mini-drive for high-n testing on Sierp base 4

[gd_barnes]

Originally Posted by em99010pepe

Impossible! It's a CPU limitation, the real problem of the false quad-core.




If you want I can easily clean those ranges in a few days....just let me know...

I'm asking people to take no more than 2 files at a time on the drive. You could grab two files to get it started, then when done, grab another 2, etc. I have a file up to n=200K that just needs to be divided up for n>120K so there are plenty to go around.

I'm trying to find a good balance between 'spreading the wealth' around and getting things searched quickly.


Gary

[em99010pepe]

Originally Posted by gd_barnes

I'm asking people to take no more than 2 files at a time on the drive. You could grab two files to get it started, then when done, grab another 2, etc. I have a file up to n=200K that just needs to be divided up for n>120K so there are plenty to go around.

I'm trying to find a good balance between 'spreading the wealth' around and getting things searched quickly.


Gary

2 files will be done in less than 12 hours in case I move all cores...I suggest you start adding more files....lol

[gd_barnes]

Originally Posted by em99010pepe

2 files will be done in less than 12 hours in case I move all cores...I suggest you start adding more files....lol

LMAO!!

Well...can you just move 2 cores to it and put 2 cores on the Riesel side? Both are in top-5K range now.


G

[em99010pepe]

Originally Posted by gd_barnes

LMAO!!

Well...can you just move 2 cores to it and put 2 cores on the Riesel side? Both are in top-5K range now.


G

I prefer to concentrate on only one side. For now I'll help to clean this mess here, I still have to move another core when tomorrow I get to work...lol

[gd_barnes]

Originally Posted by em99010pepe

I prefer to concentrate on only one side. For now I'll help to clean this mess here, I still have to move another core when tomorrow I get to work...lol

OK, I'll open it up as the exponents get a little higher and the interest begins to wane.

[tcadigan]

actually at one point there was a larger sieve file. jasong had started it, and had even emailed it to me at some point. but I seem to have lost it in both in my email and on my computer (from the hardware crash).

if I remember correctly it wasn't sieved very deeply yet but it had been at least started. sorry about that, back to square one once we reach the 1 mil mark

[gd_barnes]

Originally Posted by tcadigan

actually at one point there was a larger sieve file. jasong had started it, and had even emailed it to me at some point. but I seem to have lost it in both in my email and on my computer (from the hardware crash).

if I remember correctly it wasn't sieved very deeply yet but it had been at least started. sorry about that, back to square one once we reach the 1 mil mark

I'm not sure how to take 'larger'. Have all candidates been tested up to the range that we are starting at?

The P-value in the sieved file that you send is misleading. It only shows P=1G. I tried sieving for several hours starting at P=1G, 10G, 100G, and 1T and no factors were found. So I'm assuming that it was at least sieved past P=1T.

With the few candidates remaining, sieving would be a waste of time at this point. LLRing takes about 1.75-2.5 hours per candidate. It's unlikely that we'd get that kind of removal rate from just 900+ remaining candidates even if we found the correct point that it had been sieved to. It's faster just to test them out at this point.

Alas, we'll start anew at n=1M.


Gary

[tcadigan]

sieving what is left at this point definitely isn't worthwhile, I looked into that. what I meant by there was a larger one is:

jasong: "I'm sieving n=1 million to 5 million(base 4), which ultimately means all even values for base-2 from 2 to 10 million."

furthermore, jasong: "As soon as I get to 2000G, I'm probably going to be moving on to something else. It would be nice if someone could take over handing out the file."

(both gathered from the original base 4 thread)

I got that new dat file right after that, but now seem to have lost it.

as for the *current* file, i.e. the one that we have right now going up to 1 million (base 4) that has been sieved to 7.5T. iirc the program that I used to split up a dat file into the corresponding .npg files just puts a (mostly) bogus top line on them.

rest assured that when Jean's reservations and this one get to 1 mil it'll be an even playing field again and all of the candidates will be tested up to the 1 mil mark

[gd_barnes]

Originally Posted by tcadigan

sieving what is left at this point definitely isn't worthwhile, I looked into that. what I meant by there was a larger one is:

jasong: "I'm sieving n=1 million to 5 million(base 4), which ultimately means all even values for base-2 from 2 to 10 million."

furthermore, jasong: "As soon as I get to 2000G, I'm probably going to be moving on to something else. It would be nice if someone could take over handing out the file."

(both gathered from the original base 4 thread)

I got that new dat file right after that, but now seem to have lost it.

as for the *current* file, i.e. the one that we have right now going up to 1 million (base 4) that has been sieved to 7.5T. iirc the program that I used to split up a dat file into the corresponding .npg files just puts a (mostly) bogus top line on them.

rest assured that when Jean's reservations and this one get to 1 mil it'll be an even playing field again and all of the candidates will be tested up to the 1 mil mark

OK, thanks. Sounds good.

In the mean time, n=885K-890K is complete; no primes.


Gary

[em99010pepe]

Here's the deal Gary...can I reserve all the files? Tomorrow I'll reserve another chunk of 4 files, then I'll wait 2 days to reserve the rest....

[gd_barnes]

Originally Posted by em99010pepe

Here's the deal Gary...can I reserve all the files? Tomorrow I'll reserve another chunk of 4 files, then I'll wait 2 days to reserve the rest....

Carlos,

On this mini-drive, I'm giving no restriction other than what I stated above; that is "Feel free to take whatever files you can do in a resonable time frame". So sure, reserve them all if you want to. I personally am not all that found of such high-n searches. And based on Tcadigan's statements, he had also grown tired of the effort.

Back to drives #1 and #2...I've thought quite a while about this. I've decided to open them up. People can take as many files as they have cores to process them. See the "Edit: Feel free to take one file for each core that you have available to run the searches." statement that I just now put in each drive.

What I'll do in the future (like I've essentially done here) is somewhat limit things at the low ranges of n on drives and then as we get past n=400K base 2, I will open them up to the 'heavy hitters' such as yourself to process many files at a time. I think that will strike a good balance for everyone.

In the case of this mini-drive, of course you're welcome to reserve them all now. But if I had to give a choice, I feel that this one is a little lower priority than drives #1 and #2. I personally would like to see you knock out 5-6 files (or how many ever cores you have available) in Sierp base 16 but feel free to knock this one out first. With n=2K ranges on Sierp base 16, each file will take quite a bit longer than when we were at n<100K.

Thanks for throwing your 'mean machines' at our project! We are all about flexibility here.


Gary