GPU to 72 assignments completed prematurely?

[aketilander]

Originally Posted by ixfd64

I was able to trick PrimeNet into accepting my results by changing the factor depth to 74 bits in my results.txt file.

Never ever submit any faked results because that would make people question all your results. If some of your results were faked maybe your other results were as well? Things like that may not only jeapordize your personal results but cast doubts on the whole project as such. If you have faked some of your TF results how do we know that you have not faked your LLs?

When results are submitted to the server they should be exactly as they were when they came from the programs.

[ixfd64]

For the record, I did contact George and ask him to check in my original results for me, but he never did so. I know fake results are bad for the project, but this was supposed to be temporary measure. It shouldn't be too difficult to change the "2^74" back to "2^72" and recalculate the credits.

[Dubslow]

Originally Posted by ixfd64

For the record, I did contact George and ask him to check in my original results for me, but he never did so. I know fake results are bad for the project, but this was supposed to be temporary measure. It shouldn't be too difficult to change the "2^74" back to "2^72" and recalculate the credits.

"Temporary measure" to what end?

The only reason for George to manually add them would be to get you credit -- there is absolutely no reason the server needs to record your work. At the same time, it's probably easier for him to just give you credit rather than force duplicate results into the server. Maybe he ignored you because it would be difficult to do either -- we can't assume that it's easy.

Now you've improperly faked out the server with "results" that never actually happened. Why on Earth would you ever think that your duplicate results (admittedly not your fault) are needed by the server so badly that you add tens of fake results that serve no purpose other than get you credit?

Is credit really that important?

In order to save George the fuss of cleaning up your mess (and no, you have no reason to think "it shouldn't be too difficult"), I highly suggest you unreserve whatever assignments you have now and start crunching those results you've already claimed to have done. Yes, it might take a while, but that's totally your fault. Some people might be willing to help. (I would volunteer, but I don't have my GPU at the moment.)

Also notice that none of the other people who unfortunately did duplicate work are complaining about their credit (nor are they submitting fake results).

[kracker]

Oh come on... maybe he didn't know....
@ixfd64: If you want, I can do a few.

[ixfd64]

Assuming George doesn't fix the results:

1. Would anyone mind if I P-1'd the exponents first?

2. I appreciate the help offers. However, that gives rise to another issue: how do I transfer post-v4 credit to another user? I know it can be done for v4 assignments, but I'm not sure if it's possible for v5.

[kracker]

Credit is no biggie at all for me...

[Dubslow]

The lesson here should be "credits don't matter".

[kracker]

The only use for credit is to tell me how much work it will take. That's about almost all.

[LaurV]

Originally Posted by kracker

The only use for credit is to tell me how much work it will take. That's about almost all.

No, the credit is to make fun of other people like petrw1 when I overtake them at DC

Joking apart, I will put my heavy artillery on them. Do I need to make 72-73 too? Or only 73-74. The 72-73 will take about 12 hours and the other about 24, on 3xGTX580 (8 or 9 tasks).

At least good that he said. I suspect there are many "fake" results like this in the data base, other people did it for credit and did not say.

For confirmation:

Factor=59752981,72,74
Factor=59753069,72,74
Factor=59753131,72,74
Factor=59753179,72,74
Factor=59753209,72,74
Factor=59753219,72,74
Factor=59753279,72,74
Factor=59753329,72,74
Factor=59753339,72,74
Factor=59753341,72,74
Factor=59753357,72,74
Factor=59753377,72,74
Factor=59753479,72,74
Factor=59753539,72,74
Factor=59753663,72,74
Factor=59753693,72,74
Factor=59753741,72,74
Factor=59756927,72,74
Factor=59756951,72,74
Factor=59757041,72,74
Factor=59757149,72,74
Factor=59757209,72,74
Factor=59757319,72,74
Factor=59757391,72,74
Factor=59757433,72,74
Factor=59757449,72,74
Factor=59757571,72,74
Factor=59757587,72,74
Factor=59757713,72,74
Factor=59757721,72,74

[ixfd64]

I don't think [72, 73] is necessary because the range was already tested by the other person.

Either way, I'm going to try to reserve all 30 exponents for P-1 testing. If George does get around to fixing the results, then he'll have less work to do.

ETA: I really apologize for getting people into this mess. It seems my lack of patience got the best of me this time. :-(

[LaurV]

I can't do P-1. I am heavy CPU bottle-necked. For the GPUs I have, I might need eventually to attach few additional external CPU cores :D

But I'll do the TF for all 30 expos, then. 73 to 74 only. Next time please do not report confusing results. Please check again if they were all [72-73]-ed by a third party, I have no time to check for each. 2:25AM here, I need to go to sleep.