Mersenne Digest       Wednesday, August 30 2000       Volume 01 : Number 773




----------------------------------------------------------------------

Date: Mon, 28 Aug 2000 13:29:24 +0100
From: gordon spence
Subject: Mersenne: Credits per FFT Size

>Marc Getty wondered previously
>
>How does the CPU time contributed get calculated? I would assume that 
>there is a
>standard credit for each FFT size, but I can't find what that credit is 
>anywhere
>on mersenne.org.
>
>Can anyone help fill in the following?

Taking the data from the status page on mersenne.org where it gives timings 
per fft size per iteration, taking the mid-point exponent for each fft size 
and then multiplying by the 5.5 conversion factor that George quotes gives 
the following

FFT    Exponent         Credit (P90 years)
160     2972750 0.043
192     3612500 0.062
224     4266500 0.089
256     4924000 0.113
320     5882500 0.177
384     7122500 0.262
448     8375000 0.368
512     9670000 0.474
640     11575000        0.751
768     14050000        1.110
896     16560000        1.548
1024    19125000        2.001
1280    22865000        3.094
1536    27715000        4.515
1792    32600000        6.328
2048    37675000        8.056
2560    45075000        12.861
3072    54650000        18.967
3584    64200000        26.648
4096    74150000        33.675

And yes there are a few of us using fft's in excess of 1792, like up in the 
4096 range.....so when my exponent gets done in 2004 I should get about 36 
years of credit in one hit  <G>

regards

G




_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Mon, 28 Aug 2000 16:38:30 +0100
From: "Andy"
Subject: Mersenne: Tips on compiling source

I'm trying to compile the source code for Prime95 for a wintel machine and
am slightly lost for how to do so. I have Visual C++, gcc, a86 and just
about everything else.

My background is Java development (3 1/2 years) so I am not used to header
files and assembly language. Though do know the fundamentals.

Any help would be great.

Also any programming tutorials etc (online or hard copy) on assembler would
be great.

Andy

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Mon, 28 Aug 2000 11:56:27 -0500
From: (Mikus Grinbergs)
Subject: Re: Mersenne: Credits per FFT Size

On Mon, 28 Aug 2000 13:29:24 +0100 gordon spence wrote:
>
> Taking the data from the status page on mersenne.org where it gives timings
> per fft size per iteration, taking the mid-point exponent for each fft size
> and then multiplying by the 5.5 conversion factor that George quotes gives
> the following
>
> FFT    Exponent         Credit (P90 years)
> 256     4924000         0.113
> 320     5882500         0.177

Interesting data.  I'm running a K6-III-400, whose other chief
CPU-burners are a software MIDI synth and an occasional compile,
and which is halted from time to time for software installs.

Upon exponents LESS than the mid-point, my measured elapsed time
was  0.057 years  and  0.088 years,  respectively.

I see my K6-III-400 is  TWICE  as fast as the reference P90 !!!

mikus

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Mon, 28 Aug 2000 11:16:47 -0700
From: Eric Hahn
Subject: Re: Mersenne: CPU Time Credit Calculation

Marc Getty wrote:
>
>How does the CPU time contributed get calculated? I would assume
>that there is a standard credit for each FFT size, but I can't
>find what that credit is anywhere on mersenne.org. 

Actually, the formula isn't based on FFT size.  To get a good
estimate of how much time you'll get credited for any given
exponent, you'll need to multiply the exponent minus 1 by the
average time it takes to do each iteration.  Then multiply
by the number of times faster the PC is than a P-90.  Finally
the % of a year you get is the approx. amount of time credited.

For example: 
  On the exponent 5,593,943...
  Multiply 5,593,942 by .275 to get 1,538,334 seconds.
  Multiply 1,538,334 by 3.5 (for a PII-266) to get 5,384,169 secs.
  5,384,169 seconds is 17.073% of a year...
  So you'll get approx 0.171 P-90 years credit

>I doubt that anyone out there is using more then a 1792 K FFT.

Well....  There are actually a few of us brave souls out there
who are QA'ing exponents in FFT sizes all the way up to 4096K!!
Based on my estimates, the exponent I'm QA'ing will provide me
with ~36.348 P-90 years of credit when completed.

Eric


_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Mon, 28 Aug 2000 23:16:34 +0200
From: "Steinar H. Gunderson"
Subject: Mersenne: Re: Tips on compiling source

On Mon, Aug 28, 2000 at 04:38:30PM +0100, Andy wrote:
>I'm trying to compile the source code for Prime95 for a wintel machine and
>am slightly lost for how to do so. I have Visual C++, gcc, a86 and just
>about everything else.

You're aware you need NASM as well?

That's at least the problem I faced when I tried to compile it myself. There
should be reasonable object files included that you can link to, though...

/* Steinar */
- -- 
Homepage: http://members.xoom.com/sneeze/
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Tue, 29 Aug 2000 09:03:37 +0200
From: Henk Stokhorst
Subject: Mersenne: Prime95 and Pentium III 1.13 GHz

Hi,

for all of you who couldn't get their 1.13 GHz Pentiums working
correctly with Prime95, there is a story on that at

http://www.tomshardware.com/cpu/00q3/000801/pentiumiii-01.html

YotN,

Henk Stokhorst

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Tue, 29 Aug 2000 14:20:12 +0000
From: Alexander Kruppa
Subject: Re: Mersenne: Tips on compiling source

Andy wrote:

> I'm trying to compile the source code for Prime95 for a wintel machine and
> am slightly lost for how to do so. I have Visual C++, gcc, a86 and just
> about everything else.

The Linux version should compile without too much hassle, but you have to
remove the calls to the SEC5 function from the code, or put a security.h file
in the source base directory in which you

#define SEC5(a,b,c) 0L

In fact, this might be done automatically in the makefile, you can replace

[ ! -e ../security.h ] && touch ../security.h || true

by

[ ! -e ../security.h ] && echo "#define SEC5(a,b,c) 0L" >../security.h || true

(George, can this go the the makefile in source.zip ?)

Do not "make clean" in linux/, this will remove all the FFT object files! Only
remove prime.o, primenet.o and menu.o . Also, the GIANTS.H file is all
uppercase in the zip file so case-sensitive unix does not find it. Rename it
to giants.h .

I've been experimenting with the ECM code for a while now, trying to add
something to allow B2>2^32 and maybe computing the group order, so I've
compiled the source rather a lot recently. If you have any specific questions,
feel welcome to ask me, maybe I've run into the same problems, too.

Btw, I have never tried to compile the windows version, but the documentation
of the source mentions that this might be difficult to do.

Ciao,
  Alex.

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Tue, 29 Aug 2000 10:15:16 -0400
From: Jeff Woods
Subject: Re: Mersenne: Prime95 and Pentium III 1.13 GHz

Intel issued a RECALL of *all* 1.13GHz Pentium 3's yesterday:

http://news.cnet.com/news/0-1003-200-2634198.html

At 09:03 AM 8/29/00 +0200, you wrote:
>Hi,
>
>for all of you who couldn't get their 1.13 GHz Pentiums working
>correctly with Prime95, there is a story on that at
>
>http://www.tomshardware.com/cpu/00q3/000801/pentiumiii-01.html
>
>YotN,
>
>Henk Stokhorst
>
>_________________________________________________________________________
>Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
>Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Tue, 29 Aug 2000 23:19:55 -0700
From: Spike Jones
Subject: Mersenne: smallest possible factor

A few weeks ago, I thought someone posted something like:

2^n-1 where n is prime cannot have any factor smaller than n.

Did I get that right?  Is there a simple proof?  spike

_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Wed, 30 Aug 2000 12:03:41 +0000
From: Alexander Kruppa
Subject: Re: Mersenne: smallest possible factor

Spike Jones wrote:

> A few weeks ago, I thought someone posted something like:
>
> 2^n-1 where n is prime cannot have any factor smaller than n.
>
> Did I get that right?  Is there a simple proof?  spike

Factors of a mersenne number Mp are always of the form f=2*p*k+1, k may be
as small as 1.
The proof that factors are 2kp+1 is not simple as far as I remember and uses
the theory of quadratic residues (and thus I didn't understand it). See it
on Chris Caldwells (superb) page on Prime numbers,
http://www.utm.edu/research/primes/ .
The proof is at http://www.utm.edu/research/primes/notes/proofs/MerDiv.html

Ciao,
  Alex.


_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

Date: Thu, 31 Aug 2000 07:58:23 +0200
From: Guillermo Ballester Valor
Subject: Mersenne: Glucas beta version released. 

Hi to all:

After a long and hot summer, I've finally written a first beta version
of Glucas. This is a C-program to make L-L test. Still, there is no good
scripts to make the binaries in a easy way. I need the help of testers
to make fast binaries and fix possible bugs. 

Obviously, for intel platforms, his 'big brother' mprime is faster. In
alphas is about as fast as Mlucas. I don't know how well (or bad) it
runs other machines, I'm waiting your news ;-).

To see how you can help, and read first:

        ftp://209.133.33.182/pub/valor/README.htm

to see more details from Glucas:

        ftp://209.133.33.182/pub/valor/README.Glucas.htm

To download the source:

        ftp://209.133.33.182/pub/valor/Glucas-1.98.tar.gz

Thanks to Ernst Mayer for all his help, specially for the use of his
server. 

Regards

Guillermo.
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

------------------------------

End of Mersenne Digest V1 #773
******************************

Reply via email to