Best way to take advantage of Hyper Threading in UD?

JackHawksmoor

Senior member
Dec 10, 2000
431
0
0
I tried searching on this topic and couldn't find anything, so...

I just got my first P4 system with hyper threading. I've read that for whatever reason P4s with HT run two distributed computing type clients simultaneously with very minimal performance impact on each other.

Right now I?m running United Devices cancer research thing exclusively on all my systems, and I?m wondering how I could set it up to run two copies at once (if that?s possible). Can I do something as simple as install a second copy in a different directory, and then both will be grabbing different work units and not interfere with each other? (Even if that works, would it confuse the UD servers, when they're sending and recieving work units from "two" computers with the same name?)

Does anyone know if the performance really is close to double (or at least improved) when running two instances of UD? It's kind of counter intuitive, but I know I've read results people have posted (from Seti, I think) where (as an example) maybe running one instance a work unit takes 3 hours, and with two instances a work unit takes 3.5 hours.

As long as I'm asking on here, is there a command line version of UD that runs faster, or a way to queue work units so my 'puter always has something to do?
 

EULA

Senior member
Aug 13, 2004
940
0
0
You could follow these directions for running UD as a service; However, I don't think you'd see any significant performance increase, other than that no one will close it...
 

r0tt3n1

Golden Member
Oct 16, 2001
1,086
0
0
There is no official support for HT or dual processors AFAIK. It is a highly requested feature that `might` make its way into the next version of UD.... There have been a few instances of people using the WinXP Fast user switching feature to run 2 instances if ya really wanna do that..... There is no CLI version, but if ya wanna queue wu's use UDmonitor.
 

CyGoR

Platinum Member
Jun 23, 2001
2,017
0
0
The problem is that you can't run 2 instances because the client won't allow that. You'll get an error which says: "Another instance of the UD Agent is already running". I also read something about using the WinXP Fast User switching feature, but I don't know if it's worth the trouble.
What you could do is run a different project together with the UD agent and assign them to different 'CPU's' in Taskmanager.

Like r0tt3n1 said, UDmonitor is a nice tool to cache wu's with only one slight problem. For each 'slot' in the cache you need to make a new system profile.
Ok, that's only entering a new, non excisting, name for each cache slot. So you could name them like 'P4HT_1, P4HT_2' etc... And when someone will look
at your stats, it will show you've got a lot of device installations

What is your UD user name?
 

JackHawksmoor

Senior member
Dec 10, 2000
431
0
0
I'm tigerwolf with UD. Thanks for the info everyone! If I do decide to run a seperate project, is Folding @ Home worthy too? Will it and UD interfere with each other much on a HT CPU?

I'll probably give that UD Monitor a try.

I used to run Seti, and I'd love to be running it, but I just decieded the cancer research stuff is more important to me.
 

xbassman

Golden Member
Feb 25, 2001
1,243
0
0
Welcome to the team Jack!
Not sure about Folding@Home, but I know of other members doing multiple projects with UD.

If you cache work units with UDMonitor you can keep UD from transmitting results and let it go on to your next cached work unit. (I do this when I am using my computer.) If UD has control when it sends results your points scored per work unit will be higher. One member (that I know of) gives UD only 1 or 2% cpu time but when he sends results he gives it 100%. He can get several thousand points for a work unit that would only give him a few hundred otherwise.

When sending results UD benchmarks your computer, then it looks @ how much time it spent to analyze the work unit. So it thinks you have been working on an extremely hard one and gives points accordingly.

Now, Doing this and only sending in 1 or 2 a week might give a lot of points, but it is still only 1 or 2 completed work units. Not quite as productive as points reflect. That's why I just use it to keep UD from benchmarking my system when I am on it. If you are burning a CD or encoding a video you get raped on points.
 

CyGoR

Platinum Member
Jun 23, 2001
2,017
0
0
xbassman, but when you have a HT pc, you could run 2 processes together while they both get 100% cpu time. Although it would take a little longer for both to complete, it's the best way to use all your processing power.
A normal WU takes about 7hrs, but when you use HT and run another project with it, I doubt it will take longer then 10hrs for each wu.
And when the client benchmarks your system before sending in the result, it also gets 100% of your CPU.

F@H is also a project with a good cause. And from what I've heard, their wu's take looooooong to complete, so that should be good for when you're on dial-up
Good luck!
 

xbassman

Golden Member
Feb 25, 2001
1,243
0
0
Originally posted by: CyGoR
xbassman, but when you have a HT pc, you could run 2 processes together while they both get 100% cpu time. Although it would take a little longer for both to complete, it's the best way to use all your processing power.
A normal WU takes about 7hrs, but when you use HT and run another project with it, I doubt it will take longer then 10hrs for each wu.
And when the client benchmarks your system before sending in the result, it also gets 100% of your CPU.

F@H is also a project with a good cause. And from what I've heard, their wu's take looooooong to complete, so that should be good for when you're on dial-up
Good luck!

Admittedly, I don't know much about HT, (been an AMD fanboy for a while now) so I figured I should look a little more closely at this.

I found this thread where a user installed Microsoft Virtual PC 2004 and did some testing comparing identicle P4's w/HT, one running 1 instance of UD and another running 2 instances of UD. He showed that both machines returned the same number of WU's, but the one running 2 instances received 70% more points for the same number of WU's processed.

The points issue I have known about for a while now (from posts to this board), but it also shows some other things of note. With HT if you are running multiple projects your limiting factor will be program specific as to the resources necessary for each program to complete it's task. (hense the reason multiple instances of UD returned the same number of results as did one instance) If you were to run projects that utilized different resources (is FPU, CPU etc.) you would probably see less of a hit on results returned (per instance).

On the issue of benchmarking, For UD to be @ 100% would depend on priority of projects running.
The UD agent (which does the benchmarking) runs at normal priority. If you are running apps that are set >= normal your benchmark result will suffer. If your other running apps have <normal priority your benchmark will not be affected. This is one reason why in the test (linked to above) the computer running Microsoft Virtual PC was getting 70% more points for the same amount of work. (LigandFit runs @ idle priority while Agent runs @ normal priority)
 

ProviaFan

Lifer
Mar 17, 2001
14,993
1
0
I'm certainly not trying to do any cross-project recruiting, but since you asked about Folding@Home, I'll try to give you a few tips when using the F@H client with another DC client.

Folding@Home can be configured (be sure to say yes to advanced options if you're setting up the CLI) to varying priorities (IDLE or LOW), and certain amounts of CPU usage. This can be helpful if you can't get the two to share cycles equally; if the other project is taking all the cycles, bump F@H from IDLE up to LOW priority, and set CPU usage percentage to 50% (or 30%, or whatever you desire). F@H will take priority, but will take only its share of cycles, leaving the rest for your primary DC client. If F@H takes all of the processor time away from your other project, even on IDLE priority, then leave the priorities alone, but change its CPU percentage to the amount that you want.

Remember, if you want to run F@H for TA, to enter team number 198 during the configuration.
 

BurntKooshie

Diamond Member
Oct 9, 1999
4,204
0
0
I've never tried this out, but if someone else feels like trying, I think it's worth a shot.

Someone said that the client won't let you run two copies of it at the same time. Is this true even if it's in seperate directories, as someone suggested?

If so, I wonder if it would be possible for a second user to run another instance of it. This is similar to the Virtual PC idea, although probably cheaper (because unless you're pirating VPC, it costs money). I also don't know if this part works:

If you have WinXP, you can do a "fast user switch". Then you can log in as a different user, try running the client under that user as well.
 

xbassman

Golden Member
Feb 25, 2001
1,243
0
0
I think I remember seeing on the UD forums that using "fast user switch" the UD client will detect the first instance running and not start the second.

I didn't really dig a bunch but I saw another post that said he got it up and running (2 instances) on a dually,
but I didn't follow that thread. (UD doesn't support SMP)

The download for VPC is free for 45 days though....
 

JackHawksmoor

Senior member
Dec 10, 2000
431
0
0
Okay, I set up UD Monitor (seems to be a GREAT program), and have a cache of 3 work units, which should be enough to keep my system busy if I leave it on without internet access. Technically it bugs me that I occasionally see UD Monitor suck up 1% CPU time. Dang it, that's for UD!

I also set up the command line Folding @ Home program (as a service), using basically the default options (default priority, etc.). Right now both programs show 50% CPU usage. I'm going to keep my eye on both, especially UD, and see if work units seem to be taking about the same length of time.

Oh, and I joined group 198!

Let's see if I melt my Prescott
 

CyGoR

Platinum Member
Jun 23, 2001
2,017
0
0
Great news! Keep us posted about the time it takes for each WU (for UD, don't know anything about F@H, just that's it's a good project )

You don't have any problems getting rid of the huge amount of heat produced by that baby?
My A64 is really nice and cool...
 

xbassman

Golden Member
Feb 25, 2001
1,243
0
0
Originally posted by: CyGoRYou don't have any problems getting rid of the huge amount of heat produced by that baby?
My A64 is really nice and cool...
I remember running UD on a Palomino 2000+ and hitting 72C (socket)
That processor died an early death, but I definately learned a few things.
 

JackHawksmoor

Senior member
Dec 10, 2000
431
0
0
Originally posted by: CyGoR
Great news! Keep us posted about the time it takes for each WU (for UD, don't know anything about F@H, just that's it's a good project )<BR><BR>You don't have any problems getting rid of the huge amount of heat produced by that baby? <BR>My A64 is really nice and cool...

I've never done any scientific testing (ie running the same few different work units under different situations), but UD seems to average a little under 5 hours to around 7.25 hours per WU on this system, although I haven't paid that much attention to it (how much better is the Athlon 64?). It is proportionally faster than my 2.4GHz Northwood though. When I was running Folding @ Home at the same time, it seemed to slow both down 30-40%, but then there's always the chance that was just the work units I happened to get.

I *THINK* there's a net performance benefit running 2 clients, but without anyone doing a scientific test on it to be absolutly sure, I've decided for now to just run one project at a time.

As for the heat, this newest system is actually a Dell, so I didn't have to worry about it at all (first pre-built system I've bought since '97). I'm actually very happy with it. It's non-gimicky, came with good components (ie I was surprised that the hard drive was not only a Seagate, but one with an 8MB cache-exactly what I would have bought). It's very quiet too, and has an absolutly MASSIVE heat sync on the CPU. I've heard some people say it has some form of liquid cooling, and I saw some kind of piping (that might just have been part of the heat sync), but it's covered up and I didn't want to mess with it, so I'm not sure.

At any rate, judging by the heat let out the back of they system, it's by far the hottest running system I've owned. Prescott's heat hasn't been exagerated. The air coming out of my 2.4GHz Northwood I built last year is almost room temp. The air coming out of this new system feels quite a bit hotter-sort of like a room heater on a very low setting. I'm midly worried about it's longevity because of all the heat, and obviously I don't want to pay for the electricity I'm blowing (except in winter, I guess ), but aside from that it's been a really nice system.

It's just too bad Prescott didn't give us the kind of performance going from Willamette > Northwood did. I waited until Prescott got launched before I was going to build a new system. It's got more than double Northwood's transisters, so I was expecting it to smack the Athlon 64 all over the place...it's got to be Intel's most disapointing core since Willamette.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |