(Vanity) Free Republic Folding@Home Project update - We're in the TOP 850!!!

systematic ^ | 1-2-2006 | systematic

[systematic]

New thread for this week.

Congrats to all new members who joined this week!

We've made excellent progress so far and have smoked the DUmmies and Kossacks. Let's keep folding!

[Paladin2]

I find that that is essentially true.

[Petronski]

Most windows users have a couple dozen frickin-useless doodads in their system tray sucking up cycles.




Apple users only have half a dozen.

[papasmurf]

FReepFolders...Unchained!

FReepers, gone wild!

Ex-communicated FReepers.

FReepFolding, for fun and profit.

FReep'd from disease

FReepers, kinder, gentler...NOT!

FReepers saving the world, one protien ata time.








O.K., anybody in a better mood now?


:O)


P

[Daus]

I will oblige your flame bait, might it have to do with the complete lack of applications to run on Linux causing alot of idle time on those CPU's.

:)

[papasmurf]

Sure, because they're running the exe under VM or WINE. LOL


:O)

P

[Paladin2]

I thought this thread started out quite well before self important troll-like FReepers appeared claiming to have never seen most of those posting here.

The funny thing is that I've never run into these particular Luddites before either.

I can understand JR's lack of interest in somthing relatively divergent from the basic function/mission of the website.

[Senator Bedfellow]

Think you'd better check your math. Try dividing the number of active CPUs by the current number of teraflops, to find how many CPUs of each type it takes to add up to 1 teraflop. I get:

Windows: 1052 CPUs per teraflop
Linux: 821.5 CPUs per teraflop
Mac: 1762 CPUs per teraflop

I.e., it takes twice as many Macs as it does Linux machines to produce the same amount of work. Not that this really means much - it probably reflects how well optimized each client is, rather than real raw performance.

[Petronski]

I don't think refusing to participate is bad PR.


But taking part definitely would be good PR.





I guess if PR is not a priority, it doesn't matter anyway.

[Senator Bedfellow]

Maybe I'd better check my reading comprehension - upon re-reading your post, it looks like we said the same thing, in two different ways. LOL. I'm on a roll, what can I say? ;)

[Paladin2]

I'll respond by indicating that I'm currently primarily a Windoze user but am running ~50x the normal processor hours/day to support this team effort. The vast majority of my cpu cycles are NOT "spare".

I'd guess that the average Linux user has faster hardware than the average Windoze user. I shouldn't really speculate about the Apple users beyond noting that they may have [had] excess money.

[Paladin2]

I believe we've done the same math only using inverse divisions. Linux computers do twice as much useful work as do Apples. Linix is stwice as good as Apple.

[Paladin2]

Where were you when the Krauts attacked Pearl Harbor? LOL

[Daus]

I'd guess that the average Linux user has faster hardware than the average Windoze user

I would imagine there are some large, acedemic, Linux based clusters in the mix here as well.

Apple users beyond noting that they may have [had] excess money.

On that we can agree. :)

[Petronski]

Client statistics by CPU type

Type	Active	Active %	All	All %
Pentium	21	0.0	1276	0.0
Pentium MMX	66	0.0	3103	0.2
Athlon	56131	26.4	409036	30.4
Unknown **	30403	14.3	212404	15.8
Pentium 2 or 3	26721	12.5	290016	21.6
Pentium Pro	35	0.0	993	0.0
Pentium 4	98907	46.5	424791	31.6
Total	212284	100	1341619	100

Client statistics by OS type

Type	Active	Active %	All	All %
Windows	181998	85.7	1164961	84.6
Mac OS X	10574	4.9	65298	4.7
Other	0	0.0	0	0.0
Linux	19716	9.2	146075	10.6
Total	212288	100	1376334	100

Client statistics by Windows OS subtype

Type	Active	Active %	All	All %
Win 95	59	0.0	3122	0.2
Win 98	344	0.1	13241	1.2
Win 95_OSR2	9	0.0	410	0.0
Win ME	649	0.3	24577	2.2
Win 98_SE	2632	1.5	62925	5.7
Win 2000	23142	13.2	251674	22.8
Win NT	454	0.2	16081	1.4
Win XP	147789	84.4	730224	66.2
Total	175078	100	1102254	100

Last updated at Tue, 03 Jan 2006 15:53:41

DB date 2006-01-03 16:13:33

Active CPUS have returned WUs within 50 days.

** CPU Type is not detected for OSX or Linux

[ken in texas]

Yes, it seems to run pretty much the same speed that way. Just set your screen-saver to go to a blank screen and not spend cycles drawing pictures. :-)

[Senator Bedfellow]

D'oh!

:^)

[Petronski]

My only screensaver is a "sleep mode" for the monitor. So I'm golden.

I did shut down the nightly antivirus scans....I haven't found a virus in years anyway.

[Senator Bedfellow]

I don't know exactly why, but they've intentionally limited their x86 clients by not compiling in SSE2 optimizations, instead sticking to SSE only. If they chose to, they could boost the performance of their Windows and Linux clients quite easily - maybe something similar is playing into the relative slowness of the Mac clients.

[Paladin2]

I'm with you, let's go!

[Klutz Dohanger]

As I understand it, if I run the graphical version, but I close the window that does all the drawing, it runs as fast as the command line version. Is that true

It's "estimated" by what, I don't know, that the command line version might get you up to 5% performance increase, but I don't think anyone has stopped folding, long enough to benchmark that, and prove it right or wrong. If you've got a powerhouse video card, that offloads some of the processing from the cpu, I wouldn't think that there would be much difference at all.

And yes, closing the drawing window, removes a lot of overhead from the client, and is "close" to the speed of the command line version. But all computers are not alike, so actual mileage may vary :) But I would think that a 5% increase, would be the MAXIMUM performance gain you could expect, and realistically, a lot less.