PDA

View Full Version : [email protected] vs [email protected]: Benchmarks



Fatguy3
12-06-2003, 05:28 AM
Well after a day of testing and resinstalling and writing, the results are in. Here is the much anticipated review I did of [email protected] to see if it had any negative affect to computer performance. I'll let the results speak for themselves. BTW: I did this in tweaktown fashion to make it a bit professional, and just in case a mod thinks its awesome enough to throw it up as an article I wouldnt have to do much cleaning up. Without further ado:

Introduction:

As many of you know I just recently started posting in the Tweaktown Forums, however Ive been using
the guides here for quite a while. One of the most recent additions to my hard drive was [email protected]
For those of you hiding under a rock in the Beer Garden, [email protected] is a Distributed Computing
Program developed by Standford University to research folding proteins. In English, it's a program similar
to [email protected] that hopefully will help cure disease. Couldn't be for a better cause. To find out more
go check out the website. http://www.stanford.edu/group/pandegroup/folding/

I had participated in [email protected] in the past, but stopped because it really hurt system performance when
I was using the computer, and I was ALWAYS using the computer. So it was pretty impratical. When I
heard about [email protected] I though, ugh, more slowdown, but I decided to give it a try. Well what I found, is that
because it is set as an Idle process in Windows, it was not hurting my system performance at all. Even
when playing games such as Need for Speed: Underground and Max Payne 2, I saw no slowdown or FPS
drop whatsoever. So I figured its time whip out the benchmarks and see if [email protected] is as well written piece of
non-intrusive software.

Test Rig:
So what we have here is a fine piece of machinery I like the call Baal ;) No really, its just my computer.
Quick stats:
Aopen AK77 Pro o/c'd to 144 mhz FSB
1600+ XP o/c'd to 1500 mhz
256 mb Elixir PC2100 at CAS2
Audigy Gamer
10/100 NIC
WD ATA66 HD
Xtasy Ti4200 o/c'd to 310/580

I am using a nice stable overclock to simulate a real world situation where [email protected] would be running behind
games/apps etc. The benchmarks I'll be using are SiSoft Sandra 2k4 to check is [email protected] idle CPU usage
affects a synthetic benchmark. 3dMark2001SE to see if [email protected] running behind DX8 games would affect
performance. And Aquamark3 to simulate a real world DX9 game, and because I think the scenes are
pretty cool. ;)

All Benchmarks were run after a clean install of XP Pro, with minimal background services, and no apps
on startup besides RivaTuner to O/C the video card. Each benchmark was run 3 times and then an average
was taken. The computer was restarted after each benchmark ran 3 times. Now, on to some numbers.


The Test:

Well I dont have Excel on my computer at the moment, so you won't get any bar graphs, but what you will
get are some pretty suprising results. First lets take a look at Sandra 2k4.

CPU Arthmetic Benchmark:
Control Test: 5682 Mips / 2350 Flops
[email protected] in BG: 5681 Mips / 2350 Flops

CPU Multimedia Benchmark:
Control Test: 13687 Int / 14012 Float
[email protected] in BG: 13688 Int / 14007 Float

Memory Bandwith Benchmark:
Control Test: 2093 Int / 1935 Float
[email protected] in BG: 2074 Int / 1938 Float

Cache & Memory Benchmark:
Control was exact same as [email protected] in BG.


I was really suprised with this. I figured if any performance degredation were to show up, it would have been in
the stressful Sandra Benchmarks. Boy was I wrong. There was no performance difference whatsoever.


AquaMark3:

Control Test: 19173
[email protected] in BG: 19220

Again, WOW! Actually had a slight performance boost with [email protected] in the Background. No scores on the control group
went over the lowest score attained while running [email protected]

3dMark2001Se

Control Test: 9597
[email protected] in BG: 9625

And once again, a really cool result, with [email protected] beating out the control test. Its starting to form a pattern. Onto the
rambling conclusion.


Conclusion:
One Word. Hotness.
What I've discovered is having [email protected] running in your background has absolutely no disadvantages to your overall
system performace at all. I also tried it out with 2 real world games, Max Payne 2 and Need for Speed Underground.
Using Fraps 2.0 I found no difference in framrate while [email protected] was running in the BG. So why not run [email protected]? You're not
hurting you're system performace, and you're helping to cure disease.

Get over to http://www.stanford.edu/group/pandegroup/folding/ and download the client and get cranking out those W/U's
And dont forget to join TweakTown's Team: ID# 33272 so you can get your name in lights in our stats:
http://folding.stanford.edu/cgi-bin/teampage?q=33272

Thanks to Minibubba for starting the TT [email protected] group and having his sig get me inetersted in [email protected]
And thanks to Tweaktown for being the coolest freakin Computer Enthusiasts to date, period.
Peace.

-Fatguy3

minibubba
12-06-2003, 05:40 AM
wow. that's pretty awsome results. If anything, I expect a slight hit to performance. Glad to see I was wrong:D
Now we just need a few more people to join up so we can get the team to the top, where it belongs :)
<center>:cheers:</center>
*btw, to the guys with the power... I wouldn't mind seeing something like this on the front page to stir up some intrest in the TT DC teams ;)

Perhaps this may be a good canidate for one of the new user reviews?

kane2g
12-06-2003, 11:52 AM
I had participated in [email protected] in the past, but stopped because it really hurt system performance when
What slowdown?
i play AAO and my FPS are the same as with SETI off. Maybe I dont have a pretty numbers to throw at you, but even with my ****ty GF4 440mx I dont have any problems running seti and playing games. :D
Maybe you should also do a test with the SETI setup (as described in the link in my sig) to compare it to [email protected]

Fatguy3
12-06-2003, 12:38 PM
Good Idea Kane, as my [email protected] experience was over 2 years ago, so Im sure things have changed, should have added that in the post.

minibubba
12-06-2003, 12:41 PM
well kane2g, why don't you run the tests and add it here ;) I'm sure we would all like to see the comparisons.

besides, I wasn't kidding about wanting this (or something like it) on the main page to promote interest in both teams

Fatguy3
12-06-2003, 12:44 PM
I agree with minibubba, I think a small little review by Tweaktown on the front page would be a good idea. It would defintely stir up intrest in the User review contest, as pretty much anyone can review this stuff. Plus, dat mobo would be sweet. :clap:

kane2g
12-06-2003, 01:29 PM
no review is going to happen by me because:
all my systems are dedicated to crunching already :D
My settings (bios and windows) are so that my WUs are crunched the quickest possible.
Looks like Fatguy3 has a system with a clean install just waiting for [email protected] install. plus we could compare the numbers between [email protected] and [email protected] on the same system. which would be nice :2cents:

Fatguy3
12-06-2003, 05:25 PM
Well I just got done another round of testing. As I'm too lazy to write up a whole report right now, I'll give you the basic results.

In Sandra, there was no difference between a control test, [email protected] or [email protected] at all. 1 or 2 points here and there.

In 3dMark2001SE I found a VERY slight increase when not using a Distributed Program. However this could be attributed to fluctuating temps (snowing here in the NE) and I just got done tweaking out the vid card a little more. There was only like a 100 point difference though, nothing to write home about. However [email protected] and [email protected] were only a few points off of each other.

In Aqumark there was only a few points, 10 or 15 between all 3 tests. So basically, running [email protected] or [email protected] will not ruin performance at all, as far as I can tell.

Note: I am very interested in writing up a professional comparison to be entered into the User Review Competition. :wave: wink wink....

kane2g
12-07-2003, 03:26 PM
Thanks.
I knew [email protected] wasnt having any issues with performance! :thumb:
This deserves at least a sticky! right?

minibubba
12-07-2003, 03:28 PM
This deserves at least a sticky! right? sure, it's done :)
*and a little title revision too

waynebot
12-07-2003, 04:03 PM
*btw, to the guys with the power... I wouldn't mind seeing something like this on the front page to stir up some intrest in the TT DC teams ;)

Iv noticed most of the top teams have a link on the main web site front page. Also they put folding rite where you can see it on the main forum page, in DC forums or web Teams. I had to search to find the DC forums here. The people that run this site have the power to make our folding team a powerhouse...

elrado1
12-07-2003, 08:40 PM
... I had participated in [email protected] in the past... Are those Units parked in any meaningful location? Hint. Hint. (http://setiathome.ssl.berkeley.edu/stats/team/team_101812.html) ;)

Nice bit of analytical work Fatguy3. :thumb:

I do have some questions, however. I'm wondering how much progress was made on the [email protected] or [email protected] units during the benching/gaming vs the same length of time while running the DC program by itself? I suspect that the wu processing suffered at the expense of the benching/gaming.

I am not at all familiar with the [email protected] programs (GUI or CLI), however, there are three priority settings in SETI Driver for running the command line version of [email protected] They are 'High,' 'Normal' and 'Low.' What setting was used while benching with [email protected] running and what setting corresponds to the [email protected] setting? Or does the CLI version of [email protected] operate like the typical settings for the GUI version of [email protected] and back off completely while there are other processes with higher prioity running?

Fatguy3
12-07-2003, 09:18 PM
I only had it installed for about a week, on a P3 500. So even If I could remember my username, it wouldnt be worth while anyway

elrado1
12-07-2003, 09:55 PM
I only had it installed for about a week, on a P3 500. So even If I could remember my username, it wouldnt be worth while anyway Ok, thanks. BTW I've added some questions to my previous post.

Fatguy3
12-07-2003, 10:49 PM
#1 Although I did not record the times of units processed, performance defintely suffered during gameplay. This is to be expected, as @Home only eats up your idle processes. This was more of a test of Microsoft's CPU Priority Coding then anything really.

#2 Both progs were run in command line mode with the process set at lowest setting, Idle for [email protected] and Low for [email protected] Im sure if I upped the priority it would result in performance loss.

Thanks for the questions though. I may do some tests to see how long a certain frame takes without any progs running vs Gaming vs Sandra BurnIn Test (That would be a killer) But it really wouldnt prove anything that we didnt know. Oh well, perhaps sometime today. :)

amd_man2005
12-07-2003, 10:51 PM
Awesome work man! :thumb:

Fatguy3
12-07-2003, 10:54 PM
Danke, I just found my Works Suite 99 Disk (Super Ghetto Version) and I'm gonna draw up some sweet ass graphs and shizat, ya know, for the ladies :devil:

Wiggo
02-20-2004, 09:00 PM
Danke, I just found my Works Suite 99 Disk (Super Ghetto Version) and I'm gonna draw up some sweet ass graphs and shizat, ya know, for the ladies :devil: So what happened with those graphs? :?:

minibubba
02-21-2004, 01:42 PM
there's a very in-depth anyalsis of [email protected] and system performace here: http://www.tech-report.com/etc/2002q4/foldingimpact/index.x?pg=1

and it even includes graphs :)

here's the conclusion for those of you too lazy to follow the link and actually read it:
[b]The results of our testing couldn't be clearer, at least for the systems we've tested today. Quite simply, the impact of running [email protected] in the background is negligible. Even the most discerning users won't notice it. Windows allocates CPU time slices according to process priority, and as long as the [email protected] client is running as an idle-priority task (which is how it runs by default), one shouldn't notice any slowdown. Remember that we're also running FireDaemon in the background, and the impact of this small service management tool is also negligible.

With the results we've seen here, one more exuse for not donating excess CPU time to the [email protected] project is gone. Whether you're playing games, encoding media files, surfing the web, slaving away in a cubicle at work, or mumbling to your computer, running the [email protected] client won't slow you down. To effectively fold, however, you will need an always-on Internet connection. Also, expect to pay a few dollars a month extra on your electric bill as your system's CPU crunches away at full tilt all the time. In my book, those are small sacrifices, especially if you count completing a work unit as your good deed for the day.

So why not head on over to Stanford's [email protected] page (http://folding.stanford.edu/) and download the client?
seem right on par with what Fatguy3 has already told us :)