Mersenne Digest       Saturday, February 3 2001       Volume 01 : Number 812




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

Date: Fri, 02 Feb 2001 15:24:43 +0100
From: mohk <[EMAIL PROTECTED]>
Subject: Mersenne: idea for a new prime95 version

hi there,

the first idea is more an ideological one. the name is obsolet. :)
i vote for winprime  or prim4win.

the next idea is to give the prime crunchers the choice of doin' what they 
want to do.
for myself, i like to do double test. i could clean up the double tests to 
prove
M(6972593) very fast (even faster than a some old pentiums).
i have an tbird 800, and so i will get LLOne tests always. i could set
another lame proc in the settings, but it isnt really the same.
i'd be pleased to see in the next version a dialog where i can choose the 
test i wanna do.

regards,

mohk

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

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

Date: Fri, 2 Feb 2001 18:03:02 +0100
From: Lars Lindley <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version

Hi there mohk.

I've soon been with the project for 2.5 years and there has always been a 
choice to set if you want primenet to choose a job for you or if you want to 
do LL, doublechecking or factoring. v19? added the option of 10 million digit 
LL:s too.

You can find the settings under Primenet in the dropdown-menus.

Win32Prime would be the correct name to use there since the name reflects 
what platform it runs on.

Regards
/Lars

(It seems we need a new prime or nice milestone now. The list is virtually 
dead :(   )


> the first idea is more an ideological one. the name is obsolet. :)
> i vote for winprime  or prim4win.
>
> the next idea is to give the prime crunchers the choice of doin' what they
> want to do.
> for myself, i like to do double test. i could clean up the double tests to
> prove
> M(6972593) very fast (even faster than a some old pentiums).
> i have an tbird 800, and so i will get LLOne tests always. i could set
> another lame proc in the settings, but it isnt really the same.
> i'd be pleased to see in the next version a dialog where i can choose the
> test i wanna do.
>
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 03 Feb 2001 07:18:45 +0100
From: mohk <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version

Hi, again :)

At 06:03 PM 02.02.01, you wrote:
>Hi there mohk.
>
>I've soon been with the project for 2.5 years and there has always been a
>choice to set if you want primenet to choose a job for you or if you want to
>do LL, doublechecking or factoring. v19? added the option of 10 million digit
>LL:s too.
>You can find the settings under Primenet in the dropdown-menus.

Thanks, I found them, now.


>Win32Prime would be the correct name to use there since the name reflects
>what platform it runs on.
>
>Regards
>/Lars
>
>(It seems we need a new prime or nice milestone now. The list is virtually
>dead :(   )

I thought a voting creates a new discussion, about increcements, odds etc.
C'mon ppl, wake up and give a comment about improving the prime95 proggi.

Mohk

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

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

Date: Sat, 3 Feb 2001 09:23:13 -0000
From: "Brian J. Beesley" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version

On 3 Feb 2001, at 7:18, mohk wrote:

> >Win32Prime would be the correct name to use there since the name
> >reflects what platform it runs on.

I guess Prime95 comes from the "good old days" when Win95 was new and 
unqualified program names were expected to be 16-bit Win 3.x 
applications. BTW there still is a Win 3.x version of this program - 
I wonder if anyone's still using it???

But my vote's _against_ changing the name just for the sake of 
tracking fashion. What the program does is far more important than 
its name. We also have to bear in mind that some of the more obvious 
names have been picked up by other programs.

> >(It seems we need a new prime or nice milestone now. The list is
> >virtually dead :(   )

OK, go find us one ;-)
> 
> I thought a voting creates a new discussion, about increcements, odds
> etc. C'mon ppl, wake up and give a comment about improving the prime95
> proggi.

We need to remember what the function of the program is. It's 
designed to run unobtrusively in the background, and to have a low 
administrative overhead (be easy to set up, and easy to maintain). 
>From this point of view, the existing program does an excellent job, 
it's really hard to see how it could be significantly improved.

George has announced the development of new FFT code optimised for 
Pentium 4. The FFT code is the true heart of the program: it's really 
hard for me to put into words just how much we all owe to George for 
his unstinting efforts to make the program as efficient as it is. 
Suffice it to say that, without George's input, we would probably be 
two or three Mersenne primes short of where we actually are.

Nothing built by human hands is perfect, so, sure, the program could 
be improved! Personally I'd like to see an optimization for Athlon; 
at the expense of having to load different versions for different 
processor types, I'd like to see seperate "streamlined" versions of 
the code optimized for different processor types rather than one 
monolithic program with everything embedded in it; and I'd like to 
see the client/server security code somehow "opened", to facilitate 
the integration of non-Intel clients into PrimeNet, but without 
sacrificing the trust we have that results have really been computed.

With increasing exponent size (and therefore run time), I'd like to 
see PrimeNet evolve to track intermediate residues & also to be able 
to coordinate parallel LL testing & double-checking, so that runs 
which are going wrong can be stopped for investigation without having 
to be run through to the end.

Some people have indicated they'd like a version of the program with 
a pretty screen-saver interface. Fair enough, provided we can keep 
the "classic" version without the extra overhead.

Probably the easiest improvement to make is to have the documentation 
translated into some other languages; unfortunately I can't help 
here, as my command of languages other than English is very poor. 
However, having a copy of README.TXT and the explanatory web pages 
available in one's own native language would almost certainly help to 
popularise the program worldwide. I'm thinking of (in no particular 
order) French, German, Spanish, Japanese - and Chinese, if there is a 
variant which is standard enough to be useful. Please excuse my 
ignorance on this last point. And, of course, any other language for 
which we can find a willing translator :-)

Regards
Brian Beesley

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

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

Date: Fri, 2 Feb 2001 17:08:25 +0100
From: "Steinar H. Gunderson" <[EMAIL PROTECTED]>
Subject: Mersenne: Re: idea for a new prime95 version

On Fri, Feb 02, 2001 at 03:24:43PM +0100, mohk wrote:
>the first idea is more an ideological one. the name is obsolet. :)
>i vote for winprime  or prim4win.

Hardly any good name, as there is (at least?) one version for another
OS, namely mprime for Linux.

>the next idea is to give the prime crunchers the choice of doin' what they 
>want to do.

This already exists -- in Test/PrimeNet, I think. Uncheck the "always
get the work that makes most sense" box, and select freely :-)

/* Steinar */
- -- 
Homepage: http://members.xoom.com/sneeze/
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 03 Feb 2001 14:57:38 -0600
From: Ken Kriesel <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

At 09:23 AM 2/3/2001 -0000, "Brian J. Beesley" <[EMAIL PROTECTED]> wrote:
>George has announced the development of new FFT code optimised for 
>Pentium 4. The FFT code is the true heart of the program: it's really 
>hard for me to put into words just how much we all owe to George for 
>his unstinting efforts to make the program as efficient as it is. 
>Suffice it to say that, without George's input, we would probably be 
>two or three Mersenne primes short of where we actually are.

Possibly 4 primes.

I've been lobbying a bit for a dual-processor optimized version for Intel.  
I have little technical basis on which to judge the potential gains, but 
speculate that memory bus contention and caching efficiency would 
be improved if both processors were working on the same large exponent.

Well before learning of GIMPS and George's program in 1996, I had
coded a program to do limited trial division followed by a Lucas-Lehmer
test.  Having done that, and then seeing the efficiency of George's
program, gave me a better appreciation for how much work and skill
went into prime95.  It's highly optimized, using the counters built into
the cpus for the purpose, and using virtually everything known about
properties of potential factors and the best algorithms to speed things up
both in the factoring attempts and the Lucas Lehmer test.

Our best hopes for future speed improvements are
1) the steady march to faster computers in greater numbers (& more 
multi-cpu systems running multiple instances of the program 1 per cpu)
2) possible future discoveries of better algorithms by mathematicians.
Last I heard, there was still some space between the upper and lower 
bounds to the limit on number of operations required to perform a
long multiplication or squaring.
3) modest increments due to optimizations to specific architectures, 
additional factoring methods, implementation of additional FFT runlengths etc.
The easy large gains were implemented long ago, and medium difficulty
& moderate gains also, leaving diminishing returns requiring significant
effort.

>Nothing built by human hands is perfect, so, sure, the program could 
>be improved! Personally I'd like to see an optimization for Athlon; 
>at the expense of having to load different versions for different 
>processor types, I'd like to see seperate "streamlined" versions of 
>the code optimized for different processor types rather than one 
>monolithic program with everything embedded in it; and I'd like to 
>see the client/server security code somehow "opened", to facilitate 
>the integration of non-Intel clients into PrimeNet, but without 
>sacrificing the trust we have that results have really been computed.

Let's keep it simple; disk space is cheap (including space for paging
out unused code).
I think the current situation where the program detects cpu model
and reacts accordingly is a good one; you can still take control
by editing the ini file if need be.  It keeps it simple for both the novice
end user wanting to download a program and get going, and for the
program developers.  If cpu-specific programs were made
instead, the number of distinct programs gets large because the
combinations of cpu type and OS is large.
Prime95 supports or supported something like 11 cpu type codes.
Regarding OS, there are at least 6 types (currently maintained at V20):
Win95/NT interactive
NT service
linux
statically linked linux
freebsd
statically linked bsd

Setting CPU type in prime95 V20.4.1 and then examining local.ini,
CPUType=3 is a Cyrix 6x86
CPUType=4 is a 486
CPUType=5 is a Pentium
CPUType=6 is a Pentium Pro
CPUType=7 is an AMD K6
CPUType=8 is a Celeron
CPUType=9 is a PentiumII
CPUType=10 is a PentiumIII
CPUType=11 is an AMD Athlon
P-4 code adds another type.  (Is there another AMD type?)
Presumably cputype 386 and below have been retired 
(yes 386's were supported, in v13.2 or so)

Perhaps some GIMPS participants could offer George & Scott 
nonprivileged account access on some other architectures, so they 
could do the required development.

>With increasing exponent size (and therefore run time), I'd like to 
>see PrimeNet evolve to track intermediate residues & also to be able 
>to coordinate parallel LL testing & double-checking, so that runs 
>which are going wrong can be stopped for investigation without having 
>to be run through to the end.

In the QA effort, we've seen a few instances already of errors caught
midway by doing a manual/email version of this.  Brian Beesley had an error
detected this way in his run of a double-check of a 10-megadigit exponent.
This exponent takes a PII-400 428 days (yes 14 months) to complete,
so detecting the one error and restarting early saves about 10.5 PII-400
months.
(Thanks to Rick Pali for providing interim residues to make this savings
possible.)
Another exponent, 20295631 showed similar results; both Paul Victor Novarese's
run and mine produced errors while Brian Beesley's run matched Gordon
Spence's.

I assume that Brian means sending intermediate 64-bit residues to Primenet
for comparison.  (The intermediate save files are too big to send with any
frequency and would require a lot of storage.)

To automate checking via interim residues would require significant longterm 
storage at primenet, of quadruples containing exponent, iteration, 64-bit
residue,
and the source of the information (person or machine ID).  When two with
matching exponent and iteration but different source were available a
comparison
would be made; if a discrepancy was found, both runs should be halted while
a tiebreaker run was made via a different source, to avoid wasting cpu time
of one or both original sources.  Since the most likely cause of a discrepancy
is error in one run not both, a resume capability as well as a discard
capability
would be needed.  I feel exponents halted for a tiebreaker run should not be 
expired.


Ken

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

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

Date: Sat, 03 Feb 2001 16:48:47 -0500
From: Kel Utendorf <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

At 14:57 02/03/2001 -0600, Ken Kriesel wrote:

<snip>

 >In the QA effort, we've seen a few instances already of errors caught
 >midway by doing a manual/email version of this.  Brian Beesley had an error
 >detected this way in his run of a double-check of a 10-megadigit exponent.
 >This exponent takes a PII-400 428 days (yes 14 months) to complete,
 >so detecting the one error and restarting early saves about 10.5 PII-400
 >months.

<snip>

After hanging around the Anandtech DC forum for awhile, I'm convinced that 
this completion time "problem" might be GIMPS biggest hurdle to getting 
more participation.  Very few "loonies" like us are willing to wait 14 
months for the calculation of one result.  Those folks at Anand's are 
interested in visible changes in daily statistics, something GIMPS doesn't 
provide when doing LL testing.

Kel U.


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

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

Date: Sat, 03 Feb 2001 17:03:22 -0500
From: Jeff Woods <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

At 02:57 PM 2/3/01 -0600, you wrote:

> >With increasing exponent size (and therefore run time), I'd like to
> >see PrimeNet evolve to track intermediate residues & also to be able
> >to coordinate parallel LL testing & double-checking, so that runs
> >which are going wrong can be stopped for investigation without having
> >to be run through to the end.
>
>In the QA effort, we've seen a few instances already of errors caught
>midway by doing a manual/email version of this.  Brian Beesley had an error
>detected this way in his run of a double-check of a 10-megadigit exponent.
>This exponent takes a PII-400 428 days (yes 14 months) to complete,
>so detecting the one error and restarting early saves about 10.5 PII-400
>months.

I think this is an EXCELLENT idea, but remember that the "s" values (i.e. 
the intermediate residue/modulus) for such numbers is quite simply 
enormous.   One couldn't (and shouldn't) check the entire intermediate 
value, but merely the last "x" bits, where "x" is enough to be reasonably 
certain that a match isn't random chance -- say, the final 1024 bits.

PrimeNet would thus also have to carefully assign the exponents to similar 
machines with similar runtimes and performance, as it would do little good 
to assign the primary test to an Athlon-800 and the "real-time" 
double-check to a much slower machine, as the Athlon would quickly outpace 
the second check.

If a discrepancy was found in a real-time double-check, a ternary run on a 
different machine could determine which (if either) of the two intermediate 
residuals was correct, and the tests could proceed from there, with both 
original machines assuming the same correct residue.

Also, if this did evolve, I'd suggest that the "double-checker" be given 
equal credit with the primary machine, for purposes of credit in history 
books as discoverers, and/or EFF monies.

>I assume that Brian means sending intermediate 64-bit residues to Primenet
>for comparison.  (The intermediate save files are too big to send with any
>frequency and would require a lot of storage.)
>
>To automate checking via interim residues would require significant longterm
>storage at primenet, of quadruples containing exponent, iteration, 64-bit
>residue, and the source of the information (person or machine ID).  When 
>two with matching exponent and iteration but different source were available a
>comparison would be made; if a discrepancy was found, both runs should be 
>halted while a tiebreaker run was made via a different source, to avoid 
>wasting cpu time of one or both original sources.  Since the most likely 
>cause of a discrepancy is error in one run not both, a resume capability 
>as well as a discard capability would be needed.  I feel exponents halted 
>for a tiebreaker run should not be expired.

I'd agree.  Machines awaiting a tie-breaker could move on to factoring, or 
another smaller double-check.  I would not want to see such machines begin 
another 14-month effort, as once the tiebreaker concluded, that work would 
be suspended while the first test was concluded.

Note that there's a point of futility, at which a "tie-breaker" ought to 
merely be a triple-check, run to conclusion.  Let's say on a 14-month co-op 
effort, 13.6 months into it a discrepancy was found.   Both machines ought 
to finish, and just have it triple-checked, rather than suspending both, 
awaiting a tiebreaker.   While I'm sure someone could solve for the optimum 
cutoff point where tiebreakers are not useful, my guess would be that it is 
around 85% of the way to completion.
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 03 Feb 2001 17:04:44 -0500
From: Jeff Woods <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

At 04:48 PM 2/3/01 -0500, you wrote:

>After hanging around the Anandtech DC forum for awhile, I'm convinced that 
>this completion time "problem" might be GIMPS biggest hurdle to getting 
>more participation.  Very few "loonies" like us are willing to wait 14 
>months for the calculation of one result.  Those folks at Anand's are 
>interested in visible changes in daily statistics, something GIMPS doesn't 
>provide when doing LL testing.

Then why is SETI@home so popular, when it shows little in the way of daily 
statistics, either?
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 03 Feb 2001 17:37:08 -0500
From: Kel Utendorf <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

At 17:04 02/03/2001 -0500, Jeff Woods wrote:
 >At 04:48 PM 2/3/01 -0500, you wrote:
 >
 >>After hanging around the Anandtech DC forum for awhile, I'm convinced that
 >>this completion time "problem" might be GIMPS biggest hurdle to getting
 >>more participation.  Very few "loonies" like us are willing to wait 14
 >>months for the calculation of one result.  Those folks at Anand's are
 >>interested in visible changes in daily statistics, something GIMPS doesn't
 >>provide when doing LL testing.
 >
 >Then why is SETI@home so popular, when it shows little in the way of daily
 >statistics, either?

My understanding is that it is possible to have a significant change in 
daily statistics with  SETI@home.  A fast machine can complete a SETI work 
unit in 8 to 10 hours, I believe.

Additionally, I believe that SETI@home is "sexier" for the general public 
and has done a much more thorough job of selling itself via the general media.

BTW, I do participate in GIMPS (and have since around 1996 or so), so it's 
not as if I think it's a bad thing.  I just think it will be tough to 
attract significant membership numbers (say two hundred thousand users, 
just to throw out a number) when an exponent takes 14 months to complete.

Kel


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

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

Date: Sun, 04 Feb 2001 00:27:53 +0100
From: Alexander Kruppa <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version

"Brian J. Beesley" wrote:
> 
> Some people have indicated they'd like a version of the program with
> a pretty screen-saver interface. Fair enough, provided we can keep
> the "classic" version without the extra overhead.
> 

The screen-saver idea is important for another reason.
I asked several coworkers and secretaries to let Prime95 (NTprime,
actually) run on their PCs and they agreed, but they were less than
happy when I asked them to change the pretty 3-d screen savers for
something that lets NTprime have more cpu power. With the selection
Microsoft offers right now, that means "Blank Screen" or "Marquee" -
neither is extremely exciting to watch. Before long, most of them went
back to the old screen savers and NTprime slowed down to a halt.
A pretty screen saver that uses very little cpu time would free up a lot
of resources for background computing. It doesnt have to come in the
Prime95 package (tough a screen saver that displays progress might be
nice), just a separate program would do so we can offer something for
those pc users that dont really care what the computer is doing as long
as it looks nice (theres a lot of them).

> Probably the easiest improvement to make is to have the documentation
> translated into some other languages;

I have no idea how good I am at translating technical documentation - I
never tried before, but if theres demand, I can give it a try for a
german version of "Primzahl95" *g*

> Regards
> Brian Beesley

Ciao,
  Alex.
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

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

Date: Sat, 03 Feb 2001 19:07:36 -0500
From: Nathan Russell <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

Ken Kriesel wrote:

> At 09:23 AM 2/3/2001 -0000, "Brian J. Beesley" <[EMAIL PROTECTED]> wrote:
> 
>> George has announced the development of new FFT code optimised for 
>> Pentium 4. The FFT code is the true heart of the program: it's really 
>> hard for me to put into words just how much we all owe to George for 
>> his unstinting efforts to make the program as efficient as it is. 
>> Suffice it to say that, without George's input, we would probably be 
>> two or three Mersenne primes short of where we actually are.
> 
> 
> Possibly 4 primes.

That would obviously depend on how fast David Slowinski was progressing 
at the time.  Remember that much of George's contribution was organizing 
the project itself, though he certainly has put a huge amount of effort 
into developing the x86 software. 

> 
> I've been lobbying a bit for a dual-processor optimized version for Intel.  
> I have little technical basis on which to judge the potential gains, but 
> speculate that memory bus contention and caching efficiency would 
> be improved if both processors were working on the same large exponent.

IIRC, Slowinski usually ran one exponent on each processor, except when 
verifying a prime. 

> 
> Well before learning of GIMPS and George's program in 1996, I had
> coded a program to do limited trial division followed by a Lucas-Lehmer
> test.  Having done that, and then seeing the efficiency of George's
> program, gave me a better appreciation for how much work and skill
> went into prime95.  It's highly optimized, using the counters built into
> the cpus for the purpose, and using virtually everything known about
> properties of potential factors and the best algorithms to speed things up
> both in the factoring attempts and the Lucas Lehmer test.
> 
> Our best hopes for future speed improvements are
> 1) the steady march to faster computers in greater numbers (& more 
> multi-cpu systems running multiple instances of the program 1 per cpu)
> 2) possible future discoveries of better algorithms by mathematicians.
> Last I heard, there was still some space between the upper and lower 
> bounds to the limit on number of operations required to perform a
> long multiplication or squaring.
> 3) modest increments due to optimizations to specific architectures, 
> additional factoring methods, implementation of additional FFT runlengths etc.
> The easy large gains were implemented long ago, and medium difficulty
> & moderate gains also, leaving diminishing returns requiring significant
> effort.

Obviously, something else we can all do is encourage our friends to run 
the program.  I am a member of a website known as everything2, and have 
written articles for that site regarding GIMPS.  Feedback is welcome.  I 
am aware that my article is now extremely brief, but it's hard to see 
what more can be said in a non-technical way. 

http://www.everything2.com/index.pl?node_id=877902
I have also mentioned GIMPS on my homepage,
http://www.acsu.buffalo.edu/~nrussell/

> 
>> Nothing built by human hands is perfect, so, sure, the program could 
>> be improved! Personally I'd like to see an optimization for Athlon; 
>> at the expense of having to load different versions for different 
>> processor types, I'd like to see seperate "streamlined" versions of 
>> the code optimized for different processor types rather than one 
>> monolithic program with everything embedded in it; and I'd like to 
>> see the client/server security code somehow "opened", to facilitate 
>> the integration of non-Intel clients into PrimeNet, but without 
>> sacrificing the trust we have that results have really been computed.
> 
> 
> Let's keep it simple; disk space is cheap (including space for paging
> out unused code).
> I think the current situation where the program detects cpu model
> and reacts accordingly is a good one; you can still take control
> by editing the ini file if need be.  It keeps it simple for both the novice
> end user wanting to download a program and get going, and for the
> program developers.  If cpu-specific programs were made
> instead, the number of distinct programs gets large because the
> combinations of cpu type and OS is large.

Agreed there!  There is nothing wrong with monolithic programs.  
Compared with, eg, Netscape or even Winamp, Prime95 requires a very 
small amount of resources. 
(snip)

> 
> Perhaps some GIMPS participants could offer George & Scott 
> nonprivileged account access on some other architectures, so they 
> could do the required development.

Certainly not a bad idea.  I think an Amiga client in particular would 
attract a fair amount of interest, as would a version specially designed 
for NetBSD or OpenBSD

> 
>> With increasing exponent size (and therefore run time), I'd like to 
>> see PrimeNet evolve to track intermediate residues & also to be able 
>> to coordinate parallel LL testing & double-checking, so that runs 
>> which are going wrong can be stopped for investigation without having 
>> to be run through to the end.
> 
> 
> In the QA effort, we've seen a few instances already of errors caught
> midway by doing a manual/email version of this.  Brian Beesley had an error
> detected this way in his run of a double-check of a 10-megadigit exponent.
> This exponent takes a PII-400 428 days (yes 14 months) to complete,
> so detecting the one error and restarting early saves about 10.5 PII-400
> months.
> (Thanks to Rick Pali for providing interim residues to make this savings
> possible.)
> Another exponent, 20295631 showed similar results; both Paul Victor Novarese's
> run and mine produced errors while Brian Beesley's run matched Gordon
> Spence's.
> 
> I assume that Brian means sending intermediate 64-bit residues to Primenet
> for comparison.  (The intermediate save files are too big to send with any
> frequency and would require a lot of storage.)

Agreed there.  On the other hand, on a fast machine, the SETI project 
requires receiving the better part of a megabyte every day, and they 
have had no problem finding new recruits. 

> 
> To automate checking via interim residues would require significant longterm 
> storage at primenet, of quadruples containing exponent, iteration, 64-bit
> residue,
> and the source of the information (person or machine ID).  When two with
> matching exponent and iteration but different source were available a
> comparison
> would be made; if a discrepancy was found, both runs should be halted while
> a tiebreaker run was made via a different source, to avoid wasting cpu time
> of one or both original sources.  

The difficulty here is explaining the entire process to new users.  I 
know that when i was new I had enough trouble deciding on issues like 
how much memory to make available, even without worrying about comparing 
residues at that time. 

> Since the most likely cause of a discrepancy
> is error in one run not both, a resume capability as well as a discard
> capability
> would be needed.  I feel exponents halted for a tiebreaker run should not be 
> expired.

Note that all of this will require making changes in the server, and 
there has historically been a fair backlog of such changes. 

> 
> 
> Ken

Regards,
Nathan

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

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

Date: Sat, 03 Feb 2001 19:10:14 -0500
From: Nathan Russell <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

Jeff Woods wrote:

> At 04:48 PM 2/3/01 -0500, you wrote:
> 
>> After hanging around the Anandtech DC forum for awhile, I'm convinced 
>> that this completion time "problem" might be GIMPS biggest hurdle to 
>> getting more participation.  Very few "loonies" like us are willing 
>> to wait 14 months for the calculation of one result.  Those folks at 
>> Anand's are interested in visible changes in daily statistics, 
>> something GIMPS doesn't provide when doing LL testing.
> 
> 
> Then why is SETI@home so popular, when it shows little in the way of 
> daily statistics, either? 


I recently invested a few CPU days in SETI.  On my P3-600, I was 
completing a work unit about every 7-8 hours. 

Most people's primary machines could certainly complete one work unit 
per day quite easily. 

I might point out, though, that now that SETI is producing a new client 
version which does more testing, and requires roughly twice as long to 
do each work unit, some folks were complaining on their IRC channel 
about the time needed to finish each work unit. 

Nathan

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

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

Date: Sat, 03 Feb 2001 19:10:26 -0500
From: Nathan Russell <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version

Alexander Kruppa wrote:

> "Brian J. Beesley" wrote:
> 
>> Some people have indicated they'd like a version of the program with
>> a pretty screen-saver interface. Fair enough, provided we can keep
>> the "classic" version without the extra overhead.
>> 
> 
> The screen-saver idea is important for another reason.
> I asked several coworkers and secretaries to let Prime95 (NTprime,
> actually) run on their PCs and they agreed, but they were less than
> happy when I asked them to change the pretty 3-d screen savers for
> something that lets NTprime have more cpu power. With the selection
> Microsoft offers right now, that means "Blank Screen" or "Marquee" -
> neither is extremely exciting to watch. Before long, most of them went
> back to the old screen savers and NTprime slowed down to a halt.
> A pretty screen saver that uses very little cpu time would free up a lot
> of resources for background computing. It doesnt have to come in the
> Prime95 package (tough a screen saver that displays progress might be
> nice), just a separate program would do so we can offer something for
> those pc users that dont really care what the computer is doing as long
> as it looks nice (theres a lot of them).

I would say that the best option is either a jumping image or something 
like the game of life, updated very slowly (I'm thinking of the default 
xlock display here).  Generally, things that jump or stay fixed rather 
than sliding or, worse, warping are most likely the way to go. 

I believe Prime95 does come with a screen saver, though it's just a 
blank screen.  Personally, I've used the 'blank screen' option myself 
since getting involved in distributed computing, but I realize that some 
folks really want a pretty screen saver.

> 
> 
>> Probably the easiest improvement to make is to have the documentation
>> translated into some other languages;
> 
> 
> I have no idea how good I am at translating technical documentation - I
> never tried before, but if theres demand, I can give it a try for a
> german version of "Primzahl95" *g*

I guess I could do something in spanish, but there's a lot of demands on 
my time right now, and frankly I'd want it proofread by someone who was 
either a native speaker or had had more than just 5 high school courses!

Nathan

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

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

Date: Sat, 03 Feb 2001 20:05:45 -0500
From: "Joshua Zelinsky" <[EMAIL PROTECTED]>
Subject: Mersenne: Misc Stuff.

A few minor comments.

1. A while back I remember someone making a 2/7 or so serious comment about 
using advanced cell phones and such for factoring. That idea was shot down 
because of lack of computing power/download time + expenses.
What about some of the new gaming platforms. I think some have computing 
capabilities equivalent to P133s and they have modem hookups. However, I'm 
not sure how feasible/worthwhile it would be to write progamrs to do this.

2. Has anyone considered backing up primenet? Conceivably we could farm out 
back up to different computers which are using Prime95. Hardrives just keep 
getting bigger, and most people have tremendous amounts of free space.

3. Recently people have been discussing that the long times it takes to 
finish a batch of work makes GIMPS less appealing than other distributed 
computing systems. What if we gave out factoring in small bunches for 
multiple possible Ms. Since we know that any prime divisor of Mn is of the 
form 2kn+1, the person could check a set of values for k for a large set of 
n. This would allow us to have much shorter runs.
Unfortunately, this would require some major changes on how everything is 
done.

4. Is there ever going to be a Mac version for searching? This would be 
really helpful for a lot of people.

5. One reason SETI is more popular is that they seem to have good media 
relations. Even if we could get a small mention of GIMPS in some non-math 
publication, the effect could be enormous. The Science Section of the NYT 
would be really amazing. Any thoughts?


_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com

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

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

Date: Sun, 4 Feb 2001 04:53:15 +0000 (GMT)
From: Russel Brooks <[EMAIL PROTECTED]>
Subject: Re: Mersenne: idea for a new prime95 version, QA, primenet etc

Jeff Woods wrote:
> Then why is SETI@home so popular, when it shows little in the way of daily
> statistics, either?

Because Space/Aliens/E.T./Sci-Fi/etc is popular and SETI lets
you participate, not just watch NASA/movies/others...

That busy colorful SETI screensaver is also pretty neat to watch.

Cheers... Russ

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

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

Date: Sun, 04 Feb 2001 19:10:31 +1300
From: "Halliday, Ian" <[EMAIL PROTECTED]>
Subject: Mersenne: Distributed Computing Mandatory For Juno's Free Users

This is a cryptographically signed message in MIME format.

- --------------ms2EB503DCE9AB021E8D6E98B9
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

http://au.dailynews.yahoo.com/headlines/20010203/nbtech/981156900-2685255736.html
describes new conditions for free juno users - once again SETI is cited
as a "successful" example of distributed computing. IIRC, we have had
four successes, they haven't had any...
How will the new conditions described in their terms affect us (or any
other voluntary distributed project for that matter) ?

On a different matter, what happened to Lennart's offer of champagne to
the person who guessed a milestone date correctly? Have we reached that
milestone yet? If so, who won?

Regards,

Ian
- --
Ian W Halliday, BA Hons, MIMIS, AAIBF Snr, ATMB, CL
+64 25 245 6089
http://baptism.co.nz
Excel in all we do
- --------------ms2EB503DCE9AB021E8D6E98B9
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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- --------------ms2EB503DCE9AB021E8D6E98B9--

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

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

End of Mersenne Digest V1 #812
******************************

Reply via email to