mersenneforum.org  

Go Back   mersenneforum.org > Search Forums

Showing results 1 to 23 of 23
Search took 0.01 seconds.
Search: Posts Made By: ewmayer
Forum: Operazione Doppi Mersennes 2012-10-17, 21:15
Replies: 284
Views: 57,034
Posted By ewmayer
Not above 1T ... here is the small table I...

Not above 1T ... here is the small table I generated while searching for near-optimal params for runs around 1T:
Timing tests in 3 different ranges vs #sieve_prime, all with p = 43112609
(all run...
Forum: Operazione Doppi Mersennes 2012-10-17, 18:47
Replies: 284
Views: 57,034
Posted By ewmayer
Updated source (just the main-program .c file -...

Updated source (just the main-program .c file - headers same as before) attached to this message.

By way of clarification: For those doing deep sieving, since the code uses both stdout and stderr...
Forum: Operazione Doppi Mersennes 2012-10-11, 21:35
Replies: 284
Views: 57,034
Posted By ewmayer
Or a steady drinker. :) But what's my excuse?...

Or a steady drinker. :) But what's my excuse? (Here's one: I think the 2 and 8 in 2^8 somehow caused the nearby power of two 128 to pop into my head).


Aha - So the final details are that only...
Forum: Operazione Doppi Mersennes 2012-10-10, 22:51
Replies: 284
Views: 57,034
Posted By ewmayer
Exact result: z = 185*x^2 % m =...

Exact result: z = 185*x^2 % m = 57343615719861605804690122685

z0=z%b1;z1=(z>>25)%b0;z2=(z>>49)%b0;z3=(z>>73);

gives

z0 = 3418045
z1 = 14771380
z2 = 8030859
z3 = 6071491
Forum: Operazione Doppi Mersennes 2012-10-09, 23:49
Replies: 284
Views: 57,034
Posted By ewmayer
Aha - found the bug: In my Pari code above, the...

Aha - found the bug: In my Pari code above, the computation of the weights w0-w3 should begin with index j = 0, not 1. (I computed these individually in non-loop-style fashion when I first posted...
Forum: Operazione Doppi Mersennes 2012-10-09, 22:35
Replies: 284
Views: 57,034
Posted By ewmayer
I wanted to wait for your reply before commenting...

I wanted to wait for your reply before commenting on the results. Here is annotated Pari code (code in italics) - note that in the sample below I am so far only using the multiplier k to compute...
Forum: Operazione Doppi Mersennes 2012-10-09, 19:47
Replies: 284
Views: 57,034
Posted By ewmayer
Following your recipe, this is what I get for the...

Following your recipe, this is what I get for the weights (replacing FFTLEN with N for notational compactness):
j ceil[n.j/N] n.j/N w_j
- ----------- -------- --------------
0 0 0 2^0 = 1...
Forum: Operazione Doppi Mersennes 2012-10-09, 01:37
Replies: 284
Views: 57,034
Posted By ewmayer
OK, let's try a small example. Say I want to do...

OK, let's try a small example. Say I want to do multiplication with respect to the modulus m = k.2^n - 1. Specifically, let`s try k - 185 and n = 89, using transform length N = 4.

The wordsizes...
Forum: Operazione Doppi Mersennes 2012-10-08, 18:33
Replies: 284
Views: 57,034
Posted By ewmayer
No new factors found, then? My 10T run of k=185...

No new factors found, then? My 10T run of k=185 finished last week, sans factor.
Forum: Operazione Doppi Mersennes 2012-10-08, 00:23
Replies: 284
Views: 57,034
Posted By ewmayer
Thanks - that is interesting. But a couple of...

Thanks - that is interesting. But a couple of DWT-related points are not clear:

1. "Mersenne-like DWT on 2^{n+log k} ± c" - That should read log2(k), shouldn't it? That exponent is no longer an...
Forum: Operazione Doppi Mersennes 2012-10-05, 18:15
Replies: 284
Views: 57,034
Posted By ewmayer
Interesting - It seems as if it is doing an...

Interesting - It seems as if it is doing an implicit-DWT-mod, but one which allows only half as many input bits per word as does the Mersenne-mod DWT. Now that I see this, I seem to recall exchanging...
Forum: Operazione Doppi Mersennes 2012-10-04, 21:10
Replies: 284
Views: 57,034
Posted By ewmayer
That assumes that only one such double-width...

That assumes that only one such double-width FFT-mul is being done per powering bit processed - Unless there is a really sweet DWT-based implicit-mod method for such moduli (in which case I would...
Forum: Operazione Doppi Mersennes 2012-10-04, 18:45
Replies: 284
Views: 57,034
Posted By ewmayer
You got the gist of it right, but the work...

You got the gist of it right, but the work estimate is an order of magnitude too optimistic, because there is no fast DWT-method for multiply modulo such candidate factors, as there is for the...
Forum: Operazione Doppi Mersennes 2012-10-04, 02:47
Replies: 284
Views: 57,034
Posted By ewmayer
Since my code is currently one-k-only, I presume...

Since my code is currently one-k-only, I presume the user is keeping track of which k's have been done, need doing, etc.

For MM#47, I listed the 10 factors I found for the 30 k's surviving my...
Forum: Operazione Doppi Mersennes 2012-10-04, 00:04
Replies: 284
Views: 57,034
Posted By ewmayer
My code can start at any integer x > 0 (assuming...

My code can start at any integer x > 0 (assuming it's also < 2^64), and -- assuming the sieve-setup is working properly, which I have tested enough to have quite high confidence in -- will proceed...
Forum: Operazione Doppi Mersennes 2012-10-03, 02:14
Replies: 284
Views: 57,034
Posted By ewmayer
Here is the summary for my runs of the 30 k's...

Here is the summary for my runs of the 30 k's which survived the default (to ~32M for this size of MMp) factor sieve - These are grouped into the various residue classes k (mod 60) used by the...
Forum: Operazione Doppi Mersennes 2012-10-02, 06:48
Replies: 284
Views: 57,034
Posted By ewmayer
The ( k = 384 ) Example #2 factor in the C source...

The ( k = 384 ) Example #2 factor in the C source I first posted was a bogus one found due to the same bug I fixed in said source. (I just rechecked the alleged factor, 93934458313, via Pari, and...
Forum: Operazione Doppi Mersennes 2012-10-02, 03:30
Replies: 284
Views: 57,034
Posted By ewmayer
Code tarball (use tar -xvzf *.tgz to unpack) is...

Code tarball (use tar -xvzf *.tgz to unpack) is attached. This took a few hours longer than anticipated due to a bug discovered late in testing. The fix slows the code down ~10%; once I find a...
Forum: Operazione Doppi Mersennes 2012-10-01, 18:17
Replies: 284
Views: 57,034
Posted By ewmayer
Different MMp have different k which survive the...

Different MMp have different k which survive the standard many-k's small-prime sieve using in Mersenne TF. Also, different MMp have different allowable (k%60, p%60) combinations to begin with.

...
Forum: Operazione Doppi Mersennes 2012-09-30, 19:35
Replies: 284
Views: 57,034
Posted By ewmayer
I speeded up my hacked-together 2*k*M(p)+1 sieve...

I speeded up my hacked-together 2*k*M(p)+1 sieve by 10x over the past few days; am getting ~5e12 per day on p = 43112609 using 1 2GHz i5 core of my macbook.

Currently I am running k = 185 to depth...
Forum: Operazione Doppi Mersennes 2012-09-27, 19:11
Replies: 284
Views: 57,034
Posted By ewmayer
I hacked together a deeper version of my factor.c...

I hacked together a deeper version of my factor.c sieve (which is limited to primes < 2^32) into a small standalone code yesterday, and am doing some deeper sieving of the small-k candidates. Sample...
Forum: Operazione Doppi Mersennes 2012-09-26, 18:36
Replies: 284
Views: 57,034
Posted By ewmayer
Nice work on the sieve, guys - this matches my...

Nice work on the sieve, guys - this matches my own factor.c sieve results for the smaller-primes prescreening. For q's passing the small-prime sieve we definitely want to sieve very deeply (by the...
Forum: Operazione Doppi Mersennes 2012-09-23, 19:00
Replies: 284
Views: 57,034
Posted By ewmayer
I can say with 100% certainty that the smallest...

I can say with 100% certainty that the smallest prime factor of M(M43112609) is a world record prime, but without an explicit demonstration of such a factor, that is meaningless.
Showing results 1 to 23 of 23

 
All times are UTC. The time now is 15:37.


Sun Nov 28 15:37:23 UTC 2021 up 128 days, 10:06, 0 users, load averages: 1.21, 1.10, 1.13

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.