I noticed that my Recent Average Credit had fallen over the weekend so just checked the progress of my tasks and discovered that the vast majority of the SETI@Home v7 7.00 work units that I've run since 6 March, across 4 machines, are listed as either

Completed, marked as invalid, or
Completed, validation inconclusive.

All machines running BOINC 7.4.36 as a service

Anyone else having the same problem? Is this a known issue? Is there something I can alter to correct it?

Cheers

SusieQ

Same here. All "shorties" on both cpu and gpu wu's from 27my12ac.30270...


Keifer

---

I noticed that my Recent Average Credit had fallen over the weekend so just checked the progress of my tasks and discovered that the vast majority of the SETI@Home v7 7.00 work units that I've run since 6 March, across 4 machines, are listed as either

Completed, marked as invalid, or
Completed, validation inconclusive.

All machines running BOINC 7.4.36 as a service

Anyone else having the same problem? Is this a known issue? Is there something I can alter to correct it?

Cheers

SusieQ

Known issue. Check out the thread:
WUs with very unusual Autocorr parameters

I noticed that my Recent Average Credit had fallen over the weekend so just checked the progress of my tasks and discovered that the vast majority of the SETI@Home v7 7.00 work units that I've run since 6 March, across 4 machines, are listed as either

Completed, marked as invalid, or
Completed, validation inconclusive.

All machines running BOINC 7.4.36 as a service

Anyone else having the same problem? Is this a known issue? Is there something I can alter to correct it?

Cheers

SusieQ

Everyone else is having the same problem, it's the ongoing WUs with very unusual Autocorr parameters issue. There's nothing a user can alter to correct it.

The splitters here usually only run long enough to produce 5 groups of WUs (1280 total), and the first one or two groups of a run typically have correct parameters. The March 4 run of the splitters went much further and probably most groups are bad.

                                                                  Joe

I thought the unusual autocorr parameters only caused the legacy v6 problem. These are all running as v7.

---

David
signature sent back to alpha testing

The similar incident at the main project had autocorr_fftlen 0, which is what would be sensed if there were no autocorr parameters. The applications use that to totally skip autocorrelation processing for v6 tasks.

Here, the bad ones have autocorr_fftlen 8. Autocorrelation processing is tried, the CUDA apps fail but others apparently do the DCTs at that ridiculously short length. However, the final step of actually looking for signals in the transformed array is inhibited because the autocorr_per_spectrum is set to 0. That means no best_autocorr signal, and the v7 Validator invalidates the results of tasks which ran full length. When it's checking a pair of results, as soon as it marks one invalid there's no longer a pair so the other is inconclusive until another result is received.

                                                                   Joe

Thanks for the info guys, at least I now know it's not just me!

Rather than waste my CPU time, I aborted all the work units labelled 27my12ac.30270... in the hope I'd get some work units that worked. Unfortunately though it looks as if I've hit a daily limit as no new work units were downloaded. I thought there was just a limit on the number of tasks you could have in progress at any one time - obviously I was wrong :-)

Never mind, I'll crunch some main SETI units until I can download some more Beta ones tomorrow.

Cheers

SusieQ

For the last couple of days I've just left my PC to run through the work units I've been sent, even though pretty much all of them have been a wasted effort in terms of the results. However, each day my PC runs out of Beta work units to run, as "This computer has finished a daily quota of 33 tasks".

If the only way to clear these faulty work units from the queue is to have people run them, it would be very helpful if the daily quota could be lifted - even if only temporarily.

Any chance SETI Beta?

For the last couple of days I've just left my PC to run through the work units I've been sent, even though pretty much all of them have been a wasted effort in terms of the results. However, each day my PC runs out of Beta work units to run, as "This computer has finished a daily quota of 33 tasks".

If the only way to clear these faulty work units from the queue is to have people run them, it would be very helpful if the daily quota could be lifted - even if only temporarily.

Any chance SETI Beta?

It looks like about half the tasks your CPU is doing end up in that state. With the other half being marked as valid. Maybe they can load up some known good data sets for testing. Especially given the new GPU app releases for test.
SETI@home Beta's focus is application testing & it is my understanding the data we processes here does not get put into the science database. Not to say any of it is "wasted effort". An error here is a result that points to how something is possibly not being handled the most elegant way, or when something not accounted for occurs.

Now suddenly, my last few 27my12acs have validated and I'm getting a bunch of inconclusive/invalids on 21fe13abs. The bad ones so far have been .18498, but the ones I'm crunching now are .17120; we'll see if there's any difference.

---

David
signature sent back to alpha testing

Things are looking up my end, for S@h v7 anyway; the last few tasks have validated successfully.

---

I'm still seeing a lot of invalids and inconclusives on the 21fe13abs, but I'm also seeing enough successes to discourage me from aborting them.

Strange.

---

David
signature sent back to alpha testing