[QUOTE=NBtarheel_33;326836]Which is on its way up. 95 TFLOPS in November, 125+ TFLOPS today. 30,000 additional registered CPUs in the same time frame. The new prime (and any subsequent discoveries) will boost these numbers.[/QUOTE]
No doubt, participation will keep on increasing. But how much gain in productivity can be expected from this? C&W was maybe a handful of people doing Mersenne prime hunting at that time (let's say b/w 10-100). Now we have tens of thousands (say between 10^4 and 10^5). So about 10^3-10^4 factor of the increase came from increasing participation. Can we sustain that now? Unlikely. Keep in mind that the planet's total population puts a hard limit on how much participation can increase.

Also, C&W was a bit of an outlier. Much bigger primes were known already by then. Instead let's look at GIMPS history. In 1996, GIMPS#1 (1398269) was found. So, in 17 years (+/-) we've grown by a factor of 40 (+/-). And we want to grow another factor of 60 in 25 years? Unlikely. Given the current GIMPS LL wavefront progression of 3-4M, we could be still testing 150M bit candidates only in 25 years time. That is far short of even 100m digit (~300M bit) primes, let alone a billion. I'm willing to concede that we have an outside possibility of finding a 100m digit prime. But billion is too far out.

[QUOTE=NBtarheel_33;326836]Technological advancement is full of surprises.[/QUOTE]
This is one case where I hope I'm wrong -- and not just wrong, but spectacularly wrong :smile:

[QUOTE=NBtarheel_33;326838]Not as long as George is at the helm of Prime95 development.[/QUOTE]
Indeed. Distributed Computing is one exception where programmer productivity takes a back seat to computer productivity :smile:

[QUOTE=LaurV;326839]Yeah, he might make the calculus faster, but displaying windows (or respective command prompt/text mode on mprime) is on the OS side and that will take 90% of the time... and become slower in time... :razz:
[/QUOTE]

You underestimate George. He'll write the GIMPS/OS. It will replace BIOS. All new computer in the world will automatically run P95 -- with perfect efficiency :razz:

[QUOTE=Batalov;326742]Yeah, I think so,
...but there's no reason [B]not[/B] to drink twice to that toast! (an adapted Russian saying)

MLucas ETA 37 hrs. (7am EST on Friday)[/QUOTE]

Grab some checkpoint files and do it in parallel, slacker!

[QUOTE=NBtarheel_33;326827]Note that Moore's Law (doubling of power every 18 months) ...[/QUOTE]That is not Moore's Law.

[QUOTE=NBtarheel_33;326827]OK, so to go from Colquitt/Welsh to Cooper^3 required a 500^3 = 125,000,000-fold increase in computing power over 25 years. To go from Cooper^3 to the first billion-digit test will require a 60^3 = 216,000-fold increase from what we have now. That seems eminently doable in the next 25 years.

Note that Moore's Law (doubling of power every 18 months) predicts a factor of 2^(25/1.5) = 104,000 increase in computing power over a 25-year span. We have seemingly outdone this by *three orders of magnitude* since 1988. It seems absurd to think we can't simply *double* it over the next 25 years.[/QUOTE]
The way I look at it is (10^9/(2*10^7))[SUP]3[/SUP] ~ 2^17

So I would guess ~25 years too:smile:

D

[QUOTE=NBtarheel_33;326827]OK, so to go from Colquitt/Welsh to Cooper^3 required a 500^3 = 125,000,000-fold increase in computing power over 25 years.[/QUOTE]
I think M[SUB]110503[/SUB] took 11 minutes to test on the NEC SX/2 which is about 0.006 sec/iteration. FFT size of 8K.

Crandall's irrational base discrete weighted transform would run over 4 times faster. Implementing non-power of 2 FFTs would bring further improvement. I suppose that if George or Ernst had programmed the SX/2, M[SUB]110503[/SUB]'s iteration time would have dropped to 0.001 second.

Over a year or two, Walt probably got only a couple of percent of the machine time. It was not running in background.

Another stat I recall is that PZILLA ran at a sustained 2.4 MFLOP and was the "world's fastest program". The SX/2 was a vector machine theoretically capable of 2.6 MFLOP.

[QUOTE=akruppa;326847]Grab some checkpoint files and do it in parallel, slacker![/QUOTE]

Like the central banks (cough, cough) of the world, we need to maintain our independence, or something.

Speaking of slackerdom, I must say that the gerbils are a tad miffed that none of the guess-the-prime ferrets has nailed the new prime's exponent to within a very narrow range using the obvious clues littering this thread.

Mike, how do the snakes feel about this?

[QUOTE=M29;326865]Another stat I recall is that PZILLA ran at a sustained 2.4 MFLOP and was the "world's fastest program". The SX/2 was a vector machine theoretically capable of 2.6 MFLOP.[/QUOTE]
I find it interesting that vector hardware is making a roaring comeback, in the guise of SIMD (= vector-data), GPU (= vector-thread), etc.

[QUOTE=ewmayer;326894]Speaking of slackerdom, I must say that the gerbils are a tad miffed that none of the guess-the-prime ferrets has nailed the new prime's exponent to within a very narrow range using the obvious clues littering this thread.[/QUOTE]
I don't recall gag orders in the past. But, from the [URL="http://www.mersenneforum.org/showpost.php?p=325892&postcount=471"]Predict M48[/URL] thread:
[QUOTE=Uncwilly;325892]A prime has been reported and George is happy so far.
[SIZE=3][COLOR=Red][FONT=Arial Black]All guessing is temporarily prohibited.[/FONT][/COLOR][/SIZE][/QUOTE]

[QUOTE=rcv;326908]I don't recall gag orders in the past. But, from the [URL="http://www.mersenneforum.org/showpost.php?p=325892&postcount=471"]Predict M48[/URL] thread:[/QUOTE]

You fail to understand humor....

[QUOTE=chalsall;326909]You fail to understand humor....[/QUOTE]
Apparently, so do the ferrets. I will make my official guess: "Martin".

[QUOTE=rcv;326908]I don't recall gag orders in the past. But, from the [URL="http://www.mersenneforum.org/showpost.php?p=325892&postcount=471"]Predict M48[/URL] thread:[/QUOTE]
Sheesh! That's to determine who will have won the [I]prediction[/I] contest!

You know, like when the ball is slowing down on the roulette and the croupier makes a distinct pass over the table meaning "No more bets"? How hard is it to [I]predict[/I] what the price of MSFT was at 11:01am [I]yesterday[/I]?