TPU's F@H Team

[Alec§taar]

That's a good idea since I'm pretty shure people can find some vulnerabilities to connect with it since it's using the internet. Also, local hacks for it can also f**k you over pretty well xD

It is, for security's sake (another reason I won't run it as a service (besides lack of cpu priority granted & affinity level control) because I can compensate easily enough by running it as a std. app, but knocking out ALL other background processes, including explorer.exe shell)...

I go into HOW this is done & why, here (@ THESE FORUMS IN THE THREAD about optimizing WINDOWS):

http://forums.techpowerup.com/showthread.php?t=11119&page=2



* That explains the "how's" of how to do it, very simple... & it even provides a list of OTHER SERVICES that can have their logon entity lowered & STILL WORK PROPERLY!

APK

P.S.=> I_am_mustang_man, you passed me by in counts finished... GOOD JOB! apk

 

[Jimmy 2004]

The weekly stats update in the first post is now a few weeks out of date...

 

[W2hCYK]

sorry, i was busy yesterday. Its updated now. Sorry for the delays.

I thought i got it done last week, didnt I?

 

[magibeg]

I'm pretty sure you had last week updated because i remember checking my stats on it

 

[W2hCYK]

yea I was sure I did, i even had it in my PB before i changed the image.

 

[Alec§taar]

Guys/fellow teammates for this project, 2 questions!

How many of you run this on MORE than 1 single machine?

(Steevo & Mustang_Man are 2 that I know that do this on more than a single system... anybody else?)



* Thanks for reply!

APK

P.S.=> Also, have ANY of you run this project on a Pentium 4 2.8 - 3.2 ghz range of speed, & how did it perform especially in constrast to AMD boxes of RELATIVELY the same performance range (tough question, this 2nd part)?? apk

 

[Steevo]

I only do it as I trust my builds.



I ordered a Arctic Cooling 7 pro for our server as I do't like 145 degree temps.



So far F@H has cost me over $500 an it continues to climb. But my father almost died of cancer. It is worth it.


For any other member out there that isn't folding, you suck dick. If you have such a badass system use it for more than your own cock trip.

 

[Alec§taar]

It's a good cause, I see you have personal reasons for it... good ones!

(Pretty "passionate" response - but, I can understand it, absolutely! You have personal reasons that are better than my own, this is certain.)

* I also agree that @ this forums, many folks here do have very fast machines, which would be assets to this team & the project itself!

(However, what they choose to do w/ them is up to them... so, I agree on that account with you, but 'oh well' - you can lead a horse to water, but making 'em drink? Another situation, entirely!)

Wish we had someone with a Pentium 4 of speed/mhz rating between 2.8 - 3.2 (preferably the latter) who could tell me how CPU's of that nature perform on this project... especially vs. AMD units of "like" rating... so I would have a 'gauge' of somekind, before I "kick-in" my 2nd system here onto this project!

Heck - I have to, if I am going to 'break into' the top 2 spots (don't think I can take on a farm like you have going w/ just 2 systems here Steevo, but I will try!)

APK

P.S.=> However, I do think of it as a system benchmark as well, a long-term one... & if you think about it? It is... sort of! Not as much as I could view SETI@Home though, because the units sizes are a LOT more "variable" here/not as consistent in size.

I also "get a kick" out of how I_am_mustang_man & I are "trading places" @ #4-#5 spots for a week solid now:

http://fah-web.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=50711


It's a race! apk

 

[Steevo]

Average output on Intel VS AMD for this project, just by what i have seen and done here.



Intel Prescott 3.0Ghz 1Gb of RAM VS AMD 4000+ @ 2.88 1Gb of RAM

Just by experiance the Intel will outperform the AMD by 2X.

I have both these systems running. And have had them running for awhile.

 

[Alec§taar]

Intel Prescott 3.0Ghz 1Gb of RAM VS AMD 4000+ @ 2.88 1Gb of RAM

Thanks, I needed that type of info.!

(& it appears to be the SAME as for SETI@Home - w/ Intel rigs outperforming AMD ones (strangely imo, because on SETI? The built-in benchmark shows AMD superiority, by FAR, where it counts due to the datatype being processed (mainly f-point, not int))).

It's the compilers used, & efficiency tricks in them, favoring Intel - has to be.

Just by experiance the Intel will outperform the AMD by 2X.

Boggles the mind, honestly, due to what I stated above... again, as I have said here I don't know HOW many times - I wish they'd do a customized recompile for AMD cpu's on the FAH_Corexx.exe to favor it.

(This does help, large, on SETI @ least - the Intel results (inferior CPU's imo to AMD ones) show it)

Anyhow - it's a "dragrace" to the top & it is a competition for spots 4 & 5 all week now between I_am_mustang_man & myself (but, he's got more than 1 system kicking this constantly)... So, I have to startup my other rig & get it 'in on the action' here, & it's Intel based!

(Ought to aid me in doing better still, based on your reply & its info.!)

APK

 

[Steevo]

It is the double precision SSE2 handling that favors the Intel. It isn't that AMD handles it wrong, but from my understanding Intel SSE2 specificly offers a higher degree of precison on SSE2 FPU.

http://en.wikipedia.org/wiki/SSE2

See the area on SSE2 VS X87 Double Precision


I believe this is also the reason a AMD will generate visual artifacts instead of a lockup when too high of a overclock or oo low of voltage is used for the core. The margin of error is wider on the AMD side.

 

[Alec§taar]

It is the double precision SSE2 handling that favors the Intel. It isn't that AMD handles it wrong, but from my understanding Intel SSE2 specificly offers a higher degree of precison on SSE2 FPU.

http://en.wikipedia.org/wiki/SSE2

See the area on SSE2 VS X87 Double Precision


I believe this is also the reason a AMD will generate visual artifacts instead of a lockup when too high of a overclock or oo low of voltage is used for the core. The margin of error is wider on the AMD side.

Don't AMD cpu's also possess SSE/SSE2/SSE3 (also 3dNow)...?

APK

 

[Steevo]

After a means yes.

http://en.wikipedia.org/wiki/IEEE_floating-point_standard

80 bit precision.

So a SSE2 instruction set ran on a AMD or Intel that has no defined endpoint (IE resuts pending calculation) will have a average of results that are different.

 

[gygabite]

Ok, adding a 2nd system for F@H: AthlonXP @133MHzx10.5 with pc133 256mbsdram and gf-fx5200. I found that components on a dump and put them together, so i paid $0

 

[Lt_JWS]

Just installed F@H on my rig... very new to this so i hope i did it right lol

 

[FLY3R]

Just installed F@H on my rig... very new to this so i hope i did it right lol

Sweet, the more the marryer!!

 

[KennyT772]

hey i got two pcs folding myself!

 

[Alec§taar]

hey i got two pcs folding myself!

Thanks for joining KennyT772, & I'll tell you WHY, specifically:

You're running a Pentium 4 3.2ghz CPU according to your profile!

(One I am guessing that is using H/T enabled on it as well)...

I have been VERY curious to see someone here running a system of that CPU makeup & mhz/ghz ranges as well, because I have one here like that (& am curious on HOW it may, or may not, affect my "score" part on our team charts).

Steevo runs some of them, & says Intel rigs DOUBLE the output of AMD units here typically, from the same generation of release @ least... I tend to believe him, as I saw the same on SETI@Home (even though, imo, AMD cpu's are superior in many regards/respects, @ least the last 1-2 years now).

You'll be my "living proof example" to examine now, so thanks again for joining up & powering up on a P4 unit & in the exact ghz range I was waiting to see also!

APK

 

[CjStaal]

how do I set it up to not be a anonymmous donor since I am probally going to have 2 pentium 3's running it as well as a AMD 64 athlon rig and I have another of those spare mobo's for thre athlon so if I get another cheap athlon proccessor I can use that, and I have a pentium 2 latptop and a sempron laptop.... I also will be running multiple instances on the conroe once I get it.... I have a question though...
Since it uses the lowest priority can I have it run multiple instances on 1 core?

 

[Steevo]

I have over 14Ghz of power in my living room I brought home tonight.

 

[CjStaal]

nice, steevo do you have aim?

 

[CjStaal]

Anyone who has some money to throw around, you should start farming also
http://www.newegg.com/Product/Product.asp?Item=N82E16813128292 <-intigtrated video $43.99
http://www.newegg.com/Product/Product.asp?Item=N82E16819104249 <-processor_____$60.00
http://www.newegg.com/Product/Product.asp?Item=N82E16817154008 <-200watt PSU___$21.00
total: $124.99 for a decent single core system (AMD)
http://www.newegg.com/Product/Product.asp?Item=N82E16819112207 <-Processor Celeron D $52.00
http://www.newegg.com/Product/Product.asp?Item=N82E16813185067 <-Mobo w/onboard vid $41.99
http://www.newegg.com/Product/Product.asp?Item=N82E16817154008 <-200watt PSU_______$21.00
total: $114.99 for a decent single core system (Intel) and I think this is better for folding.

 

[Steevo]

I refuse anythig that has AOL in it, on it or close by, as a fact I will disown this post.

 

[Steevo]

Now the downside.

They are all being sold on Saturday. 9 machines, wipe and re-install before friday.




Mebey I will leave F@H as a service on them...... bastards give me no time. HARRR!!!!!!

 

[Alec§taar]

I refuse anythig that has AOL in it, on it or close by, as a fact I will disown this post.

I just don't use it - I knew it'd have a lot of "haywire madness" coming into & out of it like bugs/nasties/virus/trojans etc. - et al.

(I was into the IRC world before there was an "AIM" or any other messaging system, & saw it all degenerate into stuff like that, & even worse (because IP determination's REALLY easy on IRC too)).



* Cannot have that here... it's counter-productive lunacy - & would create work I do not need to be doing (like rebuilding a system setup again)!

APK