[QUOTE=R.D. Silverman;294560]Why in blazes does [b]any[/b] of this matter???? Why could it possibly be important to compare GHz-days with other posters???[/QUOTE]It increase the enjoyment of the task. Making something more enjoyable is beneficial to the overall effort.
Are you against people enjoying DC number crunching?

It also give a basis for comparison. By having a metric to compare with others, this tends to stimulate people into getting more resources involved.
You aren't against more G/CPU resources being applied to this project are you?

[QUOTE=R.D. Silverman;294560]Why in blazes does [b]any[/b] of this matter???? Why could it possibly
be important to compare GHz-days with other posters???

Eliminating a Mersenne candidate by finding a factor is important; It
reduces the total time required by the project.

But accumulating GHz credits??? Bah. Humbug.[/QUOTE]

You misunderstand my intent.

I do enjoy numbers, and counting things, and accumulating points...

However, in this particular thread, I am trying to find the correct formula (something all good mathematicians strive for) to calculate Ghz-days.
I am trying to create a process that I could input many exponents and determine the work-effort (aka approximate elapsed time) to process them.
I found a relatively recent post (ok 2008) from George with the formula noted but when I coded it I did NOT get the same numbers as I get on James' site.
I was hoping someone could just point me to a more recent (and correct) George post. Instead I found James source and did my best to interpret his PHP code (a language I never used before).
Then, never one to assume, I ran a real life test to verify that James' formula got the same results.

[QUOTE=petrw1;294573]Instead I found James source and did my best to interpret his PHP code (a language I never used before). Then, never one to assume, I ran a real life test to verify that James' formula got the same results.[/QUOTE]

Are you more comfortable with Perl? I converted James' PHP code into a Perl module -- I'd be happy to share this with you if you would find it helpful.

[QUOTE=R.D. Silverman;294560]Why in blazes does [B]any[/B] of this matter???? Why could it possibly
be important to compare GHz-days with other posters???

Eliminating a Mersenne candidate by finding a factor is important; It
reduces the total time required by the project.

But accumulating GHz credits??? Bah. Humbug.[/QUOTE]

I am the ghost of Christmas past.

David

[QUOTE=chalsall;294575]Are you more comfortable with Perl? I converted James' PHP code into a Perl module -- I'd be happy to share this with you if you would find it helpful.[/QUOTE]

please do.
Thanks

[QUOTE=petrw1;294580]please do. Thanks[/QUOTE]

[URL="http://www.gpu72.com/software/Timings1_0.pm"]http://www.gpu72.com/software/Timings1_0.pm[/URL]

Place this into a directory called "Mersenne" somewhere in your Perl search path or where you're going to place modules for your particular program. Then, for example:

[CODE]#!/usr/bin/perl

use lib '[COLOR="Red"][YOUR LOCAL MODULES PATH][/COLOR]';

use Mersenne::Timings1_0;

print "GHz Days to Trial Factor 46104067 from 71 to 72 is: "
. CalculateGHzDaysTF(46104067,71,72) . "\n";
[/CODE]

CalculateGHzDaysLL() and CalculateGHzDaysP1() are also included.

Thanks again for James (and, of course, George) for doing the heavy lifting for this.

[QUOTE=R.D. Silverman;294560]Why in blazes does [b]any[/b] of this matter???? Why could it possibly
be important to compare GHz-days with other posters???

Eliminating a Mersenne candidate by finding a factor is important; It
reduces the total [B]time[/B] required by the project.

But accumulating GHz credits??? Bah. Humbug.[/QUOTE]

the bold could be one answer a persons computer can operate at a certain effort/time that effort/time and the amount of effort needed for an assignment can tell you the time it can take you no ? if your computer can't handle it in a certain time maybe you could try other assignments.

[QUOTE=petrw1;294573]You misunderstand my intent.

I do enjoy numbers, and counting things, and accumulating points...
[/QUOTE]

Why on earth would "accumulating points" matter to anyone interested in
computational number theory????

[QUOTE]
However, in this particular thread, I am trying to find the correct formula (something all good mathematicians strive for) to calculate Ghz-days.
[/QUOTE]

"The correct formula".

What makes you think that such a thing exists??? I will give a hint:
It does not, at least in the sense that you seem to want.

If it does exist it will be a function of [b]many[/b] variables.
It is highly dependent on the exact hardware on which you are running,

(CPU, L1, L2, L3 cache sizes, memory latency, memory bandwidth, WORD SIZE!!, etc. etc.)

the OS that you are using, the compiler that was used to compiler the
code, etc. etc.

Any why would anyone care about such a thing? Certainly computational
number theorists/mathematicians do not.

If you want to measure your "GHz days", then just TIME YOUR CODE and
multiply by the clock rate of your machine.

This thing you want to measure is MEANINGLESS.

[QUOTE=petrw1;294573]You misunderstand my intent.

I do enjoy numbers, and counting things, and accumulating points...

<snip>

.[/QUOTE]

I fully realize that many of the posters herein are amateurs with a total
disdain for actually reading about this subject. However, I [b]strongly[/b]
urge you to read

[url]http://dl.acm.org/citation.cfm?id=621296[/url]

for an explanation of why this "GHz-Days" measure is [i]meaningless[/i].

[QUOTE=R.D. Silverman;294666]
(CPU, L1, L2, L3 cache sizes, memory latency, memory bandwidth, WORD SIZE!!, etc. etc.)
[/QUOTE]
False. With the exception of the word size (but this is fixed to 32 bits too), all the other would have no influence on the formula. One thousand different hardware doing the same job would do the same amount of GD of work. Of course, faster hardware will do the job faster. But the same amount of GD of job. What a pity I am in his ignore list... :razz:

[QUOTE=R.D. Silverman;294668]...for an explanation of why this "GHz-Days" measure is [i]meaningless[/i].[/QUOTE]

It's not meaningless from the perspective that it's the measure that PrimeNet uses to calculate "points", which some use to compare their contributions to others, to compare the relative performance of different hardware systems, etc.

As an analogy... Personally, I can't stand team sports. Don't understand the interest. But many of those who do go anal on all kinds of statistics which in my mind is pointless.

Different strokes for different folks. Deal with it, and let us have our fun.

This is, after all, about having geeky fun. :smile: