Mersenne Digest Saturday, April 17 1999 Volume 01 : Number 546
----------------------------------------------------------------------
Date: Thu, 15 Apr 1999 10:51:55 GMT
From: "Brian J Beesley" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: RE: Mersenne Digest V1 #544
> [Brian:]
> > Interesting. rpcnet.dll from the v18 distribution is much
> > smaller than that in the v17 distribution.
> >
> > Should be safe enough to keep the v17 distribution copy.
>
> Yes, but it the v17 version uses a proxy running on the old PrimeNet
> 3.1 server's box. I'd rather everyone use HTTP if possible, or at
> least use the updated v18 version dated 4/12/1999.
Ooops, sorry... I presumed that something had gone wrong which
was making the v18 rpcnet.dll misbehave and had a much smaller
file size as an indirect result ...
>
> > Actually my systems are all using either http or the special
> > rpcnet.dll used to connect to the PrimeNet Proxy server, so I just
> > don't know how badly the v18 rpcnet.dll is broken.
>
> The v18 program defaults to HTTP when you first install it, so new
> users should not run into it. The PrimeNet FAQ page also describes
> how to handle the RPC run-time library crash situation.
>
> I've updated the posted v18.1 zips with a new RpcNet.dll. I couldn't
> get it to crash. If you have an environment that can test this,
> please do so and tell me how it went.
I don't have a suitable environment, as I explained above. Perhaps
other people could have a go.
BTW, Scott, when you post new versions of software, could you
please either mail me or change the "last updates" date on the
PrimeNet News page. I didn't know that the files had been changed
until I read this message, hence by ftp mirror was still serving the
version with the "old" v18 rpcnet.dll. (I've just pulled the latest files
onto the mirror. Sorry to anyone who's been fetching over the last
few days)
Regards
Brian Beesley
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 17:05:43 +0200 (CEST)
From: Henrik Olsen <[EMAIL PROTECTED]>
Subject: RE: Mersenne: Re: Factoring & bugs
On Thu, 15 Apr 1999, Foghorn Leghorn wrote:
> >From: Paul Leyland <[EMAIL PROTECTED]>
> >You are merely restating a law of nature. After a point, everything
> >becomes useless.
>
> I am reminded of a quote from Homer Simpson: "Trying is the first
> step toward failure." :)
>
> A question for George (and Scott): Is there any chance that Prime95's
> ECM factoring will ever become automated as a part of PrimeNet? Even
> if it is never given as a default type of assignment, it would still
> be useful to dedicated number theory enthusiasts who want to run it
> on more machines than they can manage manually.
For automated ECM factoring you might want to have a look at the ECM
client/server setup at http://www.interlog.com/~tcharron/ecm.html instead,
once you get it running it works quite well.
- --
Henrik Olsen, Dawn Solutions I/S URL=http://www.iaeste.dk/~henrik/
A Pentium is a terrible thing to waste, http://www.mersenne.org/prime.htm
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 11:13:04 -0400
From: Joth Tupper <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Question & Suggestion
Message text written by George Woltman
>
At 01:02 PM 4/12/99 -0400, you wrote:
>It also seems that the number of sumout errors [that is, sum(inputs) !=
>sum(outputs) ]
>on my AMD K6-2 400 has dropped to zero in the past few days [I think] from
>about
>1 every 2 to 3 hours late last week.
My bet would be overheating or flaky memory - but I certainly
cannot prove that.
<
I understand that hardware problems certainly cause sumout errors.
If I had an overheating CPU or flaky memory, I would expect the problems to
continue but
they seem to have stopped for the moment. This is indistinguishable from
an intermittant hardware problem like a separating circuit board (or other
component).
Oh, well, and thanks.
Joth
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 17:06:57 +0100
From: Robin Stevens <[EMAIL PROTECTED]>
Subject: Mersenne: Performance hit on Pentiums with >64MB?
I currently have a P200MMX with 64MB running Primenet under Linux. Finding
64MB rather limiting on my desktop workstation, I'm looking to equip the
machine with a couple of 64MB SIMMs.
However I shall be running into the limitation of most Pentium motherboards
in that the machine will be unable to cache more than 64MB of RAM. How
much of a performance hit am I likely to encounter in running Primenet?
I've seen figures of 10-30% quoted for various applications.
Not that I'm overly bothered - I have a PII contributing rather more these
days :-)
- --
- -------------------- Robin Stevens <[EMAIL PROTECTED]> --------------------
Merton College, Oxford OX1 4JD, UK http://www-astro.physics.ox.ac.uk/~rejs/
"Never attribute to malice what can be explained by incompetence."
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 09:17:38 -0700
From: "Scott Kurowski" <[EMAIL PROTECTED]>
Subject: Mersenne: RE: Mersenne Digest V1 #545
Hi all,
[Aaron:]
> I had a question from my brother who is interested in
> getting some of his Unix boxes up and running. He was
> dismayed when he couldn't find any versions that would
> use Primenet for automatic assignment.
>
> So, I thought I'd ask, has anyone done any ports for the
> following that use Primenet, or should I tell him to
> just use the manual testing page at entropia:
I've been working with developers of several client ports.
Chris Smith is about 90% done with a PrimeNet client for UNIX and
Alphas based on MacLucasUNIX. We'll probably start testing with the
live server in early May. Until then, the manual testing page is
about it.
Richard McDonald is completing PrimeNet clients for Macs. That has
been doing well, too, maybe 60% done.
Nick Hilliard has been updating MPrime with FreeBSD improvements.
That might be ready for public testing as early as next week.
[George:]
> Note that when the next version of prime95 comes out, everyone
> will take a hit as the new faster timings are applied. Use the
> page as a rough indicator of where you stand relative to other
> GIMPS members. Don't take it as an overly accurate accounting
> of every CPU cycle you have invested.
Everyone please note this will not impact your accumulated PrimeNet
account CPU statistics.
[Foghorn:]
> A question for George (and Scott): Is there any chance that
> Prime95's ECM factoring will ever become automated as a part
> of PrimeNet?
PrimeNet already collects ECM results in its logs, but otherwise does
nothing with them; it is already equipped with almost everything else
necessary. Entropia.com has a standing offer to GIMPS to automate ECM
with PrimeNet, but I'm guessing George's priorities are elsewhere
right now (v19).
Best regards,
scott
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 15:06:41 -0000
From: "Paul Irish" <[EMAIL PROTECTED]>
Subject: Mersenne: Proth factoring problem.
hey all. i am very excited about finding a new math discovery.
i wanted to find a factor of fermat's number 20. - F20
so.. using Proth.. i set it up with
Start: For n=1048576 to 1048576, For k=2 to 2 step 2, GFN.
1048579 being 2^20.
after 3 days of processing on my Intel 300 Mhz..
it came out with this result in the log file..
1*2^1048576 + 1 is composite. (a = 3)
what is the relevance of a?
and HOW can i use my computer to find a factor of F20?
thank you.
- -paul
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 16:08:42 -0400
From: George Woltman <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Proth factoring problem.
Hi Paul,
At 03:06 PM 4/15/99 -0000, Paul Irish wrote:
>i am very excited about finding a new math discovery.
Aren't we all :)
>i wanted to find a factor of fermat's number 20. - F20
>so.. using Proth..
Questions regarding Proth should probably be directed to
the mailing list that Chris Caldwell runs on prime numbers.
Maybe, he'll post the web address so that you and others can join.
>and HOW can i use my computer to find a factor of F20?
Prime95 version 19 should be able to use ECM to find a factor of F20.
It is due out in a few months. You could try finding new factors
of the smaller Fermat numbers with the existing prime95.
There are other ways you could make a discovery. You could run
ECM on any of the Cunningham numbers. In general, the harder it
is to make the discovery, the more "famous" you'll be. Thus, don't
expect it to be easy to find a Fermat factor, but good luck
trying!
Best regards,
George
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 14:20:16 -0600
From: "Aaron Blosser" <[EMAIL PROTECTED]>
Subject: Mersenne: Java client?
Hello,
Well, my brother is really getting into the prime number thing.
He has it running on his Ultra Sparc but, as expected, it is only running in
32 bit.
He had mentioned the possibility of porting the MacLucasUnix code, in
particular, to Java...in theory anyway, the Java runtime engine on the Ultra
Sparc *is* 64 bit capable, so he's hoping for better performance, plus since
it's Java, you could run it on your Windows CE device, or anything else with
a Java engine.
Is there a demand out there for a Java port? It wouldn't be as fast as C or
ASM for most platforms, but for platforms with NO port at all, I think this
would be great.
Comments anyone?
Aaron Blosser
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Fri, 16 Apr 1999 00:04:51 +0200
From: "Yves Gallot" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Proth factoring problem.
you "just" proved that
3^(2^1048576) != 1 (mod 2^1048576 + 1)
then F20 is composite (Fermat theorem).
It doesn't help to find a factor.
I tried to find a factor of F22, using Pollard p-1 method and reached
B=2,000,000 without finding a divisor.
Does anyone know current limit of Pollard p-1 method for F20 ? Because
divisors of Fermat numbers minus one are divisible by 2^(n+2) a large
attempt with Pollard p-1 should be tried before using ECM !?
Yves
>hey all. i am very excited about finding a new math discovery.
>i wanted to find a factor of fermat's number 20. - F20
>so.. using Proth.. i set it up with
>Start: For n=1048576 to 1048576, For k=2 to 2 step 2, GFN.
>1048579 being 2^20.
>after 3 days of processing on my Intel 300 Mhz..
>it came out with this result in the log file..
>1*2^1048576 + 1 is composite. (a = 3)
>what is the relevance of a?
>and HOW can i use my computer to find a factor of F20?
>thank you.
>-paul
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 18:45:18 -0700
From: Greg Hewgill <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Java client?
On Thu, Apr 15, 1999 at 02:20:16PM -0600, Aaron Blosser wrote:
> Is there a demand out there for a Java port? It wouldn't be as fast as C or
> ASM for most platforms, but for platforms with NO port at all, I think this
> would be great.
You may be surprised at just how fast a Java implementation could be. I did
essentially the same thing for the distributed.net RC5-64 effort (see
http://www.hewgill.com/rc5/). It turned out that the Java decryption routine,
with a JIT compiler, could reach 25% of the speed of a hand optimized assembler
decryption routine.
Greg Hewgill
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 20:16:38 -0600
From: "Aaron Blosser" <[EMAIL PROTECTED]>
Subject: RE: Mersenne: Java client?
> On Thu, Apr 15, 1999 at 02:20:16PM -0600, Aaron Blosser wrote:
> > Is there a demand out there for a Java port? It wouldn't be as
> fast as C or
> > ASM for most platforms, but for platforms with NO port at all,
> I think this
> > would be great.
>
> You may be surprised at just how fast a Java implementation could
> be. I did
> essentially the same thing for the distributed.net RC5-64 effort (see
> http://www.hewgill.com/rc5/). It turned out that the Java
> decryption routine,
> with a JIT compiler, could reach 25% of the speed of a hand
> optimized assembler
> decryption routine.
Well, my brother went ahead and got it coded and working okay. According to
his tests, for larger exponents it actually begins to approach the speeds of
the C port.
He's doing some work on it to optimize it now and promises to have it
multithreading in no time. Hmm. His initial timings were based on the
Ultra Sparc running MacLucasUNIX. For example, M(3217) was only 17% slower
with Java than with the C code.
Aaron
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 21:23:12 -0700
From: "John R Pierce" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Question & Suggestion
> That sure sounds simple enough. Now I just have to determine if the
> sumouts I am suddenly seeing
> are really hardware. Hmmm.
I was getting a reproducable sumout error with Win9X but not NT on multiple
systems that had MMX extensions if and only if I ran a web plugin called
"Crescendo Live" which is a sort of streaming MIDI player
(http://www.liveupdate.com). So, they can be caused by software as well as
hardware. I tried to isolate just what was going on, but didn't have any
luck. NT was immune.
I'm gonna bite the bullet right now and see if it recurs with the current
prime95 and Crescendo plugin... Ah, Crescendo 4.0 and win98 'special
edition' beta 3 (build 2183) combined with Prime95 18.1.1 seem 'safe' on my
pentium-II 300MHz system. Crescendo's music playback is eating over 50% of
the CPU tho, sheesh.
- -jrp
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 12:33:58 -0500
From: [EMAIL PROTECTED] (Mikus Grinbergs)
Subject: Re: Mersenne: Question & Suggestion
> >It also seems that the number of sumout errors [that is, sum(inputs) !=
> >sum(outputs) ]
> >on my AMD K6-2 400 has dropped to zero in the past few days [I think] from
> >about
> >1 every 2 to 3 hours late last week.
> My bet would be overheating or flaky memory - but I certainly
> cannot prove that.
Had a strange experience - ONE instance of a sumout error in months
and months and months of running prime. The error occurred at 5 am
local time, when prime was the *only* program running, and no other
programs had started or ended since the previous evening. And *no*
occurrences of any reported prime errors since!
All I can think of is that there must have been a "brown-out" in the
utility electric supply - if the electricity had been cut off, or had
a voltage spike, my (*not* continuous supplement) UPS should have
taken over (and it usually beeps loud enough to wake me up).
mikus
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Fri, 16 Apr 1999 02:51:14 -0500 (CDT)
From: Conrad Curry <[EMAIL PROTECTED]>
Subject: Mersenne: NFS factorizations
C(2,601-) Factored
- ------------------
M601 has been completely factored by the Special Number
Field Sieve (SNFS). It was previously known that
M601 = 3607 * 64863527 * c170
The factor 3607 was found by Riesel in 1957 and the factor
64863527 was found by Brillhart and Johnson in 1960. The
c170 is a 170 digit composite number given by
c170 = 354715929158469716117612845223357182401467887957184539934\
827018425775054703022731713537808507752796939456788348401\
72667458519819530578294315203146051414787170811175493759
On April 15, 1999 it was found that c170 = prp60 * prp110, where
prp60 = 6433907608968262728069255630596126855090392650310674568\
49993
prp110 = 5513226964341686180894070234940226491991504154775763376\
9331818430511033108848146620408728664776308505356579463
The factorization of M601 was 'Most Wanted' by the Cunningham
project. It was also the smallest number of the form 2^n-1 whose
complete factorization was not known. That distinction now goes to
M617.
The sieving was done with idle time on wiglaf, a beowulf with
32 233MHz PII processors. 9.5 million relations were collected
requiring 850 Mb in uncompressed ASCII format. The resulting matrix
size was 1342267 x 1364566. The linear algebra and square-root phases
were done at Centrum voor Wiskunde en Informatica (CWI) by Peter
Montgomery. The factorization was found on the fourth dependency.
Acknowledgments are due to Bob Silverman for his NFS code and
help and to Peter Montgomery for his help. Also to CWI and the School
of Mathematical Sciences at the University of Southern Mississippi for
the use of their computers.
C(2,592+) Factored
- ------------------
CWI announces the factorization of C(2,592+):
As of April, 1999 2^592 + 1 is the largest number factored by SNFS
(Special Number Field Sieve) on a single machine. This machine is medusa,
an SGI Origin 2000 with 24 250 Mhz processors at CWI (Centrum voor
Wiskunde en Informatica) in Amsterdam.
Sieving took 115 processor-days using 8 processors March 3-20, 1999.
for an 18-day period. A full filesystem delayed us for a week.
Linear algebra on a 1709640 x 1724589 matrix took 9.5 days
on one medusa processor between March 29 and April 11, 1999.
The square root step took under an hour.
C(2,592+)
* Algebraic factor 2^16 + 1 = 65537 already known
* c174 = prp74 * prp100 SNFS CWI
31474859175390153735107856059776876864094488913562623409377402922304455233
7857910537107244650997379051129365156632153489653285061191534178529543993418
529855775986967586885057
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Fri, 16 Apr 1999 03:25:24 -0500 (CDT)
From: Conrad Curry <[EMAIL PROTECTED]>
Subject: Mersenne: Distributed NFS project
Call for Participants
- ---------------------
After M601 the next smallest Mersenne number is M617, which is on
the 'More Wanted' list of the Cunningham project. I am organizing a
distributed NFS effort to factor M617. Some binaries are available for
x86, alpha, and sparc solaris at ftp://ftp.netdoor.com/users/acurry/nfs/
If you wish to contribute download one of the binaries and email me to
reserve a range. As a test we will be factoring C(2,783+) first.
A word of warning, the program requires up to 20Mb of physical memory
free. I would suggest using a binary factor base, I should have put that
as default before uploading.
When you email me, please also include how many machines you can use
and if you would also be interested in factoring numbers on the Cunningham
'Most Wanted' list.
The programs should be considered beta versions. Let me know of any
questions, comments, or problems you may have with it.
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Thu, 15 Apr 1999 16:08:56 +0200
From: "Steinar H. Gunderson" <[EMAIL PROTECTED]>
Subject: Mersenne: Re: preventive measures
On Sat, 10 Apr 1999 08:41:16 -0600, Aaron Blosser wrote:
>Also, do you plan to optimize the assembly code in any way for the new types
>of CPU's out? AMD's K6-3, Pentium III, etc. It would certainly "seem" that
>some slight tweaking could be done to squeeze out a few extra percent of
>improvement, but I could only guess at that. Maybe just recompiling the
>assembly in a compiler that is PIII/K63 aware would take advantage of the
>processor type, and then just include each compilation in the program with
>the appropriate branches into the code depending on the CPU type actually
>being used, just as you have now for PPro/PII, Pentium, etc.
You have an entirely wrong picture of what assembly is. Assembly is the
most high level programming language existing. It is _not_ a portable language
like C, where the C compiler actually _converts_ (compiles) the C code into
assembly code for the machine that is to run the program, optimizes it
and finally runs it through an assembler and linker to create the full program.
In other words, the phrase `recompiling the assembly in a compiler that is
PIII/K63 aware' is totally meaningless. There will soon be PIII aware
assemblers (for some reason, I belive gas will be one of the first... How
strange.), but all they will do is enable the user to use the new opcodes (as
well as the more `standard' opcodes, some of which have been around since the
8086 (good old days... I had an 8088)). Regardless of which assembler you use,
the code will stay at the exact same speed, as opposed to C compilers, where
it in fact can make a big difference.
Your other ideas are already being discussed on the list, I'll leave them
alone :-)
/* Steinar */
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Sat, 17 Apr 1999 01:02:07 +0200
From: Cyril Flaig <[EMAIL PROTECTED]>
Subject: Mersenne: factor
Why factors prime95 only up to 59 bit at exponnent which are for
doublechecking?
The other (9million) are up to 62 bit.
Why is that so?
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Sat, 17 Apr 1999 01:10:00 -0500
From: "David L. Nicol" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: preventive measures
Aaron Blosser wrote:
> [the prime95 icon is ] mostly RED which would have been a good
> "alert" color. Maybe slowly flashing YELLOW or something, or make is
> usually GREEN then YELLOW for messages and RED for big problems.
Is there an "official international math color?"
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Sat, 17 Apr 1999 01:23:25 -0500
From: "David L. Nicol" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: GUI for Linux
> think there are enough Linux users out there
> to justify it.
here's mine:
xterm -display localhost:0.0 -name mprime -e tail -f results.txt &
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
Date: Sat, 17 Apr 1999 08:07:29 -0700
From: Paul Leyland <[EMAIL PROTECTED]>
Subject: RE: Mersenne: factor
Because the large the number, the more factoring is worth doing before
changing over to the computationally expensive LL test.
Paul
- -----Original Message-----
From: Cyril Flaig [mailto:[EMAIL PROTECTED]]
Sent: 17 April 1999 00:02
To: [EMAIL PROTECTED]
Subject: Mersenne: factor
Why factors prime95 only up to 59 bit at exponnent which are for
doublechecking?
The other (9million) are up to 62 bit.
Why is that so?
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
End of Mersenne Digest V1 #546
******************************
