|
|||||||
![]() |
|
|
LinkBack | Thread Tools | Search this Thread | Rate Thread | Display Modes |
|
|
#1 |
|
Member (8 bit)
Join Date: Aug 2005
Posts: 143
|
Dual Core?
Is there any way to take advantage of a dual core with F@H? Right now it only shows the cpu usage at 50% when no other programs are running.
|
|
|
|
|
|
#2 |
|
Computing Professor
Staff
Premium Member
Join Date: Jun 2001
Posts: 11,718
|
Over at Anandtech : http://forums.anandtech.com/messagev...&enterthread=y
01/20/2006 09:37 AM User is offline Print this message Author Icon kb3edk Senior Member Posts: 366 Joined: 07/11/2004 X2 FTW! I liked my 4400+ so much I built another rig around a 3800+ last week. BOINC is an especially great client for dual core DC work... unlike some other DC clients, BOINC auto-detects 2 CPUs without you ever having to configure it or run a separate instance. Good luck building your next cruncher! ------------------------- -Adam in Philly You should PM Pilot in Command after you check out his sticky at the top of this forum.
__________________
Asus M4A77D, 64 X2 6000+, 4 GB Corsair DDR2 800 ram, Radeon 5770. |
|
|
|
|
|
#3 |
|
Member (12 bit)
|
Leo, for F@H you may run two instances of the text-only client with no problems. You can install them as a service, if you like.
You just need to do a couple of things: 1. Configure your first client. If you're running text-only console right now, you need to add a flag to the target line. Right-click on the short-cut, click Properties, and in the Target line, add a space and then "-config" (without the quotes). When you run it again, it will allow you to configure. In the Advanced Settings, be sure its Machine ID is set to one. 2. You can then remove the -config flag, but then add the flag "-local" to the end of the Target line. 3. Install a second copy of F@H in a separate directory. It will configure automatically the first time you run it. It must have Machine ID = 2. In its Target line, be sure to add the -local flag at the end. Now when run, they should automatically be assigned priority to each processor. Happy Folding!
__________________
Kov Are You Foldin'? Join PCMech's Folding@Home Team and Help Save Lives! Click Here!
|
|
|
|
|
|
#4 |
|
Member (8 bit)
Join Date: Aug 2005
Posts: 143
|
Does BOINC support Folding at Home?
What exactly is a service? |
|
|
|
|
|
#5 | |
|
Member (9 bit)
Join Date: Jul 2002
Location: Goodyear, AZ
Posts: 490
|
There is no Folding client/core for BOINC yet. Stanford University is working on a client/core for the BOINC platform. That's all I know.
By the way, if you are interested in BOINC DC, check out the BOINC teams from PCMech. There are similar DC projects like Folding (i.e. Rosetta@Home and SIMAP) About your "service" question. I got some info from here: Quote:
|
|
|
|
|
|
|
#6 |
|
Member (12 bit)
|
There has been some talk of F@H being offered as a BOINC project, but it will continue to remain a stand-alone project whether that happens or not. I don't know that F@H joining BOINC is a real priority, from what I've read.
When you go through the options on installing the F@H text-only console, it will ask if you want to install as a service. Basically, it runs in the background without a taskbar icon or program box visible, and it will run on a machine regardless of which user is logged in (or even if no one has logged in yet). |
|
|
|
|
|
#7 |
|
Member (8 bit)
Join Date: Aug 2005
Posts: 143
|
Alright, I installed two versions of the text-only version on "Machines" 1 and 2 as services, cpu usage is now at 100%.
![]() if I have both usernames set to the same thing, will both instances of the program add to my points? |
|
|
|
|
|
#8 | ||
|
Foldin' For PCMech!
|
yup, I have a few labs at my old high school folding under my name and they all add points for me. You can have as many systems as you want and up to 8 clients on a single machine. The 8 client limit is to be removed with the new version of folding, whenever it comes out.
![]() What is your system specs anyway? Intel or AMD? EDIT: If ya have an Intel with 1GB or more RAM, you can run QMDs and get ~650ppd on that single machine! This will depend on the speed on the cpu tho. Fast = more ppd. Over at ocforums they got 900+ ppd out of an intel 920! Crazy stuff!!
__________________
Eric
Last edited by enhanced08; 01-26-2006 at 03:30 PM. |
||
|
|
|
|
|
#9 |
|
Member (12 bit)
|
To have QMDs assigned to your client, you need to put Yes when config asks whether you want Large Work Units. They can be 5MB uploads, so you'd only do that if you have a broadband connection. You also only want to enable that on one client per machine, as it would bog down with two.
Thanks for contributing! |
|
|
|
|
|
#10 | |
|
Foldin' For PCMech!
|
Quote:
1) you need large work units AND the -advmethods flag to get QMDs ![]() 2) with a dual core system you can run 2 QMDs without the system bogging down if you have over 1GB of RAM, if not then only run QMDs on 1 client. You should be good to go now! enjoy your points! ![]() EDIT: depending on the setup and usage you might notice a little, best thing to do is to try it! heh, we dont even know if you system can run QMDs or not! lol Last edited by enhanced08; 01-26-2006 at 08:16 PM. |
|
|
|
|
|
|
#11 |
|
Member (12 bit)
|
What's the title of this thread? Dual core...
Yes, what he said.
|
|
|
|
|
|
#12 |
|
Member (8 bit)
Join Date: Aug 2005
Posts: 143
|
Thanks for telling me how to configure two of those things... it's cool knowing all that power isn't goin to waste!
I have 2 Gigs of Ram and an AMD X2 4400+. Yes, I do have a broadband connection but I share it with my parents. What the heck is a QMD? I'm guessing that PPD stands for points per day. |
|
|
|
|
|
#13 |
|
Foldin' For PCMech!
|
yes, ppd stands for points per day. Stanford says that the average is about 40 ppd per ghz. So a 2Ghz system should get about 80ppd per day, so Stanford says. A QMD is a special work unit (WU) that is designed for Intel systems. A 3Ghz Intel can get about 400ppd if it is running QMDs, they offer a huge point increase, if you system can run them.
requirements for a QMD are: Intel + 512mb + -advmethods flag + BigWU AMD systems can't get QMDs because the compiler they used to make them in an Intel compiler which is designed to "slow down" if it detects an AMD system, making it look like Intel is much faster. AMD wont give stanford permission to use their compiler to create "AMD QMDs" the last I heard.
|
|
|
|
![]() |
| Bookmarks |
| Thread Tools | Search this Thread |
| Display Modes | Rate This Thread |
|
|