Prime95, glitch or reality?
- I've been running Prime95 with a smaller B1/B2 and a huge exponent
I realized this morning that the box has been stuck on the same curve
for the last 18 hours, when it normally finishes a curve in 2 hours.
So I checked the Prime95 folder and found a results.txt file with the
- - -
[Sat Jul 02 16:05:36 2005]
ECM found a factor in curve #11, stage #1
Sigma=4573583067221178, B1=2000, B2=200000.
- - -
The screen display shows that Prime95 has finished stage-1 and
appears to be initializing for stage-2. No mention of a factor on the
screen display and no mention in results.txt as to what the factor
supposably is. And Prime95 is still chewing up 100% of the CPU.
So thinking it may have been a glitch, I switched to another box with
more memory. I attempted to run the same identical curve with the
same B1/B2 and same sigma. Darned if it didn't go through the curve
without finding any factors.
Both boxes have the same lowm.txt and both boxes appear to have run
the same identical curve. Box-1 is still stuck (using 100% of the
CPU) and box-2 claims there was no factor.
Do I assume a glitch on box-1 and box-2 is correct? Or is there a
chance that box-2 didn't run an identical curve? Or is it time to
stick to smaller exponents since Prime95 sometimes seems to have
problems with very large factors when running ECM?
- From: "j_m_berg" <j_m_berg@...>
>With a B1 that low, it's quite likely that the factor, if it really exists,
> I've been running Prime95 with a smaller B1/B2 and a huge exponent
> I realized this morning that the box has been stuck on the same curve
> for the last 18 hours, when it normally finishes a curve in 2 hours.
> So I checked the Prime95 folder and found a results.txt file with the
> following entry.
> - - -
> [Sat Jul 02 16:05:36 2005]
> ECM found a factor in curve #11, stage #1
> Sigma=4573583067221178, B1=2000, B2=200000.
could be found with only trial division.
For each prime p of the form k*33554432+1, find 2^33554432 (mod p). If it's
one, that's a factor. That calculation is only 20 modular squarings (of 2^32),
and it should be trivial to hammer out 10^10 of those in no time at all.
Paul/David/Jim - you're the fast x86 guys - can you be of any help?
() ASCII ribbon campaign () Hopeless ribbon campaign
/\ against HTML mail /\ against gratuitous bloodshed
[stolen with permission from Daniel B. Cristofani]
Rekindle the Rivalries. Sign up for Fantasy Football