Yvan Dutil wrote:
> I was horrified to see than some individula LL-test have still
> years (no joke) of computation in front of them! Maybe, It would
> be wise to communicate with these people and:
This idea that the "snails" ought to be speeded-up or bypassed has been
discussed before.
Take it easy, Yvan, or I'll take my "snail" Pentium elsewhere, and
GIMPS will proceed more slowly _without me_ than it proceeds
_with me_ because someone else will have to take over the work
I would have been doing!
> 1-Check if they are still working on their problem.
There is already a procedure for catching dropouts that applies to all
participants.
How about this counterproposal -- Check up on the "fast" ones more
often, in order to ensure that they don't waste the large amount of CPU
cycles available to them. After all, if a guy with a 400 MHz Pentium
goofs off and doesn't run an L-L test for a day, GIMPS "loses" _FOUR
TIMES AS MANY CYCLES_ as it would if someone with a 100 MHz Pentium
fails to utilize 24 hours!
Clearly it is more important for GIMPS to keep cracking the whip over
participants with _fast_ Pentiums than to check on those with slower
CPUs.
> 2-Suggest them to upgrade their program.
There has been only a minute improvement in L-L testing speed in the
past several versions. I'll bet the loss of time involved in the
upgrading process would exceed any saving in time from faster L-L
testing.
> 3-Ask them to go on a faster machine.
Fine. Send me US$2500 and I'll be glad to upgrade! When may I expect
your international postal money orders to arrive?
> 4-Redirect them to a more productive task (i.e. Double-Checking small number).
To those of you who feel exasperated with the slowness of some other
GIMPS participants: Do you mind if we "redirect" you to a more
productive task than griping about such things?
Remember - an hour _you_ waste costs GIMPS a lot more cycles than an
hour someone with a slower CPU wastes! :-l
Richard Woods
