Mersenne Digest      Thursday, September 14 2000      Volume 01 : Number 777




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

Date: Sun, 10 Sep 2000 13:53:30 EDT
From: "Chris Nash" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: A new series of Mersenne-like Gaussian primes

Hi Mike and Yann

>I don't have any Linux software. For searching for new primes of this 
series, 
>the best program to use at present is PrimeForm/GW by Chris Nash. You 
could 
>ask him ([EMAIL PROTECTED]) whether he has anything for Linux.

Work is currently being done on a direct recompilation of PrimeForm/GW 
for Linux. Since PFGW is a command-line program, this should not be too 
difficult a task - in fact, PFGW is actually developed on BeOS using 
the GNU development tools and the Windows version is already a port.

Peter Kosinar is currently looking at the rebuild, and in fact managed 
to get the BeOS objects to link first time on Linux. There were a few 
system-specific problems (for instance, none of the timing functions 
worked) and Peter has made a few changes, however, the resulting 
program failed. It seems this might be because of some differences 
between the versions of gcc used, the BeOS version is 2.9, and I've yet 
to get 2.95 to successfully build under BeOS. Peter now has all the 
source and is doing a full rebuild, I don't foresee any major problems, 
so a Linux version may be available very soon indeed.

Best Wishes, and thanks for the recommendation!

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

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

Date: Sun, 10 Sep 2000 12:25:06 -0700
From: Colin Percival <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Volunteers wanted

At 7 Sep 2000 14:25:49 -0000, "Brian J. Beesley" <[EMAIL PROTECTED]> wrote:
>The problem could be a program bug (unlikely), an undiscovered bug in 
>either Athlon or Pentium III floating-point arithmetic (again 
>unlikely) or a random hardware glitch in either my system or the 
>system against which I'm checking (much the most likely).

  Or we might just be unlucky.  There are inputs for which the
multiplication used in prime95 will not give the correct answer --
fortunately they are very rare and it is very unlikely that any would ever
be encountered.  (It isn't worth using safe multiplications because any
problems will show up in the double check anyway.)
  I agree that a random hardware glitch is by far the most likely, but the
possibility of unlucky values should be kept in mind.

Colin Percival

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

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

Date: 11 Sep 2000 00:31:05 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Volunteers wanted

On Sun, Sep 10, 2000 3:25 PM, Colin Percival <[EMAIL PROTECTED]> wrote:

>  Or we might just be unlucky.  There are inputs for which the
>multiplication used in prime95 will not give the correct answer --
>fortunately they are very rare and it is very unlikely that any would ever
>be encountered.  (It isn't worth using safe multiplications because any
>problems will show up in the double check anyway.)

This reminds me of something I've been meaning to ask about double
checking.

I looked at a bit of the database of double-checked exponents:
ftp://mersenne.org/gimps/lucas_v.zip

I noticed some exponents where the first-time and double-check results were
submitted by the same person, using the same software, on the same machine.
If there is a reproducible hardware problem, or one of these "bad"
multiplications, then this double-check would not catch an error. 
Actually,
I kind of doubt that the LL test was done twice.  More likely one result
was accidentally _submitted_ twice, and the friendly primenet server
happily 
accepted both.  (We know that the primenet server accepts results from 
people who do NOT have the exponents assigned to them.)

Will these seemingly bogus double-checks be weeded out when the next
synchronization is done?

- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: Mon, 11 Sep 2000 06:42:45 -0000
From: "Brian J. Beesley" <[EMAIL PROTECTED]>
Subject: Mersenne: Moriarty, where are you?

Sorry to abuse the list, but Jean-Yves Canart obviously reads this 
list since he graciously replied to my recent request of volunteers.

Unfortunately his address as given in the reply I received does not 
work. I'd be grateful if he could reply again giving a working e-mail 
address.

Regards
Brian Beesley

............................................
Date sent:              Sun, 10 Sep 2000 00:41:53 +0200 (MET DST)
From:                   Mail Delivery Subsystem <MAILER-
[EMAIL PROTECTED]>
To:                     <[EMAIL PROTECTED]>
Subject:                Returned mail: User unknown

The original message was received at Sun, 10 Sep 2000 00:41:53 +0200 
(MET DST)
from smtpuser@localhost

   ----- The following addresses had permanent fatal errors -----
<[EMAIL PROTECTED]>

   ----- Transcript of session follows -----
550 <[EMAIL PROTECTED]>... User unknown

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

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

Date: Mon, 11 Sep 2000 07:54:03 -0000
From: "Brian J. Beesley" <[EMAIL PROTECTED]>
Subject: Double check validity (was: Re: Mersenne: Volunteers wanted)

On 11 Sep 00, at 0:31, Robert Deininger wrote:

> I looked at a bit of the database of double-checked exponents:
> ftp://mersenne.org/gimps/lucas_v.zip
> 
> I noticed some exponents where the first-time and double-check results
> were submitted by the same person, using the same software, on the same
> machine. If there is a reproducible hardware problem, or one of these
> "bad" multiplications, then this double-check would not catch an error.

If the offset was different then this would be a valid double check. 
However I agree it is undesirable for the first test and the double 
check for an exponent to be submitted by the same user - though, 
where purely automatic methods of obtaining assignments and reporting 
results were employed, surely it makes no difference.

Actually this is the reason I'm so keen to definitively eliminate the 
very small chance that a systematic hardware problem affecting either 
Intel Pentium or Athlon CPU families could be responsible for the 
mismatch found during the large exponent QA work. If this were the 
case we might have to double-check any exponents which had already 
been tested twice using the processor type found to have a problem.

There are also a number of (quite small) exponents where either the 
first test or the double check were done using "old" software which 
reported much less than 64 bits of residual. I'm steadily working my 
way through triple-checking these using systems which are too slow to 
be otherwise useful any more. I guess it would probably be worthwhile 
running independent triple-checks on those exponents where first test 
and double check have been done by the same user - preferably using a 
different program as well - but I don't think this task falls into 
the "very urgent" category.

> Actually, I kind of doubt that the LL test was done twice.  More likely
> one result was accidentally _submitted_ twice, and the friendly primenet
> server happily accepted both.

Hmm. In this case, if the program was George's, the offset (which is 
not shown in the lucas_v file) would be the same; hopefully George's 
procedure would catch this.

> (We know that the primenet server accepts
> results from people who do NOT have the exponents assigned to them.)

Is this relevant? More than two different people independently 
testing any given exponent is a probably unneccessary duplication of 
effort, but surely acts to _increase_ confidence in the integrity of 
the database rather than reduce it.
> 
> Will these seemingly bogus double-checks be weeded out when the next
> synchronization is done?

Hopefully, if there are any bogus double-checks, they will be removed 
and any neccessary double-checks added to the list of assignments to 
be allocated.

BTW it's a _long_ time since we had a database sychronization ...


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

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

Date: 11 Sep 2000 10:48:40 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Mersenne: Re: Double check validity

On Mon, Sep 11, 2000 3:54 AM, Brian J. Beesley <[EMAIL PROTECTED]>
 wrote:
>On 11 Sep 00, at 0:31, Robert Deininger wrote:

>> I noticed some exponents where the first-time and double-check results
>> were submitted by the same person, using the same software, on the same
>> machine. If there is a reproducible hardware problem, or one of these
>> "bad" multiplications, then this double-check would not catch an error.
>
>If the offset was different then this would be a valid double check. 

Assuming Prime95 was the program that did the test.

>However I agree it is undesirable for the first test and the double 
>check for an exponent to be submitted by the same user - though, 
>where purely automatic methods of obtaining assignments and reporting 
>results were employed, surely it makes no difference.

Don't forget the manually-submitted test results.  One way to get the
odd-looking results I saw is to simply submit the same result twice in
succession.  If the network or the server is acting up, clicking the
"submit" button more than once is a natural response.  In this case the
database might show two LL test results, when only one was actaully done.
It isn't apparent that there is any safeguard against this.

>> Actually, I kind of doubt that the LL test was done twice.  More likely
>> one result was accidentally _submitted_ twice, and the friendly primenet
>> server happily accepted both.
>
>Hmm. In this case, if the program was George's, the offset (which is 
>not shown in the lucas_v file) would be the same; hopefully George's 
>procedure would catch this.

What about the other programs?

The database does record the date and time when a result is submitted.  If
the single and double check results came in very close together, it's a
good bet they are not independent.  The time isn't in lucas_v, so I can't
tell with the cases I noticed.

>> (We know that the primenet server accepts
>> results from people who do NOT have the exponents assigned to them.)
>
>Is this relevant? More than two different people independently 
>testing any given exponent is a probably unneccessary duplication of 
>effort, but surely acts to _increase_ confidence in the integrity of 
>the database rather than reduce it.

Yes, I think it's relevant.  The server seems pretty careful about
assigning
exponents to only one person at a time.  It isn't nearly as careful about
accepting results.  A bogus double-check result, accidental or otherwise,
would prevent an exponent from ever being assigned to anyone for a real
double-check.  That does not increase my confidence!


- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: Mon, 11 Sep 2000 09:24:41 -0700
From: "Scott Kurowski" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: P-1 Credit

Hi Eric,

Eric Hahn <[EMAIL PROTECTED]> wrote:
> Terry S. Arnold wrote:
>>Does anyone have any skinny on when we will start getting credit for
>>1.      doing P-1 testing?
>>2.      finding a factor during P-1 testing?
>
>AFAIK, from what George has said, credit will eventually be given
>after BOTH v21 comes out, and the Scott has time to do some
>modifications to the PrimeNet server...  Time Frame?  ...??????...

Changes to support this are being designed.  Mostly we needed to first get
the next-generation Entropia network deployed to support v21 and other
applications.  This has been happening for a while now, so we will be
resuming a focus on GIMPS soon.

Regards,
scott kurowski

Entropia, Inc.

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

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

Date: Mon, 11 Sep 2000 10:23:04 -0700
From: "Scott Kurowski" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: P-1 Credit

Hi Eric,

Eric Hahn <[EMAIL PROTECTED]> wrote:
> Terry S. Arnold wrote:
>>Does anyone have any skinny on when we will start getting credit for
>>1.      doing P-1 testing?
>>2.      finding a factor during P-1 testing?
>
>AFAIK, from what George has said, credit will eventually be given
>after BOTH v21 comes out, and the Scott has time to do some
>modifications to the PrimeNet server...  Time Frame?  ...??????...

Changes to support this are being designed.  Mostly we needed to first get
the next-generation Entropia network deployed to support v21 and other
applications.  This has been happening for a while now, so we will be
resuming a focus on GIMPS soon.

Regards,
scott kurowski

Entropia, Inc.

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

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

Date: Mon, 11 Sep 2000 15:10:35 -0400
From: George Woltman <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

Hi,

At 10:48 AM 9/11/00 -0400, Robert Deininger wrote:

> >> I noticed some exponents where the first-time and double-check results
> >> were submitted by the same person, using the same software, on the same
> >> machine. If there is a reproducible hardware problem, or one of these
> >> "bad" multiplications, then this double-check would not catch an error.
> >
> >If the offset was different then this would be a valid double check.

The server is not responsible for determining if a double-check is valid.
That step is done by me (well, really a program I wrote) using log files
from the server and emailed results.

An exponent is considered double-checked if either tested by a different
program or tested by prime95 using a different shift count (called offset
by some).

If the first and second tests were done by the same user, then my program
prompts me as to whether I really want to accept this result.  I usually
accept the double-check for the following reasons:
         1)  It is not uncommon for top producers to get assigned a
             double-check where they did the first test.
         2)  It is not uncommon for folks to "screw up" the setup in
             multi-machine environments.  A process called "ghosting"(?)
             results in the prime95 directory being copied from a central
             server.  Now each machine has the same worktodo.ini file.
             Some users make the same mistake manually setting up
             multiple machines initially.
         3)  How likely is it that someone goes to the trouble of figuring out
             how to forge the 32-bit verification code on each results line
             and then use that info to turn in bogus *composite* results?
             After all, such a clever person might well be smart enough to
             get a variety of userids too.

Regards,
George

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

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

Date: Mon, 11 Sep 2000 15:15:32 -0500
From: Ryan McGarry <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

George Woltman wrote:
> <snipped>
> 
>          2)  It is not uncommon for folks to "screw up" the setup in
>              multi-machine environments.  A process called "ghosting"(?)
>              results in the prime95 directory being copied from a central
>              server.  Now each machine has the same worktodo.ini file.
>              Some users make the same mistake manually setting up
>              multiple machines initially.
> <snipped>

FYI, Norton Ghost is a program which allows sysadmin's to take a
snapshot of a system's hard drive to set up systems quickly.  You just
pop in the network boot floppy, select the image, and wait for an hour
while your hard disk fills up with a fresh copy the OS of your choice. 
We use it here, and it saves days of work.  Now if only Symantec would
get their Win2k support together...

Some ideas to help those of use who use Ghost:
  1) If there was a way for Prime95/PrimeNT to either store worktodo.ini
and/or results files on a separate server or 
  2) If Prime95/NT could somehow detect the machine's different name/MAC
and automatically configure itself based on the machine's name to start
from scratch...

This would make setting up PrimeNT on every machine (i've got access to
>80 Pentium II's and Celerons) MUCH more appealing...
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

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

Date: Mon, 11 Sep 2000 15:15:44 -0500
From: Ryan McGarry <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

George Woltman wrote:
> <snipped>
> 
>          2)  It is not uncommon for folks to "screw up" the setup in
>              multi-machine environments.  A process called "ghosting"(?)
>              results in the prime95 directory being copied from a central
>              server.  Now each machine has the same worktodo.ini file.
>              Some users make the same mistake manually setting up
>              multiple machines initially.
> <snipped>

FYI, Norton Ghost is a program which allows sysadmin's to take a
snapshot of a system's hard drive to set up systems quickly.  You just
pop in the network boot floppy, select the image, and wait for an hour
while your hard disk fills up with a fresh copy the OS of your choice. 
We use it here, and it saves days of work.  Now if only Symantec would
get their Win2k support together...

Some ideas to help those of use who use Ghost:
  1) If there was a way for Prime95/PrimeNT to either store worktodo.ini
and/or results files on a separate server or 
  2) If Prime95/NT could somehow detect the machine's different name/MAC
and automatically configure itself based on the machine's name to start
from scratch...

This would make setting up PrimeNT on every machine (i've got access to
>80 Pentium II's and Celerons) MUCH more appealing...  

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

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

Date: Mon, 11 Sep 2000 23:19:04 +0200
From: "Steinar H. Gunderson" <[EMAIL PROTECTED]>
Subject: Mersenne: [OT] Re: Double check validity

On Mon, Sep 11, 2000 at 03:15:44PM -0500, Ryan McGarry wrote:
>You just pop in the network boot floppy

Try network boot once -- it's great! ;-)
 
>  1) If there was a way for Prime95/PrimeNT to either store worktodo.ini
>and/or results files on a separate server or 
>  2) If Prime95/NT could somehow detect the machine's different name/MAC
>and automatically configure itself based on the machine's name to start
>from scratch...

I've been asking about a similiar thing -- I guess it's on George's todo list
:-) If I could somehow type:

prime.ini=\\primeserver\%i

...and it would look for prime.ini in \\primeserver\194.248.45.42 (or whatever),
that would increase my school's throughput by... oh, say 60-70%. ;-)

>This would make setting up PrimeNT on every machine (i've got access to
>80 Pentium II's and Celerons) MUCH more appealing...  

30 here, but yes, I know the problem.

/* 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: 11 Sep 2000 23:43:30 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On Mon, Sep 11, 2000 3:10 PM, George Woltman <[EMAIL PROTECTED]> wrote:
>Hi,
>
>An exponent is considered double-checked if either tested by a different
>program or tested by prime95 using a different shift count (called offset
>by some).
>
>         3)  How likely is it that someone goes to the trouble of figuring
out
>             how to forge the 32-bit verification code on each results
line
>             and then use that info to turn in bogus *composite* results?
>             After all, such a clever person might well be smart enough to
>             get a variety of userids too.

I'm not particularly worried about intentional cheating, just accidents.
OTOH, there isn't anything like a verification code on the manual check-in
page.

Maybe non-prime95 results are so rare they aren't worth thinking about.


- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: 11 Sep 2000 23:44:02 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On Mon, Sep 11, 2000 12:51 PM, Martijn Kruithof <[EMAIL PROTECTED]>
wrote:

>I actually had an assignment checked in twice after a communication error
>with the prime 95 client. The second time was _long_ after the first time,
>and the first time came through successfully. I am sure the test was not
>re-run. I hope this one was not counted as check and doublecheck!

Well, if you still remember what the exponent was, you can download and
decompress the lucas_v file and check.  Or drop me an email if you want me
to look.  The file is almost 11 MB decompressed, but I have a copy sitting
on my disk at the moment.

- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: 12 Sep 2000 01:25:49 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Mersenne: Re: Double check validity

On Mon, Sep 11, 2000 3:10 PM, George Woltman <[EMAIL PROTECTED]> wrote:

>An exponent is considered double-checked if either tested by a different
>program or tested by prime95 using a different shift count (called offset
>by some).
>
>If the first and second tests were done by the same user, then my program
>prompts me as to whether I really want to accept this result.  I usually
>accept the double-check for the following reasons:
>         1)  It is not uncommon for top producers to get assigned a
>             double-check where they did the first test.
>         2)  It is not uncommon for folks to "screw up" the setup in
>             multi-machine environments.  A process called "ghosting"(?)
>             results in the prime95 directory being copied from a central
>             server.  Now each machine has the same worktodo.ini file.
>             Some users make the same mistake manually setting up
>             multiple machines initially.
>         3)  How likely is it that someone goes to the trouble of figuring
out
>             how to forge the 32-bit verification code on each results
line
>             and then use that info to turn in bogus *composite* results?
>             After all, such a clever person might well be smart enough to
>             get a variety of userids too.
>


Ok, I got curious and wrote a little program that searches lucas_v.txt
for adjacent lines that are exact duplicates.  Throwing away the tedious
part of the output that lists the duplicates, the summary from the program
is:

 Found     132497 unique exponents.
        136 of them had duplicated records.
         52 of them had ONLY duplicated records.

A few observations:
1.  The counts may be slightly off due to weirdness in the data file.
    My program is pretty stupid.

2.  There are few enough bad cases that processing them by hand shouldn't
    be too hard.

3.  A lot of the duplications involve "team" accounts, and many of them
    are duplicated a lot more than two times.  This could well be due
    to point #2 in George's list above.  Or it could be some kind of
    network error.  But if 20-30 machines really ran the same LL test,
    it's a shame to have wasted the resources.

4.  The "duplicates only" tend to be the bigger, more recent exponents.

5.  The duplicates were reported by various programs - mostly the
     WT, WU, WV, and WW series of programs.  I guess that means that
     manual check-ins are NOT responsible for the duplications.


If anyone wants the program, or the output (650 lines), let me know.
Otherwise, I'll go back to waiting for my next LL test to finish.

- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: Tue, 12 Sep 2000 08:06:19 -0000
From: "Brian J. Beesley" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On 11 Sep 00, at 23:43, Robert Deininger wrote:

> I'm not particularly worried about intentional cheating, just accidents.

That seems reasonable.

> OTOH, there isn't anything like a verification code on the manual check-in
> page.

This would be fairly hard to arrange. The point is that most of the 
programs which require manual check-in have no code to generate the 
verification code, and are unlikely to get it as the algorithm is 
"secret" - publishing it would make the whole exercise of generating 
the verification code pointless!

We should probably be thinking about using digital signatures 
instead; unfortunately this would require changes to the PrimeNet 
server to verify these and also to Prime95 (etc) to generate & 
transmit the required signature instead of the verification code.
> 
> Maybe non-prime95 results are so rare they aren't worth thinking about.

(I take it you're using Prime95 as a general label covering mprime, 
NTPrime etc. as well i.e. everything generated from George Woltman's 
optimized assembler code)

Hey! Maybe there aren't _too_ many of us around, since Intel-
compatible PC hardware is so pervasive (though I wonder why there are 
so few Macintosh users contributing results). However, as a user 
running Mlucas on a couple of systems, I do most certainly consider 
that they _are_ worth "thinking about".

Double-checking at least a proportion of tests using independent 
hardware and software is surely highly desireable. 

However it is true that, to the best of my knowledge, none of the 
"other" programs are capable of safely double-checking results 
produced by the same program. So, if username & computer id 
duplicates _not_ involving "W" series programs exist, these could 
very well be "dubious".


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

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

Date: Tue, 12 Sep 2000 08:06:19 -0000
From: "Brian J. Beesley" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On 12 Sep 00, at 1:25, Robert Deininger wrote:

> 4.  The "duplicates only" tend to be the bigger, more recent exponents.

... in the range (exponents over 6 million) where we don't yet expect to 
have had double-checking assignments?
> 
> 5.  The duplicates were reported by various programs - mostly the
>      WT, WU, WV, and WW series of programs.  I guess that means that
>      manual check-ins are NOT responsible for the duplications.

I wonder whether there was an incident on the PrimeNet server where a 
batch of results collected from the network was accidentally 
processed twice. If so the resulting duplication may be detectable at 
the server log file.

Note that, for various reasons, some of us who _are_ running double-
checks on large exponents - using George's programs - tend to be 
reporting results manually direct to George, avoiding even the 
PrimeNet manual testing page since we are often working in zones that 
the PrimeNet server appears to be unaware of.


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

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

Date: Tue, 12 Sep 2000 10:40:22 +0200
From: "Hoogendoorn, Sander" <[EMAIL PROTECTED]>
Subject: RE: Mersenne: Re: Double check validity

> George Woltman wrote:

>If the first and second tests were done by the same user, then my program
>prompts me as to whether I really want to accept this result.  I usually
>accept the double-check for the following reasons:
>         1)  It is not uncommon for top producers to get assigned a
>             double-check where they did the first test.
>         2)  It is not uncommon for folks to "screw up" the setup in
>             multi-machine environments.  A process called "ghosting"(?)
>             results in the prime95 directory being copied from a central
>             server.  Now each machine has the same worktodo.ini file.
>             Some users make the same mistake manually setting up
>             multiple machines initially.

But in this case the program doesn't use a different shift count, does it?
It could also be that the p and q files are also copyed to another machine 
and if there already is an error in those files the residues still match.
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ      -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt

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

Date: 12 Sep 2000 10:00:29 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On Tue, Sep 12, 2000 4:06 AM, Brian J. Beesley <[EMAIL PROTECTED]>
wrote:
>On 12 Sep 00, at 1:25, Robert Deininger wrote:
>
>> 4.  The "duplicates only" tend to be the bigger, more recent exponents.
>
>... in the range (exponents over 6 million) where we don't yet expect to 
>have had double-checking assignments?

Yes, the biggest duplicate is around 10.3 million.  They're scattered
around
the map pretty evenly.  But the biggest double-check on the primenet server
is in the 10.4 million range.  (My lucas_v is 4 or 5 days old, so the
correlation is a bit off anyway.)

>> 5.  The duplicates were reported by various programs - mostly the
>>      WT, WU, WV, and WW series of programs.  I guess that means that
>>      manual check-ins are NOT responsible for the duplications.
>
>I wonder whether there was an incident on the PrimeNet server where a 
>batch of results collected from the network was accidentally 
>processed twice. If so the resulting duplication may be detectable at 
>the server log file.

Something like that probably happened in some cases.  The worst exponent
I noticed is duplicated 43 times.

- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: 12 Sep 2000 10:21:47 -0400
From: "Robert Deininger" <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

On Tue, Sep 12, 2000 4:06 AM, Brian J. Beesley <[EMAIL PROTECTED]>
wrote:
>On 11 Sep 00, at 23:43, Robert Deininger wrote:

>> Maybe non-prime95 results are so rare they aren't worth thinking about.
>
>(I take it you're using Prime95 as a general label covering mprime, 
>NTPrime etc. as well i.e. everything generated from George Woltman's 
>optimized assembler code)

I'm not sure what I meant.  Since I don't use any of the intel variants,
I'm not up on the lingo.  I guess I was thinking about the programs that
only allow manual check in.  Can all the intel versions talk to the server
automatically?

>(though I wonder why there are 
>so few Macintosh users contributing results). 

I don't know.  Does the mac program require manual check-out and check-in?
Those web pages are rather hard to find and are persnickety to work with.
Some of the options don't seem to work at all.  From what I've seen with
Mlucas, some folks will use any little excuse to give up.

>However, as a user 
>running Mlucas on a couple of systems, I do most certainly consider 
>that they _are_ worth "thinking about".

Agreed.  I _only_ use Mlucas.  But I don't know how far out in the
wilderness that leaves me.

>Double-checking at least a proportion of tests using independent 
>hardware and software is surely highly desireable.

Yes.

>However it is true that, to the best of my knowledge, none of the 
>"other" programs are capable of safely double-checking results 
>produced by the same program.

Yes and no.  Mlucas is the same software no matter where it runs, but
runs on a wide range of hardware.  Two different hardware platforms are
probably ok to check each other, even with the same program.  We might
get a fairly independent check of the Mlucas software by forcing the
second run to use a non-default FFT size - but probably noone does that.


- ---------------------------
Robert Deininger
[EMAIL PROTECTED]



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

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

Date: Tue, 12 Sep 2000 18:49:34 +0200
From: Paul Landon <[EMAIL PROTECTED]>
Subject: Mersenne: The quadrillionth bit of Pi is '0'

A big round of applause to Colin Percival and the PiHex distributed
computing project which finished yesterday.
The project has now ended, though their web page says that there
will be future projects under the name idlepower.net and states:-

To those who cannot or do not wish to follow me to this new project, I suggest that 
GIMPS would be worthy of your cpu cycles. It is in my opinion the most mathematically 
interesting project currently running, and needs all the cycles it can get.

http://www.cecm.sfu.ca/projects/pihex/index.html


Well done,
Paul Landon

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

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

Date: Tue, 12 Sep 2000 13:34:40 -0400
From: George Woltman <[EMAIL PROTECTED]>
Subject: Re: Mersenne: Re: Double check validity

Hi,

At 01:25 AM 9/12/00 -0400, Robert Deininger wrote:

>  Found     132497 unique exponents.
>         136 of them had duplicated records.
>          52 of them had ONLY duplicated records.
>
>2.  There are few enough bad cases that processing them by hand shouldn't
>     be too hard.
>
>3.  A lot of the duplications involve "team" accounts, and many of them
>     are duplicated a lot more than two times.

Send the results here and I'll be glad to review the lucas_v entries.
I'm sorry that the lucas_v file on the Internet does not include the
shift count.  It isn't there simply to reduce file sizes.

Regards,
George

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

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

Date: Thu, 14 Sep 2000 18:38:06 +0100
From: "Thomas Womack" <[EMAIL PROTECTED]>
Subject: Mersenne: Why do I get only double-checks

I'm using a P3-500E box at college to run mprime.

The relevant lines of the local.ini look like

LastEndDatesSent=967023734
RollingStartTime=968950008
RollingAverage=4000
CPUType=1
CPUSpeed=500
CPUHours=24

How come I get given only double-checks to do? I don't mind too much -- the
machine kills off a double-check in ten days -- but I'd be slightly
surprised if the frontier of 'obsolete; use for double-checks only' had
reached this one-year-old box yet. That Rolling Average figure looks kind of
high, too.

Tom

_________________________________________________________________________
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 #777
******************************

Reply via email to