Something went dudishly with the assignment bot today, as I can see 510 available exponents to first time TF which were cooked to 68 bits only; one at 45M and 509 at 63M. I requested to TF 400 of them to 71 and the system made fun of me... :smile:

Therefore, they are still available for TF...

Getting "LLTF to 71" assignments has been impossible for the past [URL="http://www.gpu72.com/reports/overall/graph/assigned/"]few days[/URL]. Not that I haven't tried... :wink:

The problem is not "71", but ">60M". If you try to get assignments to 73, but fill the range with 60-100M, you still got nothing. I remember the same problem when DC switched "over 30M" and chalsall fixed "by hand" and commented on the forum something about "stupid programmer error" :razz:
Maybe is the same subspecies of bacteria now?

[QUOTE=LaurV;310130]Maybe is the same subspecies of bacteria now?[/QUOTE]

Yup -- another SPE. A very old limit to below 63M when I brought in some candidates above 900M for some testing. Those candidates at 63M are available now.

Perfect! Just grabbed 400 of them to 71, this should go fast, thanks a lot.

(I would grab all but now I already took few DC tests before, and filled one GPU :ick:)

Is there a related problem picking up completed assignments in the same range?
For example [URL="http://mersenne.org/report_exponent/?exp_lo=63402701&exp_hi=10000&B1=Get+status"]63402701[/URL] still shows up as assigned but not completed in GPU272.

[QUOTE=sdbardwick;310182]Is there a related problem picking up completed assignments in the same range?
For example [URL="http://mersenne.org/report_exponent/?exp_lo=63402701&exp_hi=10000&B1=Get+status"]63402701[/URL] still shows up as assigned but not completed in GPU272.[/QUOTE]

To minimize the impact on PrimeNet, those ranges with only a few assignments are only checked a few times a day. I've "told" spidy it should consider 63M to be of interest.

Ok, makes sense. Personally, I wouldn't care if they are polled less frequently; just wanted to make sure that they aren't being overlooked somehow (to avoid them being needlessly reassigned).

Since we have started turning in some LL-D results we are surprised at how many of them are being run for the third time, due to a previous suspicious result.

We do understand that our sample size is small, but three out of eight exponents (37.5%) were suspect! We hope that is not the norm!

[URL]http://www.mersenne.org/report_exponent/?exp_lo=27220367[/URL]
[URL]http://www.mersenne.org/report_exponent/?exp_lo=27252019[/URL]
[URL]http://www.mersenne.org/report_exponent/?exp_lo=27259907[/URL]
[B][URL]http://www.mersenne.org/report_exponent/?exp_lo=27248293[/URL][/B]
[B][URL]http://www.mersenne.org/report_exponent/?exp_lo=26183239[/URL][/B]
[URL]http://www.mersenne.org/report_exponent/?exp_lo=27238933[/URL]
[URL]http://www.mersenne.org/report_exponent/?exp_lo=27237523[/URL]
[B][URL]http://www.mersenne.org/report_exponent/?exp_lo=27243217[/URL]

:unsure:
[/B]

[QUOTE=Xyzzy;310212]Since we have started turning in some LL-D results we are surprised at how many of them are being run for the third time, due to a previous suspicious result.

We do understand that our sample size is small, but three out of eight exponents (37.5%) were suspect! We hope that is not the norm![/QUOTE]

It is statistically likely that the lowest candidates issued (re-issued) by Primenet have already been DCed once, but the residue didn't match.

With absolutely no analysis behind this statement, I suspect what you are observing is nominal.

In full agreement with our observations too. If you look to older posts, and the thread "don't dc them with...", you will see that it was a time when we were very worried our DCs producing lots of mismatches. We were scared to death by the perspective that our gpus went nuts. Time (and triple checks) proved that our DC residues were right (mostly, except when we exaggerated with overclocking) and the original residues were wrong.

edit: we have some examples of exponents where quadruple and quintiple checks too, were necessary to sort the issue