Mersenne Digest Tuesday, 2 March 1999 Volume 01 : Number 518
----------------------------------------------------------------------
From: "Ethan Hansen" <[EMAIL PROTECTED]>
Date: Tue, 2 Mar 1999 01:18:13 -0800
Subject: RE: Mersenne: Fabs & Pentium III
Just a note of explanation and some (decidedly unofficial) benchmarks:
P6 family processors:
Pentium Pro: The original 6th generation CPU; had 256, 512, or 1024K L2
cache, running at core speed. Bus speed: 60-66MHz; Internal speed: 150-200
MHz.
Pentium II (Klamath): Slightly optimized PPro core with MMX instruction set.
512KB cache running at 1/2 core speed. Bus speed: 66MHz; internal speed:
266-350 MHz. 0.35um process.
Pentium II Xeon: Same core as Klamath; 512, 1024, or 2048KB L2 cache running
at core speed.
Pentium II (Deschutes): Shrink of Klamath onto 0.25um process, minor
optimizations. Bus speed raised to 100MHz, core speed 300 - 450 MHz. Also
in Xeon version.
Mobile PII (Dixon): Uses Deschutes core, speed limited to <350 MHz for
thermal and battery power consumption reasons. Packaged in a much smaller
version of the PII cartridge; uses a pin socket rather than Slot 1.
Pentium III (Katmai): Deschutes + 32MB additional L1 cache + new MMX (KNI)
instruction set. Speed up to 550 MHz (for now). Also in Xeon version.
Upcoming PIII (Coppermine): Katmai core shrunk onto 0.18um process;
additional optimizations, and on die L2 cache >= 256KB, depending on
version. Supposed to be introduced at core speeds of >= 600 MHz; bus speed
of 100 or 133 MHz.
Relative timings at equivalent core speeds for Prime95:
Baseline: Pentium II = 1; 384K FFT size.
PII Xeon: 512KB = ~3% improvement, 1204KB = +10%, 2048KB = +25%.
Improvements ~double with FFT lengths <= 1/2 L2 cache size.
PIII (non-Xeon): +6-8%; +12-15% for 96K FFT length. Speed increase appears
to be from increased L1 cache, which makes sense as accessing the L1 is 2-6x
faster than accessing L2.
Regards,
Ethan
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Gordon Spence <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 13:34:01 +0000
Subject: Mersenne: Category 1,2 or 3 ?
>From: Spike Jones <[EMAIL PROTECTED]>
>Date: Mon, 01 Mar 1999 21:57:59 -0800
>Subject: Mersenne: who has the most wicked-fast home setup?
>
>Let category 1 GIMPSers be set that contains those GIMPSers that
>use a number of commercially owned or institutional computers, such
>as the champion TempleU.
>
>Let category 2 GIMPSers be set that contains those GIMPSers that
>use a combination of commercial and home computers.
>
>Let category 3 GIMPSers be set that contains those GIMPSers that
>use only computers located in the home of that GIMPSer and owned
>by that GIMPSer.
>
>Are there any cat 3 GIMPSers in the top 100 producers? Top 200?
>Who is the highest ranking cat 3 GIMPSer?
I have been thinking about this for quite a while now. When I joined GIMPS
back in thee mosts of pre-history (well, Fall 96 actually) there were only
about 800 or so volunteers in the project.
In thos days it was quite easy to actually make a mark in the rankings
table, I seem to recall at one point just after finding M2976221 that I
moved up to about number 78 in the rankings.
The only machines I had were a P100 (home, 24 hours), PPRO200 (work, 24
hours) and about 6 hours a day on 3 or 4 P166's.
Nowadays I still have the trusty P100 that found the prime, but I also have
8 other machines that I can get some access time on, my ranking has now
disappeared into the depths.
What bothers me most I guess is that the ordinary humble joe in the street
being realistic has no chance whatever of finding the next mersenne prime.
Over-reacting? maybe, but when people like templeu have over 300 machines
of PPRO200/PII300+ running think about it. They are generating about 14
months of cpu per day.
If we are going to continue to recruit new *individual* members or thos
like me who can get time on a few computers as opposed to hundreds then
shouldn't we be thinking about splitting the rankings based on the numbers
of machines being used.
Of course, if I could get my employer signed up and use all the 25,000+
pc's then I may well change my mind ;-))
regards
G
Gordon Spence, Nokia IP Telephony
Applications Engineer Grove House, Waltham Way,
[EMAIL PROTECTED] White Waltham, Maidenhead,
http://www.nokiaiptel.com/ Berkshire, SL6 3TN, UK.
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 09:55:26 -0500
Subject: RE: Mersenne: Pentium III
At 01:10 PM 2/28/99 -0500, you wrote:
>I think that when the Celeron first came out it was a Pentium II whose
>high-speed *external* cache was broken. The on-chip cache was still
>fine, otherwise a cached 386 could probably outrun it.
There were cached 386s? I thought they introduced the radical new concept
of caching with the 486. :-)
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:01:30 -0500
Subject: Re: Mersenne: Double-checking (was Chronons)
At 02:01 PM 2/28/99 -0500, you wrote:
>Finding an error in the first LL test is not rare. I've said about 1 in
>200 are incorrect.
Yeah, but most of those are "silent mutations" in nonzero residues, not
errors in the purported primality result, right?
What's the rate of primes missed the first time or composite numbers that
cause bogus primality claims?
Also what causes the errors, bugs in the code? Is work being done on
finding subtle errors in the software? Are some of them, in the case of
Prime95, caused by Winblows? (IOW, are there more errors per P90 CPU hour
among Winblows boxes than among mprime boxes?)
I figure only a small percentage of participants have actual faulty
hardware, and that spurious cosmic ray bit flips are caught by checksumming
of some kind. (Is that what ILLEGAL SUMOUT is, a checksum mismatch?)
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:04:37 -0500
Subject: Re: Mersenne: Fabs.
At 11:02 AM 2/28/99 -0800, you wrote:
>At the risk of stating the obvious...
This was not obvious to most of the audience. Although there does seem to
be a peculiar concentration of people with lots of hardware technical
knowledge here (perhaps because the performance of CPU-intensive software
is at issue here), it is a mistake to assume that uncommon hardware-related
acronyms will be universally understood here.
[Agh, yet another posting almost missent. Why, oh why, can't this
mailing list conform?]
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:08:46 -0500
Subject: RE: Mersenne: Fabs.
At 12:05 PM 2/28/99 -0800, you wrote:
It all seemed quite believable, until suddenly:
>When I was working with some of the first digital designs in the
>100-600 MHz range over 25 years ago...
Oops. I don't know if this is an extra-zeroes issue or a "M" instead of a
"K" or what, but just 10 years ago 1 MHz was good and 7 MHz was high end.
25 years ago, I very much doubt they were bandying about 600MHz anything,
since we're only just reaching the 600MHz level now at the end of the 1990s!
[Excessive quoted material snipped]
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:18:13 -0500
Subject: RE: Mersenne: Fabs.
At 07:21 PM 2/28/99 -0700, you wrote:
>It's not really like the 486sx which were 486DX chips with the FPU
>purposely disabled.
Con artists.
It should be fraud to design something, then sell some of them crippled for
what the uncrippled product is worth, and then sell the uncrippled product
as an "upgraded" or "bigger and better and newer" product charging EXTRA
for it.
In particular, since more work goes into making a 486SX than a 486DX, the
486DX should have been sold for slightly less than what they sold the 486SX
for!
Profiteering is simply not right. A product should sell for the cost of
manufacture. That should be enough to keep the bread on the tables of the
people that are involved in making the product. Intel can do without the
billions of dollars they hoard every year after putting bread on all their
employees' tables and paying their medical expenses and paying the expenses
associated with operations... ditto Microsoft.
These companies should put back into the community everything they take
from it. Not take more money in sales than what they put back via salaries,
via paying their corporate expenses, and via substituting a manufactured
product for the money in a sale transaction.
Sales in -> Manufactured objects, salaries, and expenses out into
the community. No hoarding. That's how it should be.
And Intel's business practises in particular are especially prone to
causing me to become incensed!
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:18:30 -0500
Subject: Re: Mersenne: Fabs.
At 04:32 PM 3/1/99 -0600, you wrote:
>Yikes! Apparently the "Elbrus E2K" is a 0.18 micron *386 compatible
>running at 1.2GHz
I don't suppose you can buy one of those for what it cost for them to
manufacture per unit, plus a small overhead for paying the manufacturer's
additional expenses such as employee wages, can you?
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:22:20 -0500
Subject: Mersenne: ARGH!
I eventually *did* missend not one but TWO messages.
PLEASE start to conform and set the reply-to to point back to the list like
every other list does!
And DON'T say I can use some sort of a "reply-to-all" feature. That's
equally tedious and error prone on my mail system, because it causes it to
want to send to BOTH the list AND the original sender, and then I have to
DELETE the extraneous address or else produce REDUNDANT messages. It is,
IOW, just as screwball as the method I am using now,
but in the other direction.
It should be as simple as point, click, type, and send. For everybody!
If it's not, something is broken -- either a bug or a misfeature.
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:25:47 -0500
Subject: Re: Mersenne: Mathematica's Parallel Computing Toolkit
At 05:57 PM 3/1/99 -0800, you wrote:
>Beta
>
>"...parallel programming over a network of heterogeneous machines..."
>
>http://www.wolfram.com/news/parallel.html
Let me guess: you have to PAY for the PRIVILEGE of doing distributed
mathematical computations using their $$$-WARE so that the guys on the
Wolfram Executive Board can HOARD LOTS OF MOOLAH that they cannot
conceivably fully utilize and that they have by no conceivable stretch of
the imagination *earned*. And in particular, if you are a poor student then
YOU MAY NOT EVER DO DISTRIBUTED MATHEMATICS except, using Prime95, on one
specific task...
Oh well. I'll put the bastards out of business soon enough with a freeware
product that's better... all in good time. If only I could do that with
hardware...
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:29:05 -0500
Subject: Re: Mersenne: mersennes for exocivilizations
At 11:17 PM 2/28/99 -0800, you wrote:
>It would not surprise me at all if instead of a list of primes, they
>send us the series 2,3,5,7,13,19,31,61,89,107,127,521...
I'm pretty sure that 61 is not a Mersenne prime... 2^6 - 1 is 63 and, I
might add, composite. :-) Indeed, the exponent 6 is composite, so this is
to be expected.
Come to think of it, of those, only 3, 7, 31, and 127 are Mersenne primes.
But there IS something familiar about that list...
[Another message nearly missent. Hurry up!]
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:30:54 -0500
Subject: Re: Mersenne: mersennes for exocivilizations
At 12:07 AM 3/1/99 -0800, you wrote:
>At 11:17 PM 2/28/99 -0800, Spike Jones wrote:
>>It would not surprise me at all if instead of a list of primes, they
>>send us the series 2,3,5,7,13,19,31,61,89,107,127,521...
>
>or 2, 3, 5, 7, 13, 17, 19, 31, 67, 127, 257 perhaps?
This one doesn't even ring a bell. The same Mersenne primes in it though,
and also some Fermat primes. (3, 5, 17, 257).
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Leyland <[EMAIL PROTECTED]>
Date: Tue, 2 Mar 1999 07:36:05 -0800
Subject: RE: Mersenne: Fabs.
> >When I was working with some of the first digital designs in the
> >100-600 MHz range over 25 years ago...
>
> Oops. I don't know if this is an extra-zeroes issue or a "M"
> instead of a
> "K" or what, but just 10 years ago 1 MHz was good and 7 MHz
> was high end.
> 25 years ago, I very much doubt they were bandying about
> 600MHz anything,
> since we're only just reaching the 600MHz level now at the
> end of the 1990s!
>
> [Excessive quoted material snipped]
[Excessive signature snipped 8-]
Eh? My first job after finishing my DPhil was microcoding a AMD-2900 series
bit slice machine which had a 25ns (40MHz) clock. That was in 1983 and it
was far from rocket science then. Actually, I started (unpaid) work on it a
couple of years earlier, so we're actually talking about 18 years ago.
The good old 1970's Cray-1 had a 9ns (110MHz) clock if I remember correctly.
*Twenty* years ago, a 4MHz Z80A was a commodity chip. I still have a 1979
model 380Z from Research Machines with that very cpu. (Incidentally, RM is
one of the few companies of that era still making PCs.)
I can't remember when I bought my 25MHz 386, but it must be approaching 10
years ago.
Paul
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:37:26 -0500
Subject: Re: Mersenne: Re: Mersenne Digest V1 #515
[Bogus looking AOL address deleted and replaced with
"[EMAIL PROTECTED]" -- I probably just saved myself a bounce]
...and it was all perfectly believable, too, until he blurted out:
>They even learn a couple of things from us (General Relativity is
>unknown to them - they only have Ntromdukian gravity)...
Oopsie. There is no possible way that some notion of relativity wouldn't
become known to an advanced civilization in this universe. They'd notice
the same problems with Newtonian-like mechanics that Einstein did, such as
its complete failure to account for some aspects of some orbits of some
objects, and its prediction that the speed of light will vary in different
directions on a moving planet, which is refuted experimentally. :-)
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:38:04 -0500
Subject: Re: Mersenne: Re: Mersennes for Martians
At 03:04 PM 2/28/99 +0100, you wrote:
>A shadowy flight into the dangerous world of a man who does not exist...
>Oliver Bonham-Carter, a young loner on a crusade to champion the cause of
>the innocent, the helpless, the powerless --in a world of criminals who
>operate above the law.
>-- By Order of the Fat Monkeys --
Whuh?
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:39:48 -0500
Subject: Re: Mersenne: Re: Mersennes for Martians
[3 missent today now. Come on! I'm wasting bandwidth here because
your interface is not user friendly and is error-prone.]
At 09:26 PM 2/28/99 -0500, you wrote:
>>At 09:25 PM 2/26/99 -0800, you wrote:
>>>Can anyone think of any way to *prove* the level of sophistication
>>>of our current society, better than a list of Mersenne primes? spike
>>
>>An image of the Mandelbrot set... <g>
>
>
>How are you going to transmit an image of a Mandelbrot set?
Duh.
With points on a grid, of course. Either prime-by-prime or square.
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:45:04 -0500
Subject: Re: Mersenne: Some diagrams
At 08:38 PM 2/28/99 +0100, you wrote:
>One of account had 0.0 CPU hours so I've changed it to 0.01.
? That looks bogus to me. Why would you do that? It's probably somebody who
signed up, did nothing, and will now get undeserved credit for not doing
nothing :-)
>I don't know why, but *.ps file cannot be viewed by gs under Linux
>(but xv displays them).
Since these are Postscript files and gs is a Postscript viewer, this is a
bug; complain about it. (Be sure to inform them that the bug seems to
depend on the target machine rather than appearing on all platforms.)
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
From: Paul Derbyshire <[EMAIL PROTECTED]>
Date: Tue, 02 Mar 1999 10:46:41 -0500
Subject: Re: Mersenne: RV: billions, trillions, ...
At 12:56 AM 3/2/99 +0100, you wrote:
>
>
>> 1E15 is a quadrillion dammit! 1E12 is a trillion!!! <g>
>
>�Uhmm..! In Europe usually a billion is 1E12 and a trillion is 1E18.
>1E15 is thousand billions...
>
>Ignacio Larrosa
[Obviously someone who uses the "reply-to-all" approach to work around
the interface wart, thus occasionally emitting duplicate messages]
Okay, so an entire small continent got it wrong. Since I don't live on that
particular misguided little landmass, this isn't a big concern of mine. :-)
- --
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
- -() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
------------------------------
End of Mersenne Digest V1 #518
******************************
