Riesel base 3 - mini-drive I

[gd_barnes]

Originally Posted by michaf

I have now left 380 sequences in my sieve file

Micha,

If you can coordinate with me via PM on how many k's are remaining in your sieved file, that will help. There's frequently going to be a difference between what you have and the files that I've posted here due to the time lag between when you remove them from your file and when I remove them from the public files.

It's only causing more confusion having a different # of sequences remaining then are what in the publicly posted files.

Originally Posted by Flatlander

27556870*3^35403-1
59149054*3^35500-1
are confirmed prime.

32-34 and 34-35.5 both finished.
Who do I email the results to? (Too big to post here.)

I meant that the files I just downloaded had 401 sequences, not 383 as stated in the first post. Not a problem. I'm running them as they are, there is too much chance of me making a mistake deleting that many sequences from two files. (And I can't be bothered.)

Chris

Chris,

Send to where Max stated.

No problem running them as is. But as stated in the last post, if you see I've updated your reserved file in post 1, you can use that starting from where you are at.


Gary

[gd_barnes]

Originally Posted by Flatlander

I think Gary is looking for the lowest prime for each k and you are working lower than me so ... ?

After the two ranges I am working on I will wait for the lower ranges to finish before reserving more files.

Originally Posted by mdettweiler

Yes, we are looking for the lowest prime for each k, so ideally if a prime is found and ranges lower than it are still in progress, the k's should not be removed from the lower-range files (though of course they can be removed from any higher-range files).

However, it can obviously be quite a hassle to re-insert work for k's after having removed them and continued processing on a file; Gary, what would you recommend in this particular situation?

Max

Chris,

No, please don't stop. Keep right on going. If you find a prime for a k that later on is found to have a lower prime, I'll remove the higher prime. It's no big deal. I don't want to hold anyone up.

Max,

Why would we ever re-insert anything? No one should be removing k's when a prime is found that is higher than what is being tested. The updated reserved files that I'm posting take this into account. That's why you see a different # of k's in the various updated reserved files. On those files, I do a close balancing of k's remaining at a specific n-depth by only looking at primes <= to that depth.

As an example, you should not have been removing any k's from your n=25K-27.5K file that had a prime for n>27.5K found by someone else. The only k's that you should remove would be the ones you found a prime for in your file.

BTW, I reworded your 2nd para. in the above quoted post for clarity. As previously stated, it could have an ambiguous or dual meaning.

All,

If it gets too confusing, I would just suggest using the file that you reserved with no removal or use any updated reserved file that I post, which will only have k's removed for lower than or equal to n-ranges. If there are more than one prime for a k, I'll see it and remove the higher prime for our listings. I cut-and-paste all primes to an Excel spreadsheet and use fancy formulas to make sure they are sorted right and that each k only has one prime posted.


Gary

[gd_barnes]

k's with primes have been removed from the appropriate files.

Chris, the new files that you took dropped from 401 k's to 358 k's remaining; nearly an 11% reduction. With such a large reduction, you might consider grabbing the files now posted for your reservation. If it takes you ~3 CPU days to process a file, that would be close to an 8-hour savings per file.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Max,

Why would we ever re-insert anything? No one should be removing k's when a prime is found that is higher than what is being tested. The updated reserved files that I'm posting take this into account. That's why you see a different # of k's in the various updated reserved files. On those files, I do a close balancing of k's remaining at a specific n-depth by only looking at primes <= to that depth.

As an example, you should not have been removing any k's from your n=25K-27.5K file that had a prime for n>27.5K found by someone else. The only k's that you should remove would be the ones you found a prime for in your file.

BTW, I reworded your 2nd para. in the above quoted post for clarity. As previously stated, it could have an ambiguous or dual meaning.

Ah, thanks for fixing my post. It seems that when michaf reported having only 380 k's or whatever in his file, I immediately thought "oh no! someone's running a lower file with higher-prime k's removed!", even though that's obviously not the case as michaf doesn't have a PRP range reserved. My bad.

Max

[Flatlander]

Originally Posted by mdettweiler

michaf doesn't have a PRP range reserved. My bad.

My bad! I was confusing michaf with henryzz!

95248136*3^35651-1
51339798*3^35704-1 (51339798*3^27498-1 already found)
43958762*3^35799-1
25224356*3^35851-1
41654926*3^36038-1
63849262*3^36062-1
49679678*3^37289-1
26377066*3^37526-1
87090628*3^37592-1 (87090628*3^26732-1 already found)
Confirmed primes.

[henryzz]

47249008 28513
18094844 28679
49728916 28702
41032748 28741
21390898 28788
54386588 28793
68257172 28835
18914054 28836
62874334 28956
37858876 28957
64359712 28987
69237556 29031
30295424 30921
82997968 30976
62605838 31166
30950858 31258
76765594 31287
61142552 31290
59754124 31303
are prime

[mdettweiler]

Originally Posted by henryzz

47249008 28513
18094844 28679
49728916 28702
41032748 28741
21390898 28788
54386588 28793
68257172 28835
18914054 28836
62874334 28956
37858876 28957
64359712 28987
69237556 29031
30295424 30921
82997968 30976
62605838 31166
30950858 31258
76765594 31287
61142552 31290
59754124 31303
are prime

Note on k=37858876 and 30950858: henryzz's primes for those k's were lower than the two previously found by Flatlander, so I removed Flatlander's primes for those respective k's from the first post.

[Flatlander]

Taking:
38.5K-40K
40K-41K

[Flatlander]

96398168*3^36432-1
83655958*3^36512-1
89745514*3^36536-1
97229752*3^37767-1
21106738*3^37993-1
47177978*3^40029-1
Are prime. (But I haven't compared them with primes already found.)

[mdettweiler]

Originally Posted by Flatlander

96398168*3^36432-1
83655958*3^36512-1
89745514*3^36536-1
97229752*3^37767-1
21106738*3^37993-1
47177978*3^40029-1
Are prime. (But I haven't compared them with primes already found.)

I just checked them--they're all new primes.

[gd_barnes]

Nice work guys. Everything looked perfect! Max, if you wouldn't mind keeping the listing of primes in descending n-value sequence, that helps balance the k's remaining for each posted file. I've corrected it now.

All k's with primes have now been removed from all appropriate files.

About my estimate of slightly over 200 primes for this drive: I used the Sierp base 3 drive for the estimate and forgot that that drive started at n=35K instead of 25K like this one did.

Based on that, I would expect ~240-250 k's to be eliminated with this drive. The target will be to get to < 200 k's remaining for k<100M at n=100K.

Thanks to everyone for the great participation on this drive. I'd also like to encourage people to work on finishing up the Sierp base 3 mini-drive. It's at n=86K and we only need to get to n=100K to finish it up. Although tests take 4 times as long at n=88K vs. n=44K, we're only doing k<50M there so an n=1K file at n=88K should take about twice as long as an n=44K file here, i.e. like reserving 2 files here.

Here, once we push past n=50K, possibly near n=60K, I'll reduce the file-size again to n=500 files.


Gary