mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2002-08-18, 20:58   #1
ADBjester
 
Aug 2002

2·3·5 Posts
Default New baseline machine / figuring stats

First, I saw the Excel spreadsheet in the other thread -- excellent, and thanks! The rest of this I am crossposting from my post to the mailing list.

21000 of the 31000 participating machines are P-III or better.

Less than 2,000 true Pentium-class machines remain in the mix.

George et. al.: Could it be time to change the baseline reference machine away from the Pentium-90, and wipe the P-90 off of all pages, from rankings to status to years of computing time left to complete the task?

A couple years back, George changed the Status page reference to be a P-II/400, equivalent to 5.5 P90's. Now even that PII/400 is far less than the 'average participating machine", which given the above numbers, I'd guess is now about one gigahertz, perhaps slightly better.

I believe that a one-time re-indexing of ranks, stats, and "time left to compute" that re-indexes on either a P-III/1000 or an Athlon-1000, would make the "CPU years left" numbers on the status page a bit more realistic, as well as the number of "CPU years" I complete each day.

------------------------

Side note: Also of interest in both the benchmarks table and on the individual / top producers tables, would be a RECENT CPU hours/day comparison, as well as a machine reference back to the baseline machine, whatever it may be.

i.e. I've been with this thing from the beginning, in 1996. Obviously, my average machine has gotten better and better. My top listing says I'm doing about 1090 CPU hours a day.... but that's averaged over ALL of my submissions, dating back to when I was using 486's in 1996!

I did some arithmetic to try to figure out what I'm cranking out NOW.... (anyone want to check my logic here)?

i.e. how many CPU-hours a day is, say, an Athlon 1600+ worth?

According to the benchmarks page, the P-II/400 does a 15-17MM exponent iteration in 0.536 seconds. And we know that this machine is 5.5 P-90's. Thus, a P-90 would be expected to take 5.5 x 0.536, or 2.948 seconds.

My Athlon 1600+ takes .130 seconds per iteration.

2.948 / 0.130 = 22.677 times as fast at the P-90, so 22.677 x 24 hours means that this machine ought to be doing ABOUT 544.24 P-90 CPU hours per day.

If I add up what all my machines are doing NOW, I get 3503 P-90 CPU Hours a day, not the 1090 shown on my account and report.

------------------

What I'd like to see is:

1) On the individual account report, the above calculation (i.e. the 544.24)
shown next to the exponent/machine. This should not be ESTIMATED, but
reverse engineered from actual reported iterations per second for the
exponent, compared to 2.948 seconds for the P90 (or whatever a new
baseline might be).

2) A SUM of all of the above, to let one know how much they TRULY are
cranking out, as opposed to that slow creeping average that, after so
many years means nothing.

3) A "rolling average" for the last 6 months, for the Top XXX pages, so
that I can compare RECENT work to other recent work. i.e. I see that
I am surrounded by many others in the 1100 CPU Hours/day range....but
if my historical data is skewed so much by those old slow machines from
six years ago, how much are others skewed? Who do I have a chance to
pass? Who's gaining on me? I can't tell! A rolling average, or
perhaps the calculations from #2 above in a column instead of a rolling
average, would make comparisons in the Top XXX listings easier, and
much more meaningful.

-----------

Comments, suggestions, criticisms and flames welcome. ;)
ADBjester is offline   Reply With Quote
Old 2002-08-18, 21:31   #2
ebx
 
ebx's Avatar
 
Aug 2002

101 Posts
Default

I dont think the reference point is so important, as long as
it makes sense. P90 or PIII 1G it doesnt matter.

Your figure about your 1600+ is about right. The problem
is, the CPU clock isnt the only factor. Same 1600+ with
different memory, motherboard could run differently by
miles.

My 1800+ is about 34 P90. Although it could give me 816
P90 hours per day, there are a lot of things that reduce
this figure. First, even if I leave the box on 24 hours, it is
not running prime95 all the time. May be 95% of the time
if I do nothing else. Now, if you ever power down the box,
the number dropped dramatically. The number you saw is
daily average over the life of an account. This number are
more likely go down than go up unless you add new boxes.
I am seeing close to 700 at the beginning for my 1800+
but now sustains at just below 500.

Factoring only gets 10% credit(so I heard). That makes the
daily average smaller too.

Again, I dont care the P90 years that much either as long
as everybody is measured against the same rules. What
important is getting the job done.

However, figuring out the relative power of each CPU is
helpful.
ebx is offline   Reply With Quote
Old 2002-08-18, 21:37   #3
ebx
 
ebx's Avatar
 
Aug 2002

101 Posts
Default

I dont think the reference point is so important, as long as
it makes sense. P90 or PIII 1G it doesnt matter.

Your figure about your 1600+ is about right. The problem
is, the CPU clock isnt the only factor. Same 1600+ with
different memory, motherboard could run differently by
miles.

My 1800+ is about 34 P90. Although it could give me 816
P90 hours per day, there are a lot of things that reduce
this figure. First, even if I leave the box on 24 hours, it is
not running prime95 all the time. May be 95% of the time
if I do nothing else. Now, if you ever power down the box,
the number dropped dramatically. The number you saw is
daily average over the life of an account. This number are
more likely go down than go up unless you add new boxes.
I am seeing close to 700 at the beginning for my 1800+
but now sustains at just below 500. Just like driving. Cant
do the peak speed all along.

Factoring only gets 10% credit(so I heard). That makes the
daily average smaller too.

Again, I dont care the P90 years that much either as long
as everybody is measured against the same rules. What
important is getting the job done.

However, figuring out the relative power of each CPU is
helpful.
ebx is offline   Reply With Quote
Old 2002-08-18, 23:46   #4
ADBjester
 
Aug 2002

2×3×5 Posts
Default

Quote:
Originally Posted by ebx
Your figure about your 1600+ is about right. The problem
is, the CPU clock isnt the only factor. Same 1600+ with
different memory, motherboard could run differently by
miles.
I wasn't using the clock speed. I was using the comparative iteration times to determine how many times faster the 1600+ was than a P90.

You also say reboots and running apps affect faster boxes than slower ones, but for boxes that rarely reboot (maybe once per LL test), and have a fairly constant load, the drain ought to be about equal in terms of percentage. The comparison is fairly valid.

Quote:
Originally Posted by ebx
Factoring only gets 10% credit(so I heard). That makes the
daily average smaller too.
Depends on how much factoring you do. I've been at this since 1996, and I don't think I've done much factoring at all.... certainly no more than 1/5000th of the LL test time I've done. (About all the factoring I've done is bringing exponents up to their thressholds).
ADBjester is offline   Reply With Quote
Old 2002-08-18, 23:48   #5
Tasuke
 
Tasuke's Avatar
 
Aug 2002

23·32 Posts
Default

One point that was touched on, is that there wasn't a plethora of p90 chipsets available, differing bus, memory, and cache speeds that are abound nowadays. By using a metric that has the least amount of devience, no matter how slow or small, ensures that apples can be compared to apples. The more modern the CPU used, the more Uknowns come into play(like cache latency, cache size, memory speed, memory latency, chipset settings, non- 2- or 4-way interleaving, DDR, etc.)
Tasuke is offline   Reply With Quote
Old 2002-08-19, 02:19   #6
ebx
 
ebx's Avatar
 
Aug 2002

101 Posts
Default

Quote:
Originally Posted by ADBjester
I wasn't using the clock speed. I was using the comparative iteration times to determine how many times faster the 1600+ was than a P90.

You also say reboots and running apps affect faster boxes than slower ones, but for boxes that rarely reboot (maybe once per LL test), and have a fairly constant load, the drain ought to be about equal in terms of percentage. The comparison is fairly valid.
What I meant was given 2 1600+, their speed may vary a lot.

Reboot itself isnt that bad. Shutdown is.

If you started from 1996, on a slower machine I assume, your history is too
heavy. Not easy to bring the average up.

If it is real important, a 'sliding window' implementation may indicate recent
activities better. This is how most cpu load monitors are done.
ebx is offline   Reply With Quote
Old 2002-08-19, 03:18   #7
Complex33
 
Complex33's Avatar
 
Aug 2002
Texas

100110112 Posts
Default

I personally would not like to see a change in the baseline of the statistics. If it makes you feel better perhaps we can call it something other than P90 years... maybe points, stars, blue elephants then it doesn’t have the P90 designation. But seriously I think we should stick with what is tried and true.

I do agree that it would be quite useful to have a rolling history of our production. Something similar to RC5 might be good say with a 30 day rolling average of production or longer with all these 33M flying around. This could then give a gage of overtake dates ect. I particularly like the implementation that TPR has on their overtake page. Applied to the rankings could be beneficial.

As I have stated before I have no wish to impose on the great benefactors of this project.

:D Complex
Complex33 is offline   Reply With Quote
Old 2002-08-19, 12:10   #8
jeff8765
 
jeff8765's Avatar
 
Aug 2002
A Dyson Sphere

32×7 Posts
Default

I like the idea of the rolling average, but I like the P-90 computer years statistic. It made me feel good when I first started a few months ago that I would soon have done a full P-90 computer year.
jeff8765 is offline   Reply With Quote
Old 2002-08-23, 21:58   #9
PageFault
 
PageFault's Avatar
 
Aug 2002
Dawn of the Dead

111010112 Posts
Default

I looked in some logs from an ancient 1100 tbird and found that, in the same time it took to do a LL test worth about 1.0 P90 years, it did about 1.1 P90 years of trial factoring.

outlnder can point you to an app which calculates trial factoring credit.

Quote:
Originally Posted by ebx
Factoring only gets 10% credit(so I heard). That makes the
daily average smaller too.
PageFault is offline   Reply With Quote
Old 2002-08-23, 22:05   #10
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2·29·137 Posts
Default

Quote:
Originally Posted by PageFault
outlnder can point you to an app which calculates trial factoring credit.
http://www.teamprimerib.com/db1/p90.exe
Xyzzy is offline   Reply With Quote
Old 2002-10-04, 20:47   #11
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

26·7 Posts
Default

Well, I've not gotten any of my K5 machines swapped over to Mersenne from d.net yet - have to upgrade the LINUX kernel on all of them, they're all running 2.0.somethingold that gives the "too old of kernel" error.

They'll all be doing factoring, though - no way I'm gonna waste one of those trying to do LL with it.

8-)
QuintLeo is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Figuring Out Sequence Merges Jayder Aliquot Sequences 13 2014-05-30 05:11
What to do about a vibrating machine (never *your* vibrating machine) sixblueboxes Hardware 14 2013-03-19 21:58
CSVs for stats available + New combined stats opyrt Prime Sierpinski Project 3 2010-05-31 08:13
New machine showing in stats as unknown endless mike NFSNET Discussion 1 2005-02-06 11:47
new machine junky NFSNET Discussion 11 2004-02-21 03:39

All times are UTC. The time now is 12:57.

Wed Jan 27 12:57:12 UTC 2021 up 55 days, 9:08, 0 users, load averages: 4.63, 4.64, 4.74

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.