k=2000-3400 k's to be pulled from upcoming drives

[gd_barnes]

Attached is an Excel spreadsheet that is an analysis of the reservations currently in the k=2000-3400 area as they will relate to our upcoming drives. A link to this post will be put in the RPS threads for any comments.

Here is the thinking:
1. Reservations dormant for > 1 year are considered "expired" and will be included in our drives.
2. Reservations dormant for 6-12 months are being followed up with. I've now sent PM's to 3 people about them. A majority of those will likely be EXcluded from our drives.
3. Any reserved k's already fully searched to n=600K will be included in our drives as a double-check.

This leaves k's with the following conditions that will be EXcluded from our drives:
1. Have had activity in the last 6 months -or- have had activity in the last 6-12 months and a status update is provided.
2. Have not yet been searched to n=600K.

I am making an exception on k's that are close to n=600K; i.e. the RPS 7th drive k's, which are at n=565K. We'll include them as a double-check because there is a virtual 0% chance that we could get to n=600K before them. Regardless, I'll keep an eye on it. We will not cause a war there!

Karsten, for some reason, I thought you had unreserved k=3001-3009. Can you let me know about that? I have a question mark by them in the spreadsheet.

Right now, I believe this makes 31 k's out of 700 to be excluded...not too bad. The point is that at any one time, out of the many k's that are reserved, well less than half are active. Most have just not been unreserved by people who are no longer searching them.

I think this new drive will be a little more palatable to many people. By pulling out the k's ahead of time, there should be a smaller percentage of confirmed primes. Most of those will be from RPS's 7th drive but that's a small price to pay for such a nice juicy large range with plenty of k's that have never been touched in the top-5000 area.

I will edit this posting to post a new spreadsheet as more info. comes in from people on reservations. Like I did on the 1st drive, I'll then create an official posting of k's excluded and their status. The idea being that we'll double-check them about the time that the drives are winding down. That worked well on the 1st drive as I personally completed the double-check in the last 2 months shortly before our drive completed at the end of the year. There was only about 10-12 k's there so hopefully some folks won't mind helping me out this go around.

Whew; glad to have that arduous task off of my list.


Gary

[VBCurtis]

It looks like the sheet is posted in .xlsx format, Excel 2007 format? Perhaps a more general format would help, for those of us who don't buy microsoft products.

Steven is still working on the 2000's I reserved years ago, with three primes posted in 2009 from the sieve. I believe we're working on 2055, 2085, 2115, 2175 (2145 was given to gary in a long-ago era). I saw no status posted from him in last 6 months, thus my posting here.
-Curtis

[kar_bon]

Originally Posted by gd_barnes

Karsten, for some reason, I thought you had unreserved k=3001-3009. Can you let me know about that? I have a question mark by them in the spreadsheet.

yes, it's correct!
i'm updating that page with all primes found by FreeDC (so there's a mismatch in the table because i've not yet expanded the 'small' entries to the full columnsize).

i'll also sort that whole page by k, because they all reserved now.

hope to do so this weekend!

[henryzz]

Originally Posted by VBCurtis

It looks like the sheet is posted in .xlsx format, Excel 2007 format? Perhaps a more general format would help, for those of us who don't buy microsoft products.

openoffice 3 will read Excel 2007 format and earlier
personally i would recommend Excel 97-2003 format as more versions support it

[gd_barnes]

OK, I'll post it in Excel 2003 format shortly and edit this post when it is done.

Edit: The file has now been posted in Excel 97-2003 format.

[gd_barnes]

Originally Posted by VBCurtis

It looks like the sheet is posted in .xlsx format, Excel 2007 format? Perhaps a more general format would help, for those of us who don't buy microsoft products.

Steven is still working on the 2000's I reserved years ago, with three primes posted in 2009 from the sieve. I believe we're working on 2055, 2085, 2115, 2175 (2145 was given to gary in a long-ago era). I saw no status posted from him in last 6 months, thus my posting here.
-Curtis

Got him covered there. He was one I sent a PM to where his range status was about 8 months old or so. Those k's are being pulled out. He said he'd check the machines later today and provide a search range update.

[gd_barnes]

I have updated the spreadsheet to show Karsten's k=3001-3009 as unreserved per him. I also corrected a couple of other minor errors.

Curtis and others, attached is a text file that Karsten kindly created from my spreadsheet posting and sent to me. I'll keep the two in sync as more info. comes in.

[Flatlander]

For the double-check drive we insist on reliable, stress tested machines. Is there a danger that, by including double checking in this drive, we will increase the chance of errors unless we insist on the same restrictions?
Or will these 'double-checks' be followed up by future 'real' double-checks?

[henryzz]

Originally Posted by Flatlander

For the double-check drive we insist on reliable, stress tested machines. Is there a danger that, by including double checking in this drive, we will increase the chance of errors unless we insist on the same restrictions?
Or will these 'double-checks' be followed up by future 'real' double-checks?

we require reliable, stress tested machines on the doublecheck drive because we done have residues to compare the doublecheck to
for these doublechecks we will have have residues to compare with so we will know which results are wrong so we can triple-check the bad ones to find which result is good pretty much like GIMPS does

[gd_barnes]

Originally Posted by Flatlander

For the double-check drive we insist on reliable, stress tested machines. Is there a danger that, by including double checking in this drive, we will increase the chance of errors unless we insist on the same restrictions?
Or will these 'double-checks' be followed up by future 'real' double-checks?

Very good question. We "consider" these all to be "first pass" tests so that we can have something to compare to in the future when we do "real" double checks.

One other clarification: On the "double check" of the reserved k's that we are pulling out of the drive, those are also "considered" "first pass" tests unless they are large contiguous swaths of k's. See next para. for why it is considered such.

A question along these lines came up before related to results files that people could send us: Even though the results files may be available on some or many of the already searched k's, it's quite a bit more administrative hassle to attempt to acquire them from many sources and match them up with our sieve file to make sure they are all there than it is to just crunch them ourselves and consider it a first pass test. It came up before when I had a couple of people offer to send me results files on 2-3 k's that they had searched. Unless it is a large swath of k's, it's not worth pulling them out of our drive if we don't have to.

An example of a range that IS worth it is your k=3010-3200 effort with only 2 k's pulled out. That plus the mini drive is considered a full first pass test because it's a large almost contiguous range of k's. The same applied to Peter Benson's k=1005-1400 range for n=200K-500K. He was missing a small percentage of results so we ran those and considered it a full first pass test without including them in any main official drive.

On another topic: I just got the word from Thomas that he is unreserving his 25 k's in this range. I believe we are going to end up with pulling out just 6 k's from this drive! I'm more relieved than anything because the more k's that are pulled out, the bigger admin headache it is later on. I'll update the reservations spreadsheet and text file later on today.


Gary

[gd_barnes]

Originally Posted by henryzz

we require reliable, stress tested machines on the doublecheck drive because we done have residues to compare the doublecheck to
for these doublechecks we will have have residues to compare with so we will know which results are wrong so we can triple-check the bad ones to find which result is good pretty much like GIMPS does

This is correct but I believe doesn't quite answer what Chris was after. See my response.