Primenet web design

[Jayder]

Originally Posted by James Heinrich

Since a long time ago I've had the Poorly P-1 factored page which used to be very useful in finding exponents with badly-done P-1. Literally thousands of (small, below 10M) exponents had P-1 redone and hundreds of new factors were found.
It's still somewhat useful if you're looking to redo some P-1 work that was badly done (B1=5000 or something silly). It won't give you any info about ECM or TF though (I also have a page for less TF than expected but a lot of what's shown there is borderline cases where the rough formula I use to calculate target TF level doesn't match up exactly with the precise formula that GIMPS uses.)

Thanks James, I make good use of your site. I'm not really sure how to use your P-1 page, though. Let's say I come to this page: link. If I put pfactor lines in my worktodo.txt, Prime95 doesn't take it, and if I put pminus1 lines in my worktodo, the bounds are less than what have already been done. So I need to put the bounds in myself, which is fine, but I think there may be improvements to be made. Or maybe I am doing things wrong?

It would be nice if I could search for exponents with less than 0.2 GHzd spent on P-1, and if the form then gave me bounds that achieved that. Combined probability is a fine metric, but I lean more towards wanting to search by GHzd. Maybe it's stupid for me to want that?

[James Heinrich]

Originally Posted by Jayder

If I put pfactor lines in my worktodo.txt, Prime95 doesn't take it

Explain? What does Prime95 say?

[Jayder]

Apologies. When I take the Pfactor lines from the link I posted above, it says, "M4025627 does not need P-1 factoring." Again, Pminus1 does work, but the bounds are lower than what have already been done.

Am I doing something wrong?

[James Heinrich]

Ah, I see what's going on with those particular examples. Taking M4,024,717 for example, GIMPS would assign TF to 2⁶¹ by default, then some P-1 afterwards, both at somewhat under 1GHz-day of work. However, someone has reported a massive amount of TF work on these exponents (as in 2000+GHz-days per exponent) taking the TF level up to 2⁷⁶. A side effect of increased TF level is that the probability of a P-1 factor for any given bounds decreases as the starting TF level increases, and because these exponents (~4M) have been so over-TF'd it makes the P-1 seem to have a low probability, even though the P-1 done is 10x more effort than GIMPS default.

It's for that reason that the page is supposed to work on "combined probability", but thanks for pointing out that I'd missed some part of that and exponents were being shown as "low P-1 probability" when in fact they've had a large amount for work done. I have fixed the page now and you'll see in the 4M range that there's not much left with a <10% probability.

Originally Posted by Jayder

It would be nice if I could search for exponents with less than 0.2 GHzd spent on P-1

That would be an easy search if I had that data stored in the database, but I don't (currently). It's not impossible to add, but it would involve modifying the table of ~24 million records and then calculating GHz-d values and updating all those rows, a non-trivial task.

[Jayder]

Originally Posted by James Heinrich

That would be an easy search if I had that data stored in the database, but I don't (currently). It's not impossible to add, but it would involve modifying the table of ~24 million records and then calculating GHz-d values and updating all those rows, a non-trivial task.

Thank you, James. I certainly don't expect you to undertake such a task. I appreciate your considering it.

If I'm set on spending a certain amount of GHzd, I guess I'll have to find another way. This is why I wanted to be able to search for exponents not meeting certain bounds.

Now that I think of it, we are now able to sort results, so it wouldn't be a monumental task to collect such numbers myself. I can easily set a range (4M-4.1M) and just sort by B1. I might be able to rig up a spreadsheet or something.

[Madpoo]

Originally Posted by Jayder

Thank you, James. I certainly don't expect you to undertake such a task. I appreciate your considering it.

If I'm set on spending a certain amount of GHzd, I guess I'll have to find another way. This is why I wanted to be able to search for exponents not meeting certain bounds.

Now that I think of it, we are now able to sort results, so it wouldn't be a monumental task to collect such numbers myself. I can easily set a range (4M-4.1M) and just sort by B1. I might be able to rig up a spreadsheet or something.

I wouldn't be opposed to setting up some kind of report to show "poorly factored exponents", it's just that I wouldn't really know where to begin on figuring out the criteria there... something that could be used universally by explorers such as yourself who like to pick those out and do a little extra factoring work of some kind (TF, ECM, P-1, whatever).

The tools at my disposal are just whatever the Primenet database stores on such things... TF factored depth is one thing, but when it comes to finding poorly factored ECM or P-1, I tend to nod off.

[James Heinrich]

Originally Posted by James Heinrich

It's not impossible to add, but it would involve modifying the table of ~24 million records and then calculating GHz-d values and updating all those rows, a non-trivial task.

I decided to add the information. I'm modifying the database structure now.

• modify table structure to add 4 new fields: (not sure, session crashed after about 30 mins)
• add indexes on 4 new fields: ~45 mins
• populate fields with actual data: weeks

[flagrantflowers]

Recent Results and Recent Cleared no longer have the ability to sort.

[Madpoo]

Originally Posted by flagrantflowers

Recent Results and Recent Cleared no longer have the ability to sort.

It worked for me. Make sure you give it time to fully load the entire set of data and enough time for the table sorter script to kick in. It can take a while with large data sets like that.

Let me know if you still have any issues with it after refreshing the page. If I had to guess, the full table hadn't loaded yet or something when you tried? I hope that's all it was anyway.

[ric]

A few moments ago, I was trying to access exponent details, and I got this:

Parse error: syntax error, unexpected ';' in C:\inetpub\v5\v5server\gimps\0.95_ar2_app.php on line 287

[James Heinrich]

Originally Posted by ric

Parse error: syntax error

Sorry about that, should be fixed now.