Delivered-To: luke@ndatech.com X-Sender: rugeley@pop3.demon.co.uk X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Sat, 15 Dec 2001 22:43:59 +0000 To: Luke Welsh From: mersenne-digest-invalid-reply-address@base.com (Mersenne Digest) (by way of Gordon Spence ) Subject: Mersenne Digest V1 #910
Mersenne Digest       Tuesday, November 27 2001       Volume 01 : Number 910

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

Date: Sun, 25 Nov 2001 17:25:12 EST
From: RSNFLD@aol.com
Subject: Mersenne: Re: Mersenne Digest V1 #909

- --part1_12c.82165ee.2932c9c8_boundary
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Hi from Southern California

I've been following the planning of the GIMPS party in the Bay Area. If the
party is held near an airport, I and others from here possibly could attend.
How about it?

Irv Rosenfeld

- --part1_12c.82165ee.2932c9c8_boundary
Content-Type: text/html; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Hi from Southern California

I've been following the planning of the GIMPS party in the Bay Area. If the party is held near an airport, I and others from here possibly could attend. How about it?

Irv Rosenfeld
- --part1_12c.82165ee.2932c9c8_boundary--
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Mon, 26 Nov 2001 00:39:41 EST
From: EWMAYER@aol.com
Subject: Mersenne: Re: Mersenne prime #39

George Woltman wrote:

<<<
    My computer has completed its unofficial verification. It checked out!
GIMPS' fifth prime - and just as exciting as the first one.

    Congratulations and thanks to all GIMPS contributors. Everyone
of us had a hand in this remarkable find!

    Now comes the hard part - waiting for the official verification due
around Dec 6th.
>>>

Nice work, everyone! It's been nearly two-and-a-half years since the last
one, which is a long wait by Internet project standards (excluding SETI.)

A note of explication about the "official verification," which is being run
by Paul Novarese of Compaq on a 667MHz Alpha ev67 using my Mlucas
code (in fact a new version thereof which shall be released soon) - Paul
is getting about the same per-cycle throughput as George gets using
Prime95 on his P4, but since Paul's Alpha only runs about half as fast
as George's P4, it will take about twice as long to complete the verification.
Guillermo Ballester Valor is also doing a pair of DC's using his Glucas code -
one on an 800MHz Itanium, the other on a 1GHz Alpha ev68. Both of these
runs also get about the same per-cycle performance as the aforementioned
codes, so the latter run could be done a week sooner than Paul's, but GBV
is doing his DC's on Compaq's publicly available testdrive machines, which
have been unavailable since middle of last week, so it's not clear by how
much (if any) his DC's will be set back by this. In any event, please don't
go rioting in the streets; after over 2 years, another 9-10 days (hopefully)
won't kill anyone, except perhaps the lucky discoverer. :)

Now I just have to figure out where I'm going to hang the next prime poster,
which is going to be a whopper. Perhaps it can be printed as a diptych
( = 2 equal-sized sheets), which would make framing and transport easier.

Cheers,
- -Ernst
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sun, 25 Nov 2001 22:26:39 -0800
From: "John R Pierce" <pierce@hogranch.com>
Subject: Re: Mersenne: Re: SF Bay GIMPS party (was Mersenne #39)

> "John R Pierce" <pierce@hogranch.com> listed some sites in Monterey.
> So far, however, only he and I have expressed interest in that location.
> Unless at least five mailing list members express interest in Monterey
> by December 1, we should go elsewhere. Of course there is nothing to
> preclude multiple gathering.

I only suggested Monterey because someone else thought many out-of-the-area
list members may be attending that Western Number Theory Conference...

If in fact thats a false assumption, then, heh, yeah, whatever.

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Mon, 26 Nov 2001 19:53:33 +1300
From: "Halliday, Ian" <ian@halliday.com>
Subject: Mersenne: Meanwhile, in another part of the planet

I'm just quietly reading about the possibly enormous and possibly tiny
party to celebrate the discovery of M39. Is there anyone who would care
to join me for a couple of quiet beers in Wellington, New Zealand to
celebrate the same event?

Expressions of interest or flames to me off the list please. Thank you.

Regards,

Ian
- --
Ian W Halliday, BA Hons, MIMIS, AAIBF Snr, ATMB, CL
+64 27 245 6089 (GMT+13)
http://baptism.co.nz
Focus On Success
- --
Word documents not accepted -- see http://baptism.co.nz/word.html
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sun, 25 Nov 2001 23:22:28 -0800
From: "Aaron Blosser" <wablosser@yahoo.com>
Subject: RE: Mersenne: Meanwhile, in another part of the planet

I would think that upon official verification of M39 (as in, 39th
*known* one anyway), everyone, no matter where they are, should raise a
toast to their fellow GIMPSers... There's always that one lucky guy or
gal who actually manages to find one, but it's still a team effort,
because none of this would be at all possible if it weren't for the
thousands of machines out there, plugging away each and every hour of
each and every day.

Meanwhile, I'm estimating whether it would be worthwhile to take a
scenic drive down the coast from Seattle to join the fellow prime
hunters in the Bay area. :)

Aaron

> -----Original Message-----
> From: mersenne-invalid-reply-address@base.com
[mailto:mersenne-invalid-
> reply-address@base.com] On Behalf Of Halliday, Ian
> Sent: Sunday, November 25, 2001 10:54 PM
> To: mersenne@base.com
> Subject: Mersenne: Meanwhile, in another part of the planet
>
> I'm just quietly reading about the possibly enormous and possibly tiny
> party to celebrate the discovery of M39. Is there anyone who would
care
> to join me for a couple of quiet beers in Wellington, New Zealand to
> celebrate the same event?
>
> Expressions of interest or flames to me off the list please. Thank
you.
>
> Regards,
>
> Ian

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Mon, 26 Nov 2001 05:23:44 -0600
From: Michael Kilfoyle <MKilfoyle@home.com>
Subject: Re: Mersenne: Meanwhile, in another part of the planet

A global celebration at the same time... good idea..

"Halliday, Ian" wrote:

> I'm just quietly reading about the possibly enormous and possibly tiny
> party to celebrate the discovery of M39. Is there anyone who would care
> to join me for a couple of quiet beers in Wellington, New Zealand to
> celebrate the same event?
>
> Expressions of interest or flames to me off the list please. Thank you.
>
> Regards,
>
> Ian
> --
> Ian W Halliday, BA Hons, MIMIS, AAIBF Snr, ATMB, CL
> +64 27 245 6089 (GMT+13)
> http://baptism.co.nz
> Focus On Success
> --
> Word documents not accepted -- see http://baptism.co.nz/word.html
> _________________________________________________________________________
> Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
> Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Tue, 27 Nov 2001 00:46:16 +0100
From: "george de fockert" <g.fockert@chello.nl>
Subject: Mersenne: Trial factoring time

Hi all,
One of my machines does trial factoring.
Now in the M17900000 range, factoring to 2^66.
This takes a very long time, is it better to let it double check ?
Or is it time for a more efficient trial factor implementation (if possible
?).

George de Fockert

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Mon, 26 Nov 2001 19:18:55 -0500
From: George Woltman <woltman@alum.mit.edu>
Subject: Re: Mersenne: Trial factoring time

Hi,

At 12:46 AM 11/27/2001 +0100, george de fockert wrote:
>One of my machines does trial factoring.
>Now in the M17900000 range, factoring to 2^66.
>This takes a very long time, is it better to let it double check ?

It is a matter of personal preference. You can figure out how long
a double-check would take at http://www.mersenne.org/bench.htm

Another option for slow machines is ECM factoring. See
http://www.mersenne.org/ecm.htm for details (this is not an
automated project)

>Or is it time for a more efficient trial factor implementation (if possible
>?).

If someone wants to volunteer to improve factoring performance that
would be great. The P4 could use an SSE2 implementation. CPU-specific
factoring code over 64 bits could be written.

While factoring could probably be improved tens of percent, it has not
been a high priority item for me. A 10% boost in factoring speed lets
you factor (on average) 10% deeper, which finds about 1 more factor in
every 650 exponents. That isn't a great throughput improvement for
the GIMPS project.

- -- George

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 24 Nov 2001 12:49:42 -0600 (CST)
From: ribwoods@execpc.com
Subject: Factoring benefit/cost ratio (was: Mersenne: Fw: The Mersenne Newsletter, issue #18)

Earlier, Brian Beesley wrote:
> Eh? Doesn't it make more sense to concentrate on factoring
> Mnumbers that haven't yet been L-L tested? That way "success" in
> finding a factor reduces the number of LL tests, as well as
> (eventually) the number of double checks.

... to which Daran G. recently responded:
> Not necessarily. The marginal benefit/cost ratio of doing factoring
> work on exponant x awaiting a first time test, is twice what it
> would be if exponant x were awaiting a DC. This does not mean that
> it is greater that doing factoring work on exponant y which is
> awaiting a DC.

But ones factoring benefit calculation might ["should" would be in
line with the popular theme of prescribing what's best for other
GIMPS participants :)] include not only the time savings of
eliminating the need for one or two L-L tests, but also the extra
benefit of finding a specific factor.

In the GIMPS Search Status table at www.mersenne.org/status.htm the
march of progress is from "Status Unknown" to "Composite - One LL" to
"Composite - Two LL" to ... "Composite - Factored".

This reflects the view (with which I agree) that it is more valuable
to know a specific factor of a Mnumber than to know that a Mnumber is
composite but not to know any specific factor of that Mnumber.

So a "Factored" status is better for GIMPS than a "Two LL" status, but
calculations of factoring benefit that consider only the savings of
L-L test elimination are neglecting the difference between those two
statuses. If one consciously wants to neglect that difference ...
well, okay ... but I prefer to see that explicitly acknowledged.

Richard Woods

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Tue, 27 Nov 2001 11:14:15 -0500
From: Jeff Woods <jwoods@deltacomm.com>
Subject: Mersenne: World Test Status claims....

The virtual machine's sustained throughput* is currently 2316 billion
floating point operations per second (gigaflops), or 192.4 CPU years
(Pentium 90Mhz) computing time per day. For the testing of Mersenne
numbers, this is equivalent to 82 Cray T916 supercomputers, or 41 of Cray's
most powerful T932 supercomputers, at peak power. As such, PrimeNet ranks
among the most powerful computers in the world. (*Measured in calibrated P5
90Mhz, 32.98 MFLOP units: 25658999 FPO / 0.778s using 256k FFT.)

The link to Cray's "most powerful T932 supercomputers" is broken -- Cray
has removed that link:

http://www.cray.com/products/systems/gallery/t90.html

It appears now that this claim is also somewhat invalid, as Cray's current
champion is an 8,192 GFLOP machine called the SX-6/1024M128, a machine with
128 vector processors, or 64 MFLOPS per processing unit.

As such, it seems that GIMPS is now truly only equivalent to about one
quarter of one of Cray's most powerful SX-6 supercomputers, no?
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Tue, 27 Nov 2001 08:44:42 -0800 (PST)
From: John R Pierce <pierce@hogranch.com>
Subject: Re: Mersenne: World Test Status claims....

> It appears now that this claim is also somewhat invalid, as Cray's current
> champion is an 8,192 GFLOP machine called the SX-6/1024M128, a machine with
> 128 vector processors, or 64 MFLOPS per processing unit.
>
> As such, it seems that GIMPS is now truly only equivalent to about one
> quarter of one of Cray's most powerful SX-6 supercomputers, no?

has Cray actually BUILT one of these yet, or is it just a catalog item
waiting for a order?

- -jrp
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Tue, 27 Nov 2001 20:06:51 -0000
From: bjb@bbhvig.uklinux.net
Subject: Re: Mersenne: Trial factoring time

On 26 Nov 2001, at 19:18, George Woltman wrote:

> Hi,
>
> At 12:46 AM 11/27/2001 +0100, george de fockert wrote:
> >One of my machines does trial factoring.
> >Now in the M17900000 range, factoring to 2^66.
> >This takes a very long time, is it better to let it double check ?
>
> It is a matter of personal preference. You can figure out how long a
> double-check would take at http://www.mersenne.org/bench.htm
>
> Another option for slow machines is ECM factoring. See
> http://www.mersenne.org/ecm.htm for details (this is not an
> automated project)

Or P-1 factoring. Again not automated.
>
> >Or is it time for a more efficient trial factor implementation (if
> >possible ?).
>
> If someone wants to volunteer to improve factoring performance that
> would be great. The P4 could use an SSE2 implementation.
> CPU-specific factoring code over 64 bits could be written.

I had a good look at this a couple of years ago. There are
essentially two parts to the trial factoring "problem": one is
generating candidate factors (or, rather, quickly avoiding those
values which cannot for one reason or another possibly be factors),
the other is actually trying the factor.

Whilst I was able to get reasonably close to the performance of
George's code for trying the factor - possibly even a tiny bit better
for factors with only just over 2^64 - I got nowhere near the
performance of his candidate elimination code.

One of the lessons I learned from the effort is that multiple
precision arithmetic is _hard work_ on systems which have only
one carry flag. If SSE/SSE2 enables you to have multiple carry
flags then you could effectively try two or more factors in parallel,
this would yield a considerable performance increase. But I'm not
sure offhand whether the instruction set supports this. In any case,
slower systems (the ones that _should_ be running factoring?)
don't in any case support SSE or SSE2.
>
> While factoring could probably be improved tens of percent, it has not
> been a high priority item for me. A 10% boost in factoring speed lets
> you factor (on average) 10% deeper, which finds about 1 more factor in
> every 650 exponents. That isn't a great throughput improvement for
> the GIMPS project.

Indeed.

The other way to look at this (same conclusion) is that factoring is
way, way ahead of LL testing, so any perceived inefficiency is not
critical.

Regards
Brian Beesley
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Tue, 27 Nov 2001 17:15:56 -0500
From: Jud McCranie <jud.mccranie@mindspring.com>
Subject: Mersenne: start Prime95 at bootup

I used to have Prime95 starting at bootup on my Windows XP system. I had
some problems, though, and now Prime95 doesn't start at bootup. It is not
listed under msconfig as it used to be. I have "start at bootup"
checked. I deleted prime95 and reinstalled. How can I make it start at
bootup (it used to do that).

+---------------------------------------------------------+
|     Jud McCranie                                        |
|                                                         |
| Programming Achieved with Structure, Clarity, And Logic |
+---------------------------------------------------------+

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Wed, 28 Nov 2001 00:05:30 +0100
From: "george de fockert" <g.fockert@chello.nl>
Subject: Re:Mersenne: Trial factoring time

>Depends on the speed of the machine(s) you are using. Look at the
>benchmark info at www.mersenne.org, or let me know your speed and
>machine type and I will check it for you.

>Best Wishes,
>Stefanovic

>It is a matter of personal preference. You can figure out how long
>a double-check would take at http://www.mersenne.org/bench.htm

>Another option for slow machines is ECM factoring. See
>http://www.mersenne.org/ecm.htm for details (this is not an
>automated project)

>If someone wants to volunteer to improve factoring performance that
>would be great. The P4 could use an SSE2 implementation. CPU-specific
>factoring code over 64 bits could be written.
>
>While factoring could probably be improved tens of percent, it has not
>been a high priority item for me. A 10% boost in factoring speed lets
>you factor (on average) 10% deeper, which finds about 1 more factor in
>every 650 exponents. That isn't a great throughput improvement for
>the GIMPS project.

>-- George

My factoring machine speed is k6 400Mhz , and I know of the benchmark pages,
but there are no factoring benchmarks, only LL.

>George Woltman suggested ECM factoring to you as a possibility.
snip snip
>Good luck! And I would be happy to answer any questions you might have.
>
>Phil Moore
>Lane Community College
>Eugene, Oregon

I will look into the ECM and P-1 factoring, it somehow frustrates me that LL
test tell you
that it is no prime, without knowing a factor.

>Whilst I was able to get reasonably close to the performance of
>George's code for trying the factor - possibly even a tiny bit better
>for factors with only just over 2^64 - I got nowhere near the
>performance of his candidate elimination code.

Hm, do I see here a chalenge for the dark winter nights ?

snip snip

>The other way to look at this (same conclusion) is that factoring is
>way, way ahead of LL testing, so any perceived inefficiency is not
>critical.

>Regards
>Brian Beesley

So, forget trial factoring for the next months, and do only LL ?

George de Fockert

_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

End of Mersenne Digest V1 #910
******************************