Tweak to assignment rules

[Prime95]

I've just implemented a minor change to the assignment rules. Computers that ask for "what makes the most sense" work type and have never returned an LL result will be given double-checks rather than first-time LL tests. There are two reasons why I'm trying this:

1) Stress testers and newbies are more likely to finish a DC before giving up.
2) Stress testers that have no intention of ever finishing an LL are causing a "churn" in cat 4 LL tests making it difficult for GPU72 to keep up with optimal TF. Hopefully this will move some of the churn to the DC range.

[chalsall]

Originally Posted by Prime95

I've just implemented a minor change to the assignment rules.

...

2) Stress testers that have no intention of ever finishing an LL are causing a "churn" in cat 4 LL tests making it difficult for GPU72 to keep up with optimal TF. Hopefully this will move some of the churn to the DC range.

Excellent idea George!

So everyone knows, there is just short of a year's worth of DC Cat 4 candidates already appropriately TF'ed, while we're still releasing some LL Cat 4 at only 73 bits.

Or, in other words, for those doing TF'ing, please don't feel the need to do any additional DC TF'ing. If anything, anyone doing DC TF'ing should move back to LL TF'ing (if so inclined).

[kladner]

Hope you and SWMBOK'd are doing better, Chris.

[LaurV]

Originally Posted by chalsall

Excellent idea George!

+1 to this! I was thinking about it some time ago, or well... at least to something close, but never voiced it out. This is an excellent idea George!

[chalsall]

Originally Posted by kladner

Hope you and SWMBOK'd are doing better, Chris.

Thanks for the thoughts. Getting there... Still can't type at my usual 80 WPM for hours, but at least can do a bit (for a while).

[Prime95]

This wasn't as minor as I thought! Nearly 1000 cat4 DCs have been assigned in the last day. At that rate, we'll chew through 60000 cat4 DC assignments in the next 60 days! I don't see that as much of a problem.

I wonder why our retention rate is so low or why so many stress testers can't read the initial startup screen and not get Primenet assignments.

[retina]

Originally Posted by Prime95

I wonder why our retention rate is so low or why so many stress testers can't read the initial startup screen and not get Primenet assignments.

Reading takes mental effort and delays the opportunity to do other more exciting things like watch cat videos. IOW: People are lazy. You can't change people.

[Mark Rose]

Originally Posted by Prime95

This wasn't as minor as I thought! Nearly 1000 cat4 DCs have been assigned in the last day. At that rate, we'll chew through 60000 cat4 DC assignments in the next 60 days! I don't see that as much of a problem.

I wonder why our retention rate is so low or why so many stress testers can't read the initial startup screen and not get Primenet assignments.

Waaaay back in the day in 1999, on PII 350 running Win98, when I first ran Prime95, I gave up on the assignments because they took so long. Obviously not much can be done to change that, but I'm sure other people have the same thing happen.

I guess it's good that I've been DCTF'ing lately. I can only do ~135 ->71 per day at the moment though.

[TheMawn]

Originally Posted by Prime95

I wonder why our retention rate is so low or why so many stress testers can't read the initial startup screen and not get Primenet assignments.

Would it be worth the work to have the stress-test available as a much smaller stand-alone program? On the other hand, we bump into the same problem of lazy people grabbing the wrong piece of software, I guess.

[kracker]

Originally Posted by TheMawn

Would it be worth the work to have the stress-test available as a much smaller stand-alone program? On the other hand, we bump into the same problem of lazy people grabbing the wrong piece of software, I guess.

I think it would be almost the same size(assuming it's the assembly taking most of the space).

[Mark Rose]

Originally Posted by TheMawn

Would it be worth the work to have the stress-test available as a much smaller stand-alone program? On the other hand, we bump into the same problem of lazy people grabbing the wrong piece of software, I guess.

I'd rather have them do DCLL, but with some additional mechanism to communicate to them that if they get a matching residue that their system can be considered stable. DCLL takes less time, and that's much more of an incentive to actually finish the job than with LL.