinator

Yes, but none of the publicly released applications keep track of the individual signal counts when they're stopped and restarted. The 2.0_A Chicken cruncher went through that three times, so its one reported Gaussian was found after the last restart.

I don't remember whether Crunch3r's apps left any evidence of stop/restarts, but I'm pretty sure the 1.41 Chicken build Dirk's using doesn't.

A difference in the number of reported signals definitely puts the results in the "Weakly similar" realm. But for a signal very close to the reporting threshold it can happen even if both versions of the application are working correctly.

My guess is when the 4th result is returned everyone will get credit. The average turnaround for that system is over 3 days, so the resolution may take awhile.

                                                            Joe

Interestingly, they all ran on different app's and none of them agree on the number of gaussians found.

Not too surprising there's no concensus yet.

Alinator

Yes, but none of the publicly released applications keep track of the individual signal counts when they're stopped and restarted. The 2.0_A Chicken cruncher went through that three times, so its one reported Gaussian was found after the last restart.

I don't remember whether Crunch3r's apps left any evidence of stop/restarts, but I'm pretty sure the 1.41 Chicken build Dirk's using doesn't.

A difference in the number of reported signals definitely puts the results in the "Weakly similar" realm. But for a signal very close to the reporting threshold it can happen even if both versions of the application are working correctly.

My guess is when the 4th result is returned everyone will get credit. The average turnaround for that system is over 3 days, so the resolution may take awhile.

                                                            Joe

Hello!

Thanx for replies!

No, no... my PC is the slowest with ~77400 seconds...
With:
Version: Windows MMX 32-bit based on seti V5.15 'Chicken Good!'
Rev: (R-2.0|QxB|FFT:IPP_SSE2|Ben-Joe|)
CPUID: 'AMD K7 Athlon (Argon)'

...and the WU was stopped 4 times , or ?


O.K., then I'll wait to the fourth result...


Greetings!

---

Can anyone explain how out of a quorum of 3 results 2 got credit and I got zero (invalid) for this wu-

http://setiathome.berkeley.edu/workunit.php?wuid=102210396

I have had this happen for 3 or 4 other WUs over a period of time.

---

Can anyone explain how out of a quorum of 3 results 2 got credit and I got zero (invalid) for this wu-

http://setiathome.berkeley.edu/workunit.php?wuid=102210396

I have had this happen for 3 or 4 other WUs over a period of time.

I've had this on rare occasions too, but only on machines running:

Version info: Windows SSE4-Core 2 32-bit V5.15 'Chicken Good!' (R-1.41|+more_vec|xT)

which I see you're using too. Let's hope F. lets us see/lets Simon use the Core 2 SSSE3 code enhancements he offered so that Simon and the rest of the coding team can release a real version 2.0 for Core 2.

Thanks Richard glad I am not the only one, it is rather rare. Been a couple of weeks since F posted and patiently waiting for the super code.

---

Calling who?, where are you who?? Haven't heard a peep outta him for a couple of weeks, might still be playing with that hot rod Jag of his.
But I wish that soon we'd be hearing from, you know, who?!

---

"Time is simply the mechanism that keeps everything from happening all at once."

Can anyone explain how out of a quorum of 3 results 2 got credit and I got zero (invalid) for this wu-

http://setiathome.berkeley.edu/workunit.php?wuid=102210396

I have had this happen for 3 or 4 other WUs over a period of time.

The validator waits until there are three results returned.

If two are strongly similar (read nearly identical), then these two are granted credit. If the third is at least weakly similar to these two, it is granted credit.

ELSE If all three are weakly similar, they are all three granted credit.

ELSE, there is no match yet. Wait for the fourth (or send another).

In the case where 2 results are granted credit and the third is not granted credit, the first two were strongly similar, but the third was not similar.

---

BOINC WIKI

OK so what would happen when the 4th result is returned and is strongly similar to my invalid result - it would be, 2 with credit and 2 without?

---

I looked around (Core 2 Duo, Core 2 Quad, Xeon) and sometimes they have invalid results with the SSSE3 app...
Maybe it's better to use the SSE3 app?

How much less performance have this SSE3 to the SSSE3?


Greetings!


BTW. The fourth WU I was waiting for... I think it'll not come... deadline 6 Dec 2006 18:16:58 UTC (in ~8 hours)...
http://setiathome.berkeley.edu/workunit.php?wuid=99171644

---

The validator waits until there are three results returned.

If two are strongly similar (read nearly identical), then these two are granted credit. If the third is at least weakly similar to these two, it is granted credit.

ELSE If all three are weakly similar, they are all three granted credit.

No, the Seti-validator demands 2 results to be "strongly similar" for the wu to pass validation. If "weakly similar" had been enough, it means a signal only showing-up in one result could be added to the Science-database...

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

OK so what would happen when the 4th result is returned and is strongly similar to my invalid result - it would be, 2 with credit and 2 without?

The Validator never checks against a result that has already been marked invalid.

Anyway, the 4th. result is in, and it passed validation.

This indicates the 2 most probable reasons is either a software-problem with your optimized seti-application, or a hardware-problem like "too much overclocked", so computer is randomly spitting-out garbage...

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

This indicates the 2 most probable reasons is either a software-problem with your optimized seti-application, or a hardware-problem like "too much overclocked", so computer is randomly spitting-out garbage...

I've reported before that I've seen the same issue, and mine was on brand-new, stock speed Dell kit (Dimension and Precision desktops).

So I think there's a (small, rare) problem with Simon's v1.41 optimised app for Core 2.

I would agree with Richard as I have checked other C2D results and everyone checked so far has the very rare invalid result using Simon's v1.41 optimised app for Core 2. Is Simon aware of this?

---