OFFICIAL: SlickDeals.net Folding@Home Team Getting Started, Discussion, Bragging, and Community Support Thread
 

moved to wiki
mod: please remove this post

This post can be edited by users like you :)
 

This guide is designed to make getting started easy and convenient for novice computer users running Windows that want to participate for Team Slickdeals. Participation is optional, any information listed here could be wrong, install the suggested software at your own risk, enjoy!

Introduction:
Folding@Home (F@H) is a way to participate in a distributed computing project done by Stanford Univeristy to research the way human proteins assemble (fold) themselves. This research will assist biologists to greater understand the causes of disease such as Alzheimer's Disease, Cancer, Huntington's Disease, Osteogenesis Imperfecta, Parkinson's Disease, Ribosome & antibiotics, etc. The purpose of participating is to donate your otherwise idle computer power to assist modern medical science in finding cures to these diseases.
More info: http://folding.stanford.edu/

How it works:
You install a small software program on your computer which downloads data, processes it, and send the results back to Stanford. Ideally this program should be running whenever your comptuer is on, and ideally running on computers that are left on all the time. The software program is designed to work when your compuer is idle by setting the process priority to low. This basically means that the F@H program pauses when applications you are using are busy, and it resumes when the applications you are using are idle. You should not see any performance loss when the program is running. There should be no exceptions to this, even if you are running processor intensive applications like 3D games or CAD, but you can easily pause the F@H program temporarily if for some strange reason you find it interferring with whatever application you are using. The research project is open to anyone with an internet connection and a Windows, Mac, or Linux computer. There are also optional "teams" which you can join to give the participation an added sense of accomplishment and competition.
More info: http://vspx27.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=49191

About the software:
There are two versions of F@H, the "graphical" version and the "console" version. Technically these descriptions refer to the configuration portion of the software, once you configure the software you no longer do anything with it except let it run and do its work in the background. There are many pros and cons for either version:

Graphical version:

• configuration is fast and easy through a typical window-style user interface.
• gives you instant access to pause, resume, or quiting the software.
• gives you instant access to its status, current activity, team and user stats.
• gives you an animated visual representation of the current work being done.
• takes up slightly more resources because its system tray application must always be running, which can take anywhere from 900kb to 12mb of RAM.
• some of the advanced options are not in the user interface, such as "MachineID" so you can run it on multiple computers, but this option is available through the INI file.

Console version:

• all configuration is done through the command prompt, like DOS.
• uses Windows' services control panel, or you can create a shortcut, to pause, resume, or quit the software.
• has an option to install as a service, so it starts automatically before you log into Windows, and there are other benefits of running as a service which advanced users may want to use.
• all status and activity are saved to log files.
• takes up slightly less resources because it doesn't use any graphical interface.
• every option in the software, including all the advanced options, are available through the command line interface.

[B]If you want to use the graphical version for Windows, follow the steps immediately below.

For the console version for Windows, skip down to the next section.

Graphical version Installation and Configuration for Novice Users:
http://s90802609.onlinehome.us/fah_w1.png [onlinehome.us]
1) Go to http://folding.stanford.edu/
2) Click on Downloads.

http://s90802609.onlinehome.us/fah_w2.png [onlinehome.us]
3) Click the download link for the Windows 2000/XP/Vista Graphical Client (or the client that corresponds with your correct operating system).
4) Follow your internet browsers prompts to save the file to your computer. I suggest you save the file to a separate folder, perhaps a new folder named "FoldingAtHome" inside your "My Documents" folder.

;) I always recommend manually scanning downloaded files for viruses before running them. Most antivirus software offer the option to scan specific files in the right-click menu. Right-click the downloaded file and select scan for viruses (your antivirus software may not have this feature).

5) If the file you downloaded is a ZIP file:
(the beta client is typically a ZIP file), then you will need to unzip the file: if you have Windows XP/Vista you can right-click the zip file and select "Extract All" then follow the on-screen prompts to extract the file to the current location.

http://s90802609.onlinehome.us/desk2.png [onlinehome.us]
Locate and double-click the EXE file. The beta client will probably run from its current location without going through any installation. The non-beta client will probably go through an installation, follow the on-screen prompts to complete the installation. You can typically use all the default options during the installation.

;) if you are prompted by your software firewall to grant access (or unblock) the F@H client, select the option that allows the program to access the internet.

Configuration:
http://s90802609.onlinehome.us/desk.png
6) You should now see a new icon in the system tray.
7) If the configuration window does not automatically open, then right-click on this icon and select "Configure".

http://s90802609.onlinehome.us/fah1.png
8) Select the "User" tab.
9) Enter a desired username where it says "user name". You might as well use your Slickdeals username.
10) Enter 49191 where it says "team number".
11) Click Ok.

http://s90802609.onlinehome.us/fah.png
12) Double-click the F@H icon in the system tray.
13) You will now see the main window and the program doing its work.
14) You can close this window by clicking the "X" in the upper-right corner, the program will continue running in the background.
15) If you need to force the program to quit for whatever reason, you can do so by right-clicking the F@H icon in the system tray and select "Quit".
16) The program will automatically start when you turn your computer on, but it will not show the main window, it will only display the F@H icon in the system tray.

;) if the F@H system tray icon does not appear when you turn your computer on (make sure you click the little arrow to show all the icons), then the program is not running, and you will need to manually add a shortcut to your startup folder. This will most likely be the case if you downloaded the beta graphical client, which does not appear to go through an installation. This will work for Windows 2000/XP, not sure about Vista: open "My Computer" and browse to the EXE file, right-click the EXE file, select "Create Shortcut", right-click over the shortcut, select Cut. Go to Start, Programs, right-click over the "Startup" folder, select Paste.

17) After a few weeks of participating, you should see your stats listed on the F@H Team Slickdeals website:
http://vspx27.stanford.edu/cgi-bi...mnum=49191

Remember, the purpose of this software is to let it run silently in the background. You don't necessarily do anything with it.

If you just followed the directions above you can stop now you are done. This next portion is for the console text-only version, not the graphical version.

Console version Installation and Configuration for Advanced (Windows) Users:

Most of the console prompts offer a pre-determined default value which you can select by just pressing the Enter Key. I cannot verify that these default values will remain the same for each operating system or version of the service, so I provided a recommended value for each setting. The values I recommended may not work best for everyone, but they should work well enough to get anyone up and running quickly and easily with a minimal amount of problems.

1) Create a new folder for the application. This is where the program will run and register itself as a service, and where its log files will be stored. This must be a permanent folder for the program. The program will register itself as a service the first time you launch it. Moving the EXE file to another folder or renaming or otherwise changing the program path can cause problems.

2) Go to http://folding.stanford.edu/

3) Click on Downloads.

4) Download the appropriate console version for your operating system, save the file to the new folder you created.

5) Run the program.

6) "User name" = enter a desired name, you might as well use your Slickdeals username.

7) "Team number" = 49191.

8) "Launch automatically at machine startup, install this as a service" = yes, unless you are re-configuring the software then this question changes to "Do not this program automatically, remove the service?" and you should enter no. When first configuring the software, entering no would be absolutely idiotic, and would defeat the entire purpose of using the console version. It would be better to install the Graphical version than to install the console version without the service install option.

9) "Ask before fetching/sending work" = no.

10) "Use Internet Explorer Settings" = no. Make sure you select no, selecting yes may prevent the F@H application from accessing the Internet. When you select no, you may be prompted if you want to use a proxy, or to enter proxy settings.

11) "Allow receipt of work assignments and return of work results greater than 5mb in size" = yes, unless you are using dial-up or for some reason dont want to download large assignmnets then enter no.

12) "Change advanced options" = yes, unless this is the only computer you are installing F@H on and you always leave your computer on then enter no and skip to item #23.

13) "Core Priority" = idle.

14) "CPU usage requested" = 50. I recommend starting with 50 until you can determine that there is no significant change in CPU temp. If everything runs smoothly, reconfigure the service at a later time (instructions are at the end of this post) and enter 100.

15) "Disable highly optimized assembly code" = no.

16) "Pause if battery power is being used" = no, unless you are using a laptop then enter yes.

17) "Interval, in minutes, between checkpoints" = 15.

18) "Memory, in MB, to indicate" = press the Enter key to use the default (max) setting.

19) "Request work units without deadlines" = no, unless you turn your computer off frequently then enter yes. Keep in mind if you enter yes, that sometimes there are no work units without deadlines meaning that the service wont do anything until some are made available. If no work units are available, the logs/console will report errors when attempting to fetch work units.

20) "Set -advmethods flag always, requesting new advanced scientific cores and/or work units if available" = no.

21) "Ignore any deadline information" = no.

22) "Machine ID" = 1, unless this is an additional installation of F@H then enter a unique number for each computer. For example, on the first computer enter 1, on the second computer enter 2, on the third 3, etc.

23) The installation is now complete and the console client and service is running. At this time you will have a console window open showing some status and logging information. Do not close the F@H console window now, otherwise the service will stop. Leave this program running until it displays "Completed 0 out of XXXXXXXXX", then close the console window.

24) To easily start and stop the service, create shortcuts with the following targets. The format is "net.exe start " followed by the service name surrounded with quotes. You can find the service name by going to start, run, services.msc. Its basically "FAH@" followed by the path to exe with backslashes replaced with plus symbols. Don't forget to replace C:+Program Files+FAH+FAH504-Console.exe in the examples below with the F@H service name on your computer. Don't forget to start the service after you have closed the console window!

• Here is an example shortcut target to start the service:
  net.exe start "FAH@C:+Program Files+FAH+FAH504-Console.exe"
• Here is an example shortcut target to stop the service:
  net.exe stop "FAH@C:+Program Files+FAH+FAH504-Console.exe"

25) To easily re-configure the software, create a normal shortcut to the exe and add " -configonly" (without the quotes) to the end of the target. Stop the service, run the config shortcut, configure as desired, close the window, then start the service.

Other commands are available which you can also create shortcuts for. Here is a list of all the console commands:
http://folding.stanford.edu/conso...guide.html



Feel free to ask questions on how to setup/configure the program in this thread.

Lots of new users need help figuring out which version of the program to install.
The original guides (written by darksarcasm in 2006) are still valid for those with SINGLE CORE PROCESSORS. If you have a dual or quad core processor, your computer can probably handle "more difficult" work units and provide a larger contribution (and earn more points) by installing the SMP version of F@H.

Here is a basic summary of all of the program options:
Version 5.03 - graphical install (easier for novice users) for single core processors
Version 5.04 - console install (answer approx 20 yes/no questions to configure the program, see post #2 below) for single core processors
SMP 5.91 beta 6 - this is the typical version for a dual-core or quad core computer, console based install, can now be setup as a service fairly easy (automatically runs when you turn on the computer, no open window)
The three versions above cover the vast majority of users, and there are also the various clients for linux, mac, GPU (hgih end ATI graphics cards) and PS3 (scores a lot of points).

If you dont know which version(console or graphical) to install, install the console. it is much more light weight(will effect your computer less) and runs in the backround seemlessly


How to show off your folding score in your slickdeals signature:
You may now show off your folding score in your signature, or in the forum posts by enclosing the user name in the [fah] tags, for example:
yields




If you have special characters in your username, you can use the [fahuid] bbcode. You'll need to find your userid # over at EOC.

For Dr. Who, it would be



For detailed info on team or individual stats go to:
http://folding.extremeoverclocking.com/

Helpful Links:
foldingforum.org
fahinfo.org
fahwiki.net
Points per Work Unit
Useful install guide [xtremesystems.org]that has worked well for those who want to use the SMP (multiple CPU) client

During setup, answer no to the question "Use Internet Explorer Settings"
This often causes a problem submitting completed work units. In repeated threads on foldingforum.org they tell you to change that to "NO"

For those of you REALLY serious about folding, here is a basic setup you can get for cheap:
CPU: Intel Core 2 Quad Q6600 boxed $199.99 at microcenter
Motherboard: Asus P5B, supports the new Penryn 45nm too! but don't count too much on overclocking. $60
Memory: 256MB is enough, I bet many of you have tons of 1GB left around due to 4GB deals lately.
USB: 128MB USB drive is enough. You don't need a harddisk!!
Total is $260 if you have memory and USB key laying around, which most of us do. Add in $40 for the power supply.
There, $300 and a very powerful crunching machine.

For a bit more money, here's an updated system that will net many more points a day with -advmethods and/or -bigadv
I'm assuming if you know how to overclock, you know how to pick out your own stuff, so this is for a NON overclock build which is cheaper and doesn't require a dedicated video card, since sandy's have one built in.
Case.. Antec 300 or Cooler Master Centurian 5.. around 50 bucks
CPU: Intel Core i7 2500 ($199 average price) or 2600 ($299 average price, 8 threads lets you fold "bigadv" units)
heatsink.. an aftermarket heatsink is highly recommended but not required.... Cooler Master Hyper 212 Plus ... 25 bucks avg price
Motherboard: ASUS P8H67-M LE ... $99 ... there are cheaper ones but sometimes it's worth 20 bucks for good reviews and better build quality
Memory: GSKILL RIPJAWS DDR3 memory can be found for 50 bucks for 4 gigs, or 100 bucks for 8 gigs. RAM is cheap these days. I recommend 1.5 volt or less for the ram's volt rating.
hard drive: well pick up whatever SATA drive you want, they are in all sizes and price ranges.
Power supply: Antec Earthwatts 650 watt power supply... 65 bucks

Total price 400-600 range depending on your CPU and hard drive choices, less if you reuse stuff or get stuff from relatives. This system will, at stock speeds, blow any other processor away, at their respective stock speeds. Remember to do the latest a3 units, or -bigadv units if you have a 2600. There are links above that tell you want you need to do, or use this link below this paragraph. Overclocking is so easy your grandma could do it, and will net significantly better results as it's easy to take these right up to 4.3 GHZ with a bit of extra cooling, but it will cost more too. PM me (dissident) if you want me to post a list here of overclock friendly components to supplement this list which is strictly for non overclockers.

http://www.hardwarecanucks.com/fo...-home.html

Here's some information on -advmethods.. basically with a good quad core processor and the SMP client (for multicore processors) and the right settings you can fold the newer projects and you can get bonus points (if you use a passkey and fold 10 bonus qualifying units first, then you start getting points) and with the system I suggested above net a good 15000 points a day or more, to above 30,000 if you are using an overclocked system with 8 threads (2600k processor)

http://www.stanford.edu/group/pan...tings.html

*added addendum for an alternative Folding setup: Purchase a Playstation 3 and youll get a free blu-ray player and a gaming system for $349 or less.


Reported EUE's (Early Unit End):
2665 (R2, C188, G16) Client-core communications error: ERROR 0x7b - 61%
2665 (R2, C306, G14) Client-core communications error: ERROR 0x7b - 1%
2665 (R2, C707, G20) Client-core communications error: ERROR 0x7b - 35%
2665 (R2, C882, G13) Client-core communications error: ERROR 0x7b - 23%
2665 (R3, C646, G15) Client-core communications error: ERROR 0x7b - 7%
2665 (R3, C985, G13) Client-core communications error: ERROR 0x7b - 3%
3062 (R2, C42, G78) Client-core communications error: ERROR 0x7b - Warning: long 1-4 interactions
3065 (R4, C59, G45) Client-core communications error: ERROR 0x7b - 27%

2665 (Run 2, Clone 335, Gen 10) Client-core communications error: ERROR 0x7b - 0% (savethelinoleum)


Flag Options

moved to wiki
mod: please remove this post

console > graphical

<edit>
bean's tear, blood and possibly pain free guide to update your GPU2 and 5.91 beta client.

These are the clients I will be using:

Windows XP/2003 GPU Console client (6.20) [stanford.edu]
Windows XP/2003/Vista/2008 SMP client console version 6.22 beta2 for MPICH (32-bit or 64-bit) [stanford.edu]

1. Stop the current client you are updating and backup everything first!
2. Stop the MIPCH2 Service (services.msc) for SMP as well.
3. Install the new program from the above repository into the folder with your work.
3a. I put the GPU client in the Application Data\Folding@home-gpu
4. I've renamed my clients 'fah.exe' and 'fahGPU.exe' just for easier CLI tasking, optional.
5. Run each client using the -configonly switch.
6. Setup each client how you want but make sure to use the following additional parameter is necessary.
6a. SMP: -local -smp -forceasm -verbosity 9
6b. GPU: -verbosity 9
7. Change each service like the following replacing 'Administrator' with the login information you used for installation:
http://img.photobucket.com/albums/v2...3/LogOnTab.jpg
8. Check your services to make sure they all checkout and try them out.

Credit for full service install (5.91) is here [xtremesystems.org]
</edit>

Console version Installation and Configuration for Intel Mac Users:

Most of the console prompts offer a pre-determined default value which you can select by just pressing the Enter Key. I cannot verify that these default values will remain the same for each operating system or version of the service, so I provided a recommended value for each setting. The values I recommended may not work best for everyone, but they should work well enough to get anyone up and running quickly and easily with a minimal amount of problems.

1) Open a new terminal window.

2) Copy/paste the following into the terminal, and wait for the program to download. If the program does not automatically launch after its downloaded and unzipped, press Enter.

3) "User name" = enter a desired name, you might as well use your Slickdeals username.

4) "Team number" = 49191.

5) "Passkey" = press the Enter key to leave blank.

6) "Ask before fetching/sending work" = no.

7) "Use Proxy" = no.

8) "Acceptable size of work assignment and work result packets" = normal.

9) "Change advanced options" = yes.

10) "Core Priority" = idle.

11) "Disable highly optimized assembly code" = no.

12) "Interval, in minutes, between checkpoints" = 15.

13) "Memory, in MB, to indicate" = press the Enter key to use the default (max) setting.

14) "Set -advmethods flag always, requesting new advanced scientific cores and/or work units if available" = no.

15) "Ignore any deadline information" = no.

16) "Machine ID" = 1, unless this is an additional installation of F@H then enter a unique number for each computer. For example, on the first computer enter 1, on the second computer enter 2, on the third 3, etc.

17) The installation is now complete and the console application is running. At this time you will have a terminal window open showing some status and logging information. Do not close the F@H terminal window otherwise the application will stop. Leave the terminal application running all the times. To get it out of the way, just right-click over the Terminal icon in the dock and select Hide.

18) To automatically start the application:
a) Go to System Preferences, Accounts, select your user account, select Login Items.
b) Drag-drop the StartFaH file located in ~/Library/Folding@home.
c) Select the Hide checkbox if you do not want the terminal window visible.

You can stop now you are done!

19) To reconfigure the software:
a) Shutdown the Folding application by closing the terminal window.
b) Open a new Terminal window.
c) Copy/paste the following into the terminal window and press Enter:

d) You will now be prompted to go through the configuration.
e) Don't forget to restart the Folding application once you have finished the configuration: copy/paste the following into the terminal window and press Enter:
20) Optional: you can download FahMon for Mac here: http://fahmon.net/download.html. This application will show statistics of your progress, current work assignment, deadline, PPD, etc.

CUDA 181.20 is out.
Direct Link XP [nvidia.com]

Core 1.19 Thread Manual Download [foldingforum.org]
Direct Download [stanford.edu]

Inspiring everyone or demolishing? :lmao: He had another 15k PPD jump yesterday.

for quadro nvidia cards right?

Is there a difference between the CUDA drivers and regular drivers? Nvidia standard drivers are at 181.22 now.

Yes, the folding bug got me and I am currently running five 9800gx2 in two builds plus some minor extra contributions. I was hoping it could inspire at least the previous big folders to keep up and someone else having spare computing power to add to the production. :)

I have some hardware issues to fight with so my PPD will probably drop some.

What rigs are you running? I am impressed with all the 1920-pointer mr slickdeals keep submitting.

Now, go team slickdeals!

My current hardware is:

Folder 1:
AMD BE-2400
MSI Crossfire Motherboard ($28 :clap: )
2 x eVGA 8800gs 384mb

Folder 2:
Intel e7200
Abit IP-35E
1 x XFX 8800gs 384mb

Folder 3: (at mom's house)
Intel Q6600
Asus G31 Motherboard
1 x eVGA 9600gso 384mb

Main Rig:
Intel Q6600
Asus P5B Deluxe
PNY 9800gtx

Folder in progress:
Intel E6600
MSI P6N Diamond
1 x PNY 9600gso 768mb
1 x eVGA 8800gs 384mb

Will probably remove the 2x8800gs from folder 1 and put them in the P6N Diamond for now (if I ever get around to it), turn off folder 2 and put it's card in folder 1. Then I've gotta finish a new computer to replace my main rig, and once that's done I can put the 9800gtx from my main rig in folder 1... man, this gets complicated. I run SMP clients on the Q6600 at my mom's house, and also on the E6600 that is in progress. Don't know why, but I can't get the SMP client to run on the e7200 build - it just crashes, so I gave up. Running SMPs on the BE-2400 was too slow (risk missing deadlines) and the PPD per watt numbers weren't favorable.

I've been getting a lot of EUEs on the GPU clients lately, on all of my systems and none of them are overclocked - so I think something is up with the newer WUs. :dontknow: I have to babysit them a little and restart the clients to get them running again (otherwise they stop for 24 hours). It's very frustrating to be adding GPUs to the folding effort but still have my PPD drop because of unstable work units.

I just use my regular desktop, with an eVGA 9600 GSO 384 mb. When I wake up in the mornings I turn it on to use it, and leave it running until I goto sleep. (And close F@H whenever I want to play a game or something ;) )

dempa's push is inspiring!

i am in the same boat with raven with when i am folding. i also turn off the gpu client when i am gaming but leave the cpu's on even thru gaming. i find the gpu client not being able to throttle back as well as the cpus.

um, since i recruited two friends to not only join in the project but also team SD could i tally up their PPD to mine? :D

that would make about 4000-5000 ppd total!

not :(

I am running a q6600 and an 8800gt. I fold in two week spurts, about once every 2 months.

i was @ 12 or 15k total ppd.... i let all the clients go till they crashed or eue....i'm left with one linux client that runs perfectly... they all started to consume too much of my time (and money) to keep them running, if it were more stable i'd be more apt to get my ppd back up.

How much were they costing you to run? Were they taking that much power? (Haha, and now that I think about it, I have 5 roommates that get to split any cost of my folding)

I agree, it is amazing how stable Linux SMP is, where the other clients just never seem to get ready for prime time. When I had a dedicated linux folder, I don't think I ever had to touch the thing - literally would go for months at a time until I installed a patch or update that required a re-boot. Would love to see a Linux GPU client - Dempa posted a link a few days back to a tutorial that would allow you to run GPU on Linux via WINE, but native GPU on Linux would be awesome.

It's best not to calculate how much the folding adds to the electricity bill. :D

For me, two dedicated computers right now draw 300w loaded... so 300w x 24h x 30d = 216kwh. Multiply by your average $/kwh. I don't like to do that last step because it's depressing - but it'll be at least $40 a month, my rates are tiered so my first kwh costs 11c, but quickly goes to 22c, 33c, 37c depending on how much total power I use for the month.

I'll have to see about getting 5 roommates to split the folding bill.... Not sure the wife would go for that. ;)

I tried a couple weeks back, but couldn't get the CUDA drivers installed. The NVIDIA drivers for linux don't come with it pre-installed.


Ouch. That is pretty steep. And tell her it's about sacrifice for a good cause. Because for all she knows, one of those 5 might have another folding box to contribute.(I know it wouldn't work....I'm not looking forward to my folding future when I tie the knot....although her desktop does have a free PCI-E slot....)

This is what drives me nuts about the current crop of unstable GPU work units and all the Early Unit Ends I've been getting. With three GPUs running (I think my mom turned her computer off), WUs take about 3 hours, so I should be posting 2 or 3 WUs per update: Point updates are newest to oldest:

Time......Points....WUs
6pm.......0............0 (all three still dead)
3pm.......0............0 (all three died)
12pm.....864.........2 (one died)
9am.......1,533......3 (restarted all before went to work)
6am.......511.........1 (this morning, all clients dead with EUEs)
3am.......1,217......3
12am......895........2
9pm.......1,533......3 (last night, all running)

At least with the SMP client, if it got a NAN or EUE, it would just restart on a new unit. The dang GPU2 client decides to wait 24 hours before downloading a new work unit. Which is why my PPD is so freakin' erratic lately.

I just don't know what to do with the thing. The build with 2x8800gs has a 650w Antec Trio power supply and only draws 180w at the wall socket, so I can't believe the EUE's have anything to do with power fluctuations. And the build with 1x8800gs has a 500w earthwatts and draws 120w, so no issue there. Thought maybe the GPUs were dying, but I get the same frequency of EUEs on my 9800gtx and a brand new 9600gso - so the only way I'm buying hardware failure is if I have 5 defective GPUs.

Anyone have any suggestions? Configuration tips, additional parameters to add to the command line like "-stability=on" of "-don't $*#(ing crash!!"

I was getting EUE's every little bit with my 8600GT with the old 178 drivers, with the beta ones it wouldn't even run. Haven't had a problem yet with the new drivers yet though. You might also try the new nvidia core 1.19 which is supposed to help with EUE's if you have a problem, I put it up on my earlier post.

Thanks - is the nvida core 1.19 a re-install of the GPU client (uninstall, get the new 6.23 version from the downloads [stanford.edu] page), or just delete the old core in the work directory and let it download a new core?

I have deleted all work files and the fahcore_11.exe (or whatever the filename is) to let it download the latest core, but I don't recall what version it is. I did install the latest drivers (jan 20 version), but maybe I need to uninstall and re-install the latest GPU client. :dontknow:

edit: Never mind - found your previous post.

SMP 6.23 Expired February 2nd.
http://foldingforum.org/viewtopic...=46&t=6642
New Executable Direct Link [stanford.edu]

Thanks! I was actually scratching my head trying to configure another machine for SMPs - I'd install it, and then try to run the executable - and it would do nothing at all. No error, no warning, no message, just nothing. I'll have to try the new one next time around.

i am not sure if this was posted earlier or if anyone else also knows about this statistics site but i recently found out about so i thought i would share it: http://folding.extremeoverclockin...s=&t=49191

i know about the other one, http://www.xcpus.com/folding/fold...191&page=1

cheers.

I've been using the first one since I've been introduced to folding. I think the site may be in the first thread as well...

just double checking, if i install the F@H client onto another computer i do not need to change the machine ID number, right? i can leave it as machine ID #1 if it is the first and only insistence on that computer?

That is correct. Each computer is independent of the other.

over the past two days ive installed the client on other computers but on F@H's site it reports the number of active processors within the last 7 days as being the same.

odd.

no worries. it's not always accurate, and additionally, it may take a while to update correctly.

ty to the both of you!

ok something is happening with my f@h and it is becoming annoying. for the past 3 days on my computer ive not been hitting my ppd avg. barely 1WU in 24hours.

skimming through the console i see that i am getting "Got status 503" in trying to return the finished work load. oddly enough it is able to DL a new workload when it has not finishied uploading the previous one. it states that it could not transmit unit 05 to collection server and it is keeping it in queue .

what could be the problem with uploading? does this mean i have 4 other units in queue waiting to be uploaded? will it eventually upload it at all? is there any commands to force it to try again?

it is one thing to not get points but another to have 3 processors working for the past few days and none of the work to be returned ... feels deflating.

is it my internet connection? ive havent changed anything. i do use opendns though.



p/s.: idk if this helps but the memory usage for all three clients have blown up to 200k per each f@H within the task manager which is something ive never seen the f@h taking up so much. and their folders in c:/ drive is a combined 80mbs of hard drive space now o_0

for f@h application i should only overclock my gpu with the highest memory OC possible and forget about the core clock for faster folding?

would i be able to achieve a higher memory OC if i were to restore my core clock speed?

i have a 8600gt running at 815/855 (680/800 default) right now.

Hi Guys,
Joined folding a good while back but gave up until i built my new rig, anyway im now running a Q9950 with a GIGABYTE GA-EP45-UD3R Intel P45 board and a Asus Radeon HD4850 512MB with 4G of DDR2 800. Anyway when i was folding back in the day i was able to run the CPU's at 100% when the PC was idle, now with this new software Version 6.23 i can only get one of the cores to run at around 70% and the rest are all lower then that, is it the software that is resticting the CPU's?
Thanks
Paul

Forgot to add, im running vista ultimate 64x. i haven't VMware setup on it yet

I can't speak to the CPU clients but are you running a GPU client, you'll probably get better points out of CPU/GPU combo than just CPU???

i am uncertain myself but its true what the other poster said that if you were to run the gpu client when you can itll return greater points. in fact i think your gpu would net more than four insistence of the cpu client. or a combination of the SMP clients.

Hey PhucilliJerry, how did you setup the signature

You just put

[xxx]your user name[/xxx]

then replace "xxx" with "fah".

Sweet thanks

Paul

Are you running the client with the -smp flag?

my console and client cpu are both working on workloads that have taken two days and at this rate would probably take another two more days to finish. usually at most it would be a day... i wonder if its an error. and in the console the timestamp of it writing local files is off.

should i restart them?

i think the only difference couldve of been the fact that i set both of them to work on workloads greater than 10mbs where as before only one of them was set as so. could that have slowed it down?


p/s. is running one smp on a duo core cpu better than two consoles in terms of ppd ? would there be a noticeable lag? with two consoles i havent experienced any lag in daily computering.

Yes, in most cases SMP produces greater PPD than two CPU clients on a dual core.

I recommend you to install FahMon so you can better track the size and speed of the work units which sould answer your question. Maybe you received 905 pointers which should take "days" on your duo core.

fahmon? done and done! thank you :)

btw, fahmon reports that my duo core has been working on the two work loads for the past 3 days and will be done in about 1.5 days and that they are each worth 2400 points!
not bad for nearly 5 days of work and no ppds returns from my cpu :D

i hope theyll give me another chunky work load after these two sets. usually the ppd for my cpu is around 200 ppd per core. 2400 points for 4-5 days is around 450-500ppds!!!


i am a breaking a sweat just thinking of how big this work load is. 5 days to finish?! i would like to think that f@h now trusts me with more demanding and bigger workloads due to donating for a while and returning WU's within a "timely" fashion. that and i broke 30k in points. i wonder what f@h will give me to work on when i get past 50k points? WU's that would take my cpu a whole month to crunch?!

i cant wait!

the WU's 'difficulty' remains the same regardless of how reputable a folder you are :P

also, is EOC's server working today? i've only got 1K logged for some reason...

ah well it was a fleeting idea :).
the servers due seemed swamped. it took a lot of attempts for a finished WU to be uploaded and to dl a new one.

question! i have deleted the previous two consoles i was running and dled and installed the SMP verison.

now that i have it up and running (with -smp -deino -verbosity 9 set in config) it doesnt seem to take advantage of both of my duo cores. only one core is being pushed to the max. any ideas behind this? i though SMP console was suppose to use both cores...

It should use both cores but not 100% unless you use the affinity changer, anyone have a link to that? I don't do SMP in windows anymore. Linux is much more efficient and stable

Affinity Changer
http://distributed.org.ua/forum/i...topic=1149

folding kings amongst men! ty.

Dempa, your PPD is sick, 61st out of ALL folders.
Very cool seeing a 'Slickdeals' member listed in the top 100.

i would like to know as well. i started using riva tuner and i hear that the higher the shader setting the better for F@H?

also what program does everyone use to benchmark the stability of an GPU OC? before i was using ati tool but i cant have both ati tool and rivatuner running. and ati tool doesnt allow unlinked OCing of the core and shader.
and any recommendations on a gpu monitor that can sit within my tray that is not ati tool?

i currently have two OC-ed profile speeds and i wonder which would be best for f@H?:
1) 680 core/ 1960 shader/ 800 (1600) ram
2) 820 core/ 1640 shader/ 855 (1710) ram

although again i need to bench it to make sure its stable.

bleh, sorry again, but i dled the affinity changer and installed it. at first i thought it worked and i left my computer but now looking at it , both cores are not being pushed.

heck one core is not even maxed all the way. perhaps 90% all the time.

any other ideas? i would like to run the SMP version at full force.

any information i am omitting ?


EDIT UPDATE: i was able to resolve it. i just needed to restart it for it to verify the flags. sorry about the numerous questions. a bit green to this. but i am happy that now i am folding with full force.


p/s.: still wondering about the optimal settings for gpu though :)

I just setup F@H on one of my PCs (with Machine ID = 1) and then on my PS3... I changed the user id and team to match my account here, but it never asked which Machine ID I wanted on the PS3--and its nowhere in the settings. Am I screwing myself over on points here? Do I need to change my PC machine id to 2?

Don't worry, you're fine. MachineID is only computer specific to be able to distinguish between multiple clients on the very same machine.

Thanks, was going to edit my post--found this explanation on the first page which I somehow missed...

Got both GPU and SMP running, and made sure to set the other id to 16 :)

Individual
 

Hi,

My experience from OC'ing GPU's is that it is quite individual what you will be able to use on your particular setup. E.g. the same 9600gso card works at very high OC settings in computer A running Intel q6600 and Dell 350W PSU. The very same card can not run with same settings in computer B running AMD X2 6000+ and cheap 500W PSU. I needed to run with considerable lower OC not to have constant EUE = "Early unit end" in F@H.

I've noticed the same for other cards/systems. Since I don't have time for any gaming or such, I let F@H be the benchmark for how high clocks the gpu can take. As long as F@H doesn't produce EUE's, I'm good.

Using Rivatuner, for folding generally I'd recommend to not touch core and memory clock and concentrate at cranking up the shader speed as much as possible. This makes by far the biggest difference in PPD. However, in systems with single GPU setup, after doing the shader I'd start with trying with the core clock as well. Can give a tad additional boost. Memory clock is the most sensitive to EUE's and also gives least effect on PPD.

When running single 9800gx2, two gpu's in one card I was running up to around core 700, shader 1836. Additionally I noticed you can not always run same speed on both gpu's on the same card. In multi-9800gx2 setup, I leave core at default 600 and shader maybe at 1730-1782. The biggest challenge with multi-GPU is heat. I've had 9800gx2 running stable at 90C+ for days.

Real shader clock speed used by NVidia go in 54MHz steps, e.g.:
[...]
1404
1458
1512
1566
1620
1674
1728
1782
1836
1890
1944
1998
2052
2106
2160
[...]

Setting to 1900 will mean 1890 etc.

Another question here---I'm trying to grab a WU for Windows SMP 5.91 beta client (for MPICH), and based on this server status list [stanford.edu] my 171.64.65.65 is currently in standby mode. Is there any way to switch the server I'm linked with? I'm seeing people complaining elsewhere that sometimes they end up waiting more than a day at a time while these servers are down.

If anyone knows how to change which server my SMP client is associated with, it would be much obliged

***EDIT*** The server went back up--but I'd still like to know if its possible to manually switch your server if necessary! I can see that being a problem (especially for uploading WUs in the future) if a server is down for a prolonged period of time.

Good info Dempa. On the EUEs, do you get any with your setups? I started getting EUEs constantly with some of the newer work units, so my quad 8800gso setup will sometimes go a day without an EUE, and other days I will have all 4 GPUs down with too many EUEs in a day. Currently, I am not overclocking anything since I thought that may have been the problem - but I still get the EUEs and their frequency doesn't seem to have gone down with the reduced shader clocks. It's frustrating having to babysit the machine and restart clients all the time.

My PSU should be good - 750w PCP&C with 60amps on the 12v rail :D Maybe I should try reducing the memory clocks, since that won't effect ppd. It seems in reading around the forums, a lot of people are having trouble with EUEs, so I don't think it's necessarily hardware. :dontknow:

ty you once again, dempa. ive seem to have hit 1900 for the shader and 820mhz for core. i have lowered the ram to default as per your advice. i produced an EUE at 1970 and perhaps i can find a sweet spot around 1950's or so but it is a bit too time consuming to keep watch of it and to produce another EUE imo.

i hope these couple of OC mhz's will help in folding. i think ive noticed half an hour taken off? of course this varies a lot. i want to say an hour with some WU's but perhaps that is too much. my 8600gt default is 680/1440/800 (now 820/1900/800).


p/s.: there is no actual way to watch for EUE's other than checking the times and see if it actually ended early, huh. it doesnt seem like the EUE's produce any points at alll even though i thought it would.

My EUEs usually only occur right when the WU is starting, so I don't worry about losing points because unit crashed, but I get really mad when the client stops because I've exceeded the EUEs allowed and it pauses for 24 hours.

I check for EUEs by watching FAHmon and reading the log file. :dontknow:

So, I decided to try the system tray client v.6.23 on my netbook. Unfortunately, this stable edition appears to be buggier than the Betas! As soon as I configure and restart the F@H.exe, the process ends up dying within 10 seconds! Restarts do squat, and the only way to get it working again is to nuke all of the folder files, repair, and try again from scratch.

Is anyone else dealing with this? The group even admits in the faq that the viewer doesn't work either. Best to avoid the sys tray version entirely, even on my core solo machines?

when i first set out to fold i did not think i would even break near team rank 100 but i have!

i am listed on the first page of any rank sites with rankings 1 thru 100 as their first page :D

of course this is after a lot of newbie-ish questions regarding gpu, smp, and etc :)

but thank you to everyone within this forum and to slickdeals.net as a whole.


fold on!

p/s.:i am also on the top 20 list of producers! albeit rank #20~ i would like to thank all the 720p and 1080p season bittorrent bundles that i have been dling for the last 7days that have forced me to leave my computer 24/7 on and to my friends that i have hanged out with causing me to not game or watch 1080p which in turn makes me pause or throttle down my gpu folding.

question: which desktop video card is equivalent to the ps3's?

will there be a difference in what kind of video card to compare to the ps3 in terms of it being a video card overall and in terms of its folding ability? that it may be a higher end card or lower end if we were to compare it in terms of its folding ability or as an overall video card?

The PS3 runs about 1000ppd, which is somewhere between an 8500gt and 8600gt. Those are lower end cards which definitely don't compare to the PS3s gaming capabilities. Overall, the PS3 isn't a very efficient folder, especially when you consider power usage.

odd i thought it was. the more you know!

Just wanted to drop in and introduce myself to the rest of the SlickDeals folders. I just recently started folding@home after learning about the Stanford project which I think is a great cause. With all the great deals I have found through SlickDeals, it was an obvious choice for me to join this team. Hopefully I can help a little and may have some questions in the future as far as tweaking some of my rigs for maximum gain. My love of working with computer hardware and building new machines combined with folding to develop a greater understanding of diseases sounds like a new hobby right up my alley!

Welcome to the folding team! Looks like you'll be in the top 100 of us in a week with the current PPD you seem to produce. What equipment are you starting on?

Cheers

welcome TheEvil! i am glad you decided to not lurk and say hello :)
i was wondering who you were moving up the ranks so fast. thank you for not only joining our ranks but also helping out with the folding cause.

cheers.

Thanks for the warm welcome everyone.


I have a few different setups running at the moment. My largest PPD producer would be my 2008 Server (x64). It has an Intel Q9550, 8GB PC2 8500 RAM, Geforce GTX 260, on a Gigabyte GA-EP45-UD3P motherboard. Main OS sits on a VelociRaptor 300GB 10k RPM hard drive, with 3 other Western Digital 1TB Caviar Black hard drives in a RAID 5 array. I'm just about finished setting it up, but it's meant as a test bed, and run some virtual machines in the near future.

The other PCs I have folding are using the uni processor client and include my HTPC running on an Athlon 64 X2 5000 Black, my "main" machine on a Intel Core 2 Duo E6320, oh and my old server running on a Xeon that I'm about to decommission.

I work in IT and have a lot of spare parts just lying around, so I wouldn't put it past me to put together a few Frankenstein PCs to do some extra folding. We will see... :)

Welcome to the team!

The PS3 is a VERY efficient folder. Stanford has simply set the work unit value at 300 points each for 8 hours of folding. The PS3 does combination GPU/CPU folding, and pound for pound its 8/9-cell processor blows just about everything out of the water.

Concerning power requirements, the newer PS3 architecture (the last year or two) has only required 115w, versus the original 200w. You're looking at about $0.15-$0.20 to power your PS3 for a full day, which is less than the $0.40 or so it takes to power your high-end PC under high load. If it was on a linear scale, the PS3 would end up netting you 10,000+ PPD. Not inefficient, very big difference. If you're using one of the original PS3s (20/60gb), then yes, it would be an inefficient use of power.

That certainly makes sense, it seems to me like the hardware in the PS3 should be capable of folding very well. But bottom line, my PS3 was drawing about 130w and putting out 1000ppd and my PC with an 8800gt draws 200w and puts up around 4500ppd. So from that perspective, it certainly seemed inefficient.

ooo a shiny new thread :)

Hi guys... i'm having a little bit of trouble here... seems like my machine is stuck trying to send results for the last two days. any help would be appreciated. I am running -Folding@Home Client Version 6.23 - I have quit and restarted about 5 times already and it keeps giving me the same thing. I'm using a dell P4 1 gb of ram and radeon 9550. its and old set up, but it still contributes :crazylaf:

TIA

Disable "use proxy server" would be my only guess? Or maybe it's just the FAH server that you are trying to send the data to is down - look in your log file and it should tell you the IP address of the server, then you can check if it's up or down here [stanford.edu]. It happens. :dontknow:

I have a funny one - didn't realize one of my VGA dummy plugs came unplugged, so Windows 7 lost the video card so the GPU client kept crashing - but every time it crashed, it would send a "partially complete" work unit back to the server. I guess 0 = partial. Check out how many WUs I reported at 6am! :yahoo:

Time....Points....WUs
12pm....864.......2
..9am ...1,858....5
..6am....384.......51 :woot: :woot:

current SD team rank: 80

we have moved up!

well we have been moving up but it seems like we will continue to :)

good work everyone!!

Welcome TheEvi1 :yahoo:

Welcome also to all the other new folders that I may have missed while I was 'away' (kept my machines going though).
It's nice to see DempaSD still kicking incredible amounts of a$$.

Fold on!!

Nice going to all of you! Based on my future production estimates it will be around mid July before I hit 1million. I did break into the top ten producers (based on 24 hour average) this week though. :)

ok this is not F@H related but perhaps someone out there who has knowledge on this can help chime in.

earlier today i noticed my partitioned C-drive was missing 3-4gbs and that is my windows xp pro OS drive and i dont save anything on there. poking around i do not notice anything new that could take up 4gbs. i did have a computer freeze up before i noticed this.

so i defrag and clean out any temp files and i am still missing 4gbs. i go into safe mode and i notice in safe mofe i have my 4gbs back. at first i thought it was the paging file and i set it to ZERO and then defrag. and then i set my paging file back but when i load into windows normally i am missing 4gbs again.

any clues fellow folders?


p/s.: under normal windows boot ... in the paging file setting area it reports my c-drive with the 4gbs there but in My computer window the 4gb is not anymore.

Your hard drive could be going bad, have you ran any diagnostics on it? Good program to find where your hard drive space is being used is TreeSize Free [jam-software.com].

GPU update heads up:
GeForce Release 182
Version: 182.06 WHQL
Release Date: February 18, 2009

interesting read on why drivers are huge in file size: http://www.anandtech.com/weblog/s...aspx?i=571

everything checks out to be green and normal on all scans regarding HD health.

Been folding at home for a few months now. I'm not making the kind of progress you guys are, but I don't think it's too bad considering I'm using a 3 year old laptop(Lenovo T60) and the $350 Dell E521 I picked up nearly 3 years ago as well. Thinking about getting a Radeon HD 4830 and running the GPU client, but money is kind of tight.

I used to fold many years ago, but with all the support and the friendly competition here, it has motivated me to be a little more dilligent when it comes to folding.

David

hey david! top 200 rank is notthing to sneeze at and any folding helps, period. not only for points but mainly for helping disease research :)

i cant talk for everyone but i am glad there is a fellow slickdeal member that is giving it their all.

p/s.: the 4830 is a nice gpu for folding but just remember to make sure that your dell mobo can take the connection. i think 3 years ago the slot was still AGP. if it is any form of pci-e (pci-e 1.0 vs 2.0 doesnt matter) then itll be good. not to encourage anything out of your budget but the prices are pretty cheap now. i am waiting for a 8800 or 9800 to drop around 50 to throw to the folding cause.

Thanks, I've been checking Tom's Hardware's monthly Best Video Card for the Money and they say the 4830 is around $90. I'll need to check what kind of connection, but a lot of cards would be off limits due to PSU limitations and slot limitations. I'll definitely look into it.

There have been deals recently on the HD4830 for around $70 after rebate, but holy cow the HD4870s have been down below $150 so prices just keep dropping. Most of my 8800gs/9600gso cards I have managed to get for about $35 to $45 AR - great folding cards. I don't have any ATI cards to compare to, but I think with the new GPU work units the ATI cards are holding their own PPD-wise. Initially, when the GPU2 client first came out, nVidia was much faster for folding@home - but I think they are very close now.

35 to 45?! give me a link or a PM if you see another one. that is CHEAP!

I will post here if I see one, but they usually make the front page if they get that low.

This was the Amazon deal for $36 AR. I just got my rebate check yesterday. :D

How many PPD do you get with the 9600 GSO? Around 4,500?

Used to get that much. Now, it varies by a bit depending on what work unit you get. I get anywhere between 2500 to 4100 ppd - probably averages a little less than 3000 ppd overall. This is for the "old" 9600gso that has 96 stream processors. There are newer 9600gso's (because nVidia likes to confuse us) that only have 48 shaders - and they probably do about half the ppd.

Quick question on Nvidia drivers. Do you suggest using the newest 182.06 drivers or the 177.35\174.55 drivers as per the Folding@home FAQ?

always the latest--since its the greatest.

does an OC-ed CPU affect the GPU?

because the last few days i had to keep lowering my GPU OC and now its at default but even then from time to time i still receive an "UNSTABLE MACHINE" error in the log and the gpu client goes and starts over in a sense.

my cpu cilent has yet to ever encounter this.

Yes, i had issues with my 9800GTX and my 8800GTS when i had my Q6600 at 3.0Ghz. Didn't matter what i did, i got the unstable machine. I have overclocked my 9800GTX mildly with no errors, but if i went a little too high, i got the same error. My 4870 Does not get the same errors, the display driver simply crashed if it was too high.



Josh

thank you for the insight on that. bleh, i would default my cpu if this machine was a purely folding machine seeing as how the GPU cilent has better ppd (more points , better folder and etc), but this is also my daily computer so no deal.

ill restore my GPU OC settings for folding and just take the unstable errors.

Thanks, that's what I figured.

I might be back to folding... hehehe. Kinda surprised I still rank...

FYI

PNY is having a Folding@home contest at XLR8Gaming. PNY Rules\Prizes [xlr8gaming.net]

Contest Rules:
Contest Prizes:
While I highly doubt I could make it into the top leader boards, every folder has a chance at the random draws. So go download the client and make sure you join the Slickdeals folding team!

Thank you TheEvi1 for sharing this! Best of luck to everyone with the SD Folding team in winning a prize :)

p/s.: i except to see dempa within the top 10 ;)
p/s.: here is the link in getting your passkey: http://fah-web.stanford.edu/cgi-b...passkey.py which is something i was folding without .

:celerylift: Congratulations to Dempa for taking over the number 4 spot on Team Slickdeals (and passing me :crying:) :celery: :celery:

Signed up. Definitely won't make it in the top, but hopefully I can win one of the random prizes. It's really nice how the Folding on here has taken off. I remember that the team was at 95 when I started, and we're at 79 currently.

is it just me or is the PNY contest website a bit on the flimsy side?

when i log in to see the random winners it seems like it wants me to re-register my folding info.

or when i go to "my account" it doesnt seem to be there.

edit: congrats to TEAM SD for reaching rank 79!

Yeah they have definitely had some hiccups getting this thing up and running smoothly. I think a lot of it has to do with the passkeys and making sure the data collection is working properly with the appropriate user name\passkey. If you go to your account page and look under contest registration and see your Folding@home user name across from Folding, then you should be set. Where you are seeing the ability to re-register is just where they left the ability to change your Folding info in case you messed up or something.


On a side note, does anyone have any suggestions for a multi-gpu capable motherboard for Folding? I'm currently looking at getting an MSI K9A2 Platinum [newegg.com] open box from Newegg on the cheap.

Nice, I think I'll sign up. THanks

I bought 2 of those, open box and have three gfx cards on each running 24/7. Good value for folding, no doubt.

how does this one look: http://slickdeals.net/permadeal/1...rboard-160 ?

seems like an alternative imo since you also receive a solid cpu for 60bucks more or so.



EDIT: its sold out.

on another note, anyone else find it at odds that they are giving out higher end GPU cards to top folders who obviously have... higher end GPUS since they are able to fold so well?
ugh, stuff like this in life that makes me shake my head. it should just be random and the end.

Hi Guys,
Looking for some help in setting up a few VM ware PC, what the best free software your there, i have an x64 OS and am looking at getting more ram for my rig.

Anyway any help would be great
Thanks
Paul

Check this [planetamd64.com] out.

Wow, it looks like Stanford is putting out some big proteins for us to fold with our GPUs.

I just got project 5902 which will take 9.5 hours to finish on my 8800GT and is worth 1680 pts.

Gimme some of those! I'm still working on a bunch of 384 and 511 pointers. :vomit:

Welcome TheEvi1 to Bold!!!

Anyone keeping a rough track of about how much it costs to Fold on a 24/7 bases?

Using this site i have a system that draws 288watt (100% peak loads selected): http://www.extreme.outervision.com/PSUEngine

Just rough estimates using the monthly electric bill i believe there is a $15-$20 increase per month since ive started folding. I know there are varying factors and that looking at a households bill to gauge is not accurate but i do believe just by guessing that i am not off by much.

My avg PPD varies from 1000 to 2000 and this is due to when i have work that must be done and as such the resources for Folding is taken away.


My Folding Production by Months:
Month Points WUs
03.09 14,222 31
02.09 35,801 104
01.09 14,198 66
12.08 9,878 38

So about $70 donated in the last four months to produce 74,099 Folding points. Eh, i am ok with that.

At first i expected a $5-$10 increase with Folding 24/7. Before folding i was basically running my PC 24/7 with dling and uling. So it would seem that an OC cpu and gpu running at max does make a difference in electricity drawn.

Of course i could be wrong about how much it is costing me to Fold. I am interested in hearing what anyone else thinks or know what it is costing them if they dont mind sharing. I dont mean to bring this up as to deter Folding.

I do believe that during summer i will only Fold when i am on the computer and during sleeping hours as to avoid peak hours (i am out during the day to noon) as much as possible :)

p/s.: idk how they would do it but it would be neat if there was a stats ranking that lists ppl based on electricity paid per point :D

Get a Kill-A-Watt if you really want to know. That power supply estimator is usually quite high.

My rig (E6600 with an 8800GT) pulls 200w when running the GPU client, so about $15/mo. But I leave my computer running 24/7 anyway and it idles at about 150w, so realistically it's only costing me about $5/mo MORE to Fold.

I'd been getting the 420 point 5900 WUs, but looks like I've got one of the 1680s right now. About 4000ppd for either WU, so that's a bump up from those 384/511 pointers that only gave me 3500ppd.

Either the new projects or the new FahCore (mine just downloaded 1.24) are doing something different. I'm still running the old nVidia drivers that would keep one CPU core maxed because the new drivers don't seem to like my machine and I haven't felt like messing with it yet. But... it's not maxing one core anymore, just spiking it every few seconds. Seems like they must have optimized something.

perhaps i am off with the PSU estimator and the monthly bill. according to this article, which is a bit dated but should be still good to use, for folding it seems like there is a 30W to 70W increase from idle.

http://www.silentpcreview.com/art...page4.html


p/s.: electricity billing also varies for ppl to ppl :D but a 30w to 70w increase is not really double the watt drawn so i cant explain the nearly $20 gap jump in price.

Thanks! Off on vacation this week so not browsing the forums as much, but as you can see I can't stay completely away. Perhaps I have a problem. :rolleyes:

Are you on Win Vista or 7 and using NVidia GPU?
 

Try the old beta drivers 185.20 for PPD boost on the new big WU's:

http://foldingforum.org/viewtopic...=52&t=9075

Let us know how well it works for you!

Cheers

I will have to try that when I get home tonight. I don't remember what drivers I'm on now. Seems like whatever driver I have, these new WUs don't get my cards nearly as hot - they are running around 40-45c now, wherease with the "old" WUs they ran closer to 70c most of the time. So maybe the GPUs aren't getting fully utilized and the different drivers help.

I will say that the new work units have had zero EUE/NAN errors. Haven't had to restart a client for about two weeks. Very stable now, so I don't have to babysit the things. :thumbsup:

Wow, ridiculous gain here!

9800GTX on Server 2k8 x64
My 5903 WU went from 5hr 50min remaining (5800ppd) to 3hr 15min remaining (10kppd)!!!

Whoa, after giving that a read, I may have to try those drivers since I'm on Vistax64.

On the opposite side of the fence, any input on optimizing PPD on SMP folding? I took a hiatus to build a new system (Phenom 9850 + M3N-HT Deluxe + dual 9600 GSOs), and I've mostly just been folding to slowly break in my OC 9850. Right now I'm tinkering with SMALL files (project 4463) and I'm pulling an average of 250 PPD at 100% core use!! I should be pulling in 1500-1900 PPD @ 2.9ghz... As long as I'm tweaking I don't care, but I'm obviously doing something wrong here--any suggestions?

Are you sure you're using the SMP client? Project 4463 isn't an SMP project, it's a regular gromacs. Do you have the SMP client but aren't using the SMP flag??? You should be getting a lot more ppd out of that processor on SMP, probably 2 instances would be your best bet for just processor folding, but if you fold on the 9600 GSOs, doing 2 GPU instances and 1-2 SMP instance(s) might be best. Anyone else?

185.20 on Vista
 

I only have one box running Vista with one 8800gs GPU. Performance increase on the 59xx (1680p and 420p) WU's around 75%!

I agree 2xSMP on Quad core is probably the optimal together with 2xNVidia. I might try a very similar setup (Phenom II 940 + 2x8800gs) on Vista in a week or so.

Double check start command for the smp client.

Checked my client file and I'm using the -smp flag... I'm also running the same 6.23 beta for MPICH (x64) SMP client version that I was before the rebuild. But yeah... it's definitely showing up as Gromacs in Fahmon. Not a darn clue what I'm doing wrong with this...

Unless I'm reading that wrong, I had no idea that multiple instances of the same clients could be assigned to specific cores or cards... I thought the only way to see results from two cards in one machine was to simply operate in hybrid or SLI mode. I guess I've got some googling ahead of me!

***EDIT*** I think I follow what you're getting at--I simply copied my gpu folding folder and set it up with another machine id when running... or is there a better method?

Okay, after testing this with repeated reinstallations and copy/pastes, I've discovered that I seem to alternate between receiving .a0/82/78 cores (gromacs33/Amber/SMP respectively). I could receive an amber work pile, delete the client/work folder, and then receive an SMP work core... what the heck? Is it supposed to alternate? Is there a guaranteed way to make sure that I will ONLY download SMP Gromacs? If I have to continually check my folding, I'll just turn the darn thing off, because at 220ppd it's just not worth my electricity!

Does anyone know how to consistently get A2 core WUs on a Windows box? I'm on Vista x64 and I sometimes get them when I manually setup an SMP client, but more often than not I end up with a 78/82/A1...

The A2 is greatly preferred because each instance only takes up 25% of my processor capacity and still gives huge PPD--everything else gives less PPD and just plain bogs down the CPU at 100%. Apparently Linux and OSX are the operating systems that get A2, but I'm definitely getting them from time to time...

***********************************************************************
Okay, about the only solution I can find (without simply going to Linux) is to run a couple VMWARE sessions on my Vista box. I'll only be able to handle two of them, but that should still increase my PPD by about 2k...
Instructional on setting up VMWARE to regularly get A2 core WUs [bjorn3d.com]

Bought the $100AR GTS250 and replaced card in the Vista-box = very close to 11k PPD with slight OC for the shaders.

About the SMP-issue.. I could not get it to work until I googled some extra and found good guide suitable for my setup. I am running 32-bit OS's all over the place so DEINO setup it is. Ran VMware/Linux on laptop for a while until it was burnt out :rolleyes:. Will follow example above and run it on desktops when I find time.

GPU clients are to be prioritized first though.

I wish I woulda bought one. $100 for the card, ebay the game for $20, sell my 8800gt for $50, would have be a nice gain for $30.

Edit: Meh, not that big of a gain.

I installed the 185.20 drivers on my Windows 7 64-bit box and saw about a 30% increase in PPD (3000 before, 4000 ppd now) for my 8800gs cards. :thumbup: Installed the 185.20 drivers on my Vista 32-bit build with a 9800gtx and went from 6000 ppd to 11,000 ppd. :eek::omg::drool:

Now I just gotta replace some 8800gs cards with gts250/9800gtx cards. :woot:

It's not THAT good--it comes to 2x ppd over the 8800, and it's still technically twice as expensive. Assuming you've got the extra machines, it's not necessarily a bad thing to simply spread the cards out and get 5k ppd on each---things do overheat and explode, and I'd feel a lot better about it if it was the cheaper card. Yeah yeah, warranty and all that---but they're a lot like rebates in my mind.

Nice results with the 185.20 drivers. I am going to have to try it out myself later tonight. Has anyone seen a big difference on the XP side of things?

Also have a question on multi-gpu setups. I have two 8800 GS cards that I have gotten off eBay on the cheap. It was my intention to get another two 8800 GS's, but I'm having a hard time finding any more. I was thinking about completing the 4xGPU setup with two 8800 GT's, but wanted some confirmation that there wouldn't be any problems with having two different types of cards. From what I have read though, as long as they use the same drivers, everything should be fine. If anyone could put this to rest for me, I would appreciate it.

Gaming wise I won't see any difference with what I am playing. It would be nice I cant justify it just yet tho. see how long it takes a card comparable to the gtx 290 to get around $200.

It sounds like the drivers really only help for 64bit OSes. My results on Server 2008 x64 were terrific. My 8800GT on WinXP 32bit in my home machine showed no gains whatsoever.

Unless you're actually trying to SLI them, it doesn't make any difference.

Agreement--I tried a dual card config with hybrid SLI (in Vista, no SLI Bridge), and the gains through folding were basically zero.

And I don't know about you guys, but I have two systems cranking with dedicated cards, and since last night my average PPD has gone back down to its pre-beta levels... can someone confirm that they're still getting the 70% PPD boost from using the Beta drivers?

My 9800GTX is now working on a 5773 WU worth 768 points, showing 5100ppd. Maybe they pulled the 590x WUs. :(

I'm on 5775 worth 768 points, showing 3400ppd on each 9600 GSO... so it sounds like we're getting similar results today--I was still getting 5-6k ppd as of about 5am this morning. I can't seem to get any larger WUs to check against. Anyone else? I'm not seeing any complaints on the other folding boards

***EDIT***
Confirmed, there was a fix put out today. We're back to regular PPD now...
Since the "fix" was released on Tuesday morning, I have witnessed MAJOR INSTABILITY PROBLEMS with my GPU clients operating with the Beta drivers. My WUs almost always fail at some point and restart at 0% again. I have since switched back to the standard manufacturer drivers and my clients are back to operating without a hitch. I highly recommend you transition away from the Beta drivers--reinstalling without uninstalling the Beta cleared up all issues for me.

Just a quick comment on this: With my Phenom 9850, the optimal configuration is one SMP instance--if I attempt two, it simply eats from the efficiency of my GPU client, and the overall PPD lowers by about 5-10%.

This all goes out the window with A2 core WUs though... when I can get those, the optimal configuration is GPU + 4x SMP clients, and I end up capping out at over 14,000 PPD.

i wonder if the SMP is slowing down my gpu folding.

Wow, that sounds very interesting! My big Q before going VMware is how to manage system prio for VMware so GPU client performance is not affected. I mean, if running Win only, as long as I run SMP clients with lower prio ('lowest possible') than GPU ('sligthly higher'), SMP clients have not affected GPU clients performance.

EDIT: Setting VMware prio: http://www.overclock.net/5691927-post809.html !! Will try when I finally get around to installing SMP under VM...

That's a good question--I've already downloaded all the software and I'm going to give VMWARE/Linux a go this weekend. It really depends how much the two instances of vm/linux eat up... because when I run each SMP client with A2 clients on Vista, it only adds 25% to my cpu load (versus 100% for any other SMP core). I can't imagine that dual instances of VMWARE could possibly add 50%, so I expect to see at least a 2k ppd increase over my current configuration. From what I'm reading, VMWARE Server 2.0 may NOT work with Vista x64---here's hoping it works, because I'm not switching back to XP x64---spending my weekend reinstalling software is NOT my idea of fun.

An edit to my previous post--what I meant before was that my OVERALL ppd value dropped 5-10%... Actually, looking at it now, my GPU efficiency dropped 30%, the first SMP dropped 20%, and the second SMP client was at about 60% of optimal. Adding any SMP clients above the 100% use mark simply causes all values to degrade. It's amazing how efficient those new A2 cores are.

Considering your processor, it SHOULDN'T be, if you're only running one SMP client.

If you've got FAHMON [fahmon.net] installed, it'd be really easy to find out for certain. Simply run the GPU client for about 30 minutes, check PPD, then begin running your SMP client. Give it a good 2-3 hours (not just 30 minutes, it should go about 20-30% for the value to really average out) and then check your GPU ppd value. My first SMP instance (that adds 100% cpu load) has absolutely ZERO affect on my GPU ppd values.

I'm also going to try 1 VM running Linux along with my GTX 260 this weekend and see how that goes. Afterwards I'll try 2 VMs with the 260 and compare the results as far as efficiency goes. This is on a Q9550 running @ 3.3GHZ

A bit off topic here, but what do you guys recommend for cooling your constantly folding GPUs? I'm thinking of going with something like this. [tigerdirect.com] I've mainly heard BAD things about pci exhaust fans--they simply don't blow air on the cards...

I've got two 9600 GSOs in an SLI configuration, and they were both at about 70-80c on high load--this was with a standard case with the PSU on top. I then switched to an Antec Three Hundred (with the PSU on the bottom--fan pointing up towards pci-e slot#3), and now my bottom card is sitting pretty at 43c with air blowing ON THE CARD, whereas my top card is 70-80c.

I've got 2x front intake fans, 2x exhaust in the rear, and an intake on the side vent. Suggestions? Having my card sitting at 75-80c while folding 24/7 is just plain unacceptable. Fans are expensive--I don't want to keep buying the wrong ones. If you've got ANY experience with this at all, please share your input.

Picture for reference--don't mind the missing SLI bridge, have a 90mm flex version on order showing up tomorrow :)

Yeah definitely stay away from the PCI exhaust fans. I've tried using them before and they were all but useless in my opinion. Your best bet would be a third party GPU Cooler to replace the stock one one your card. This would probably show the biggest improvement, with the fans you linked coming in behind them. The GPU Cooler would probably be more expensive though and you would need to find one that would fit your 9600 GSO.

Thanks. I forgot to comment on vga coolers--I definitely don't want to put an aftermarket cooler on my card because it's cost prohibitive, would take up too much of my time, carries risk (if I frak it up), and would likely eat into the PCI slot just below it. I'm pretty sure swapping out your stock fan will only void the warranty--I'd consider it on a card no longer under warranty.

I've been reading that exhaust fans were the way to go, but I'm now seeing firsthand just what direct air cooling on the card surface can do. Keeping it 30c cooler (with a PSU fan no less!) is not something you simply brush off.

I think I will spring for the pci cooler I linked--it's obviously not an exhaust fan (no grill on the pci retainer), and should provide more CFM than the PSU fan...

EDIT**********
after taking the time to search for a pci slot fan, I've finally settled on this one:
http://www.chiefvalue.com/Product...1924010060
$13 shipped is the cheapest price I could find on one of these things, and this one is highly customizable. I'm thinking of adding two more fans to the rails:
http://www.ocforums.com/showthread.php?t=599056

I thought this deserved a new post:

Is there any way we can give the team folding a bit more presence on the site? Maybe a link or image could be placed somewhere on the main Slickdeals page? There's almost no way for other members to know about Folding without seeing the sigs from current folders posted in the forums... and if they've got sig viewing turned off, they'll NEVER know!! There aren't that many active folders on this site, despite the huge numbers of viewers Slickdeals gets.

I didn't find out about this for nearly half a year until I got curious enough and actually clicked on someones sig and realized it wasn't just a picture

A lot of people give up on this right away because it's too hard to figure out---and the guides on the first page are getting outdated. You simply could not follow the guides to the letter and accurately setup client folding.

Obviously this is intended for a mod or admin to consider, but I'm sure other members here agree that this little project deserves just a little more attention (especially considering most of the staff appear to be folders)

By the way, yet another good thread concerning the setup and execution of VMWARE/Linux on your x64 systems to capitalize on A2 core efficiency [xcpus.com]

i second this. perhaps its own tab/section along the menu on the site.

has anyone contacted SD itself?

p/s.: congrats to TEAM SD for slotting into rank 78!!

Thought about it--haven't tried it. I figure it would be easier to convince them to spare some free space on the main page IF a large enough following of site folders agreed this would be a good idea.

If you actually check the list of folding teams, you'll notice that Slickdeals is beginning to slow down... there are few teams ahead with lower PPD values, and there are a bunch of newer teams with monster values just behind and about to pass. We need new blood to keep the boat afloat.

100 average folders on a site with viewing this massive? C'monnnnnnn :)

Anyone see this thing? Someone needs to make one of these for Team SD.
http://www.engadget.com/2008/08/1...oints-day/
http://www.blogcdn.com/www.engadget....8fahnvidia.jpg

I like this solution [atlasfolding.com].

Huh, I always wondered how people were able to have more than 16 separate clients running. I always thought you were limited to 16 TOTAL per account name. Seeing people running 51 GPUs... it's obvious there's a workaround for limited machine ids, eh?

Hmm, machine ID's are per system.

Now I know--and knowing is half the battle!

I would love for the SlickDeals Folding team to have more of a presence on this site. On most other websites, the Folding team gets their own section, and I would think this would raise the awareness here on SlickDeals.

And now onto other things...

Ok, so I said that I would be working on setting up a Folding VM this weekend and after a lot of problems, I have finally got it up and working...and am very pleased with the results. :)

After installing VMWare, I setup a Ubuntu VM running the SMP client on two cores of my Q9550 w/ GTX 260. I kept getting blue screens and finally found the problem to be with my memory when all 4 x 2GB sticks were running in dual channel mode and a VM was running. I ended up having to take one of the sticks out just to make some progress with the VM setup and will come back to my RAM issue later ( I believe its being undervolted and\or timing issue).

When I finally got a stable VM going I took note of my PPD on my GTX 260. Although I didn't note what project I was working on, I was getting roughly 6600 PPD on a 1680 pointer. When I started the VM up, my PPD on the GPU client took a huge dive to a little over 5000 PPD, with the VM running the A2 core giving about 2200 PPD. While this was a net gain of 600 PPD, I was not very pleased with it.

Finally after doing some research, I set the CPU Affinity for the VM to Core0 & Core1 while giving the GPU Client access to all 4 cores. With this setup, my GPU Client is running project 5904 (1680pts) at 6509 PPD and my VM is running project 2675 (1920 pts) at 2865 PPD. With this setup, my GPU client was barely affected, if at all.

I'm not for sure if I will try and get two VMs going at the same time since I use this machine for other things besides folding, but I might try sometime just to see if I can get two VMs going without adversely affect my GPU client. Bottom line for now though, it would be worth it to have a VM + GPU client going at the same time.

Glad other people agree--even in reduced screen mode, there's still room between the top ad and the Slickdeals.net logo. There's room for a button at least. If not there, then maybe on the Toolbar next to the searchbar!!!

Props for looking into and documenting this little project. I was going to try this weekend, but VMware didn't want to install, and I decided to hold off until I get more time. Every blasted time I start a project it ends up taking 10x longer than I'd planned. Sounds like you had your share of issues, and that was probably a 5+ hour investment!! You should really try the second VMWARE setup---people with lesser processors than yours have been doing it, and not just i7 people.

Question, are you running Vista x64? If so, which version of VMWARE were you using?

Just found two useful folding links:

1. General folding guide (includes VMWARE instructions!) [overclock.net]
2. EX Nvidia Folding guide [overclock.net]

My experiences with SLI and multiple GPU clients
 

Okay, I finally got my sli bridge today, and I noticed zero improvement in my gpu folding... with one client or two.

After doing some research, I found out that you need to DISABLE sli in order to see any benefit. These are the steps:

1. Disable SLI in Nvidia Control Panel (SLI bridge can remain connected)
2. Create a dummy adapter for your second card to initialize it [overclock.net]
3. Create a SECOND gpu client
4. In the parameter section of each client.cfg file (for each gpu instance), add "-gpu x" where "x" is 0 through 4. You cannot use the same number. What you're doing is assigning a different graphics card to each client.

That's it. My now pulling 3300 ppd off each card, stock settings. I'm now pulling 8500 ppd on my system. next up is running dual VMWARE clients. Hopefully I'll be chugging along at 10,000+ ppd in the very near future. Tip, you can use basically ANY of the mid-range resisters--it doesn't have to be 68/75... and they don't even have to match.

Yeah I don't have any of my video cards in SLI mode, and you only need a dummy plug if you are folding multi gpu's in Vista. In XP, just extend your desktop on that video card and you are good to go.

I'm thinking about making a VM already set up and ready to fold to make it easier on those that aren't familiar with Linux and VMWare. The only thing left for the user to do would be to configure it with their Username, TeamID, and Passkey if they use one. If you think there would be interest in this, let me know and I'll see if I can get one made along with some instructions before the weekend.

I enjoy learning as I go here, but these days I'm happy to save an hour or two where I can. I know I'd make use of it, since I probably won't have much time to invest until next week.

Heads up though--someone already created a simple VMWARE application over on overclock.net. [overclock.net] I was going to give it a go, but I wasn't seeing a location in the tutorial for entering your team number. Presumably that much has already been preset for the other team. I was thinking of backward engineering to make it team slickdeals friendly :P

i finally set this up on my desktop which will be running 24/7 with an e5200 at 3.7 and ati 4830. i don't really understand this though i have one gpu and one normal one running. is that all im supposed to be doing?

congrats on your new computer!

with a dual core cpu you could be running the SMP version which is more efficient and nets more points--more points is a rough estimate of how much workload you are processing anyway.

here is the guide: http://folding.stanford.edu/Engli...GuideDEINO
its about a 5min read and about a 5minute set u.

as regarding the gpu, you are good to go! i also have the client version. just make sure your drivers are up to date and you should be good :)

p/s.: http://www.fahmon.net/about.html is a status monitoring program. all you need to do is show it the folder where you installed either of your clients and itll show you ETA stats, points and etc.

alright, thanks for the tips! ill install the mpich one tomorrow. i have administrator right off in vista x64 right now and i just dont want to restart lol. fahmon was not very friendly in the add client divison but i figured it out none the less.

edit: hmm i cant get the mpich to even set up, it opens up command with an error every time. i can at least open up the install window with deino, just haven't gone through it since the guide says not to if im using a 64 bit system. any idea's?

I'm pretty sure you NEED administrative rights to install--anyhow, that's likely your issue. Turn off user account control or right-click and use as administrator... You'd use mpich--you can't use deino on your x64 rig

In a completely unrelated matter, I've decided to put a hold on 24/7 folding, because I'm royally pissed off. The last two nights I've wasted more than 4 hours of my time trying to correct a multi-gpu folding issue (where upon restart of your secondary gpu client, you get an error stating "At present your GPU is not supported or you need a current driver"). I've followed multiple guides, I've uninstalled and reinstalled the latest Cuda drivers (several times in fact), and I even went out and bought a pack of 100ohm resisters (confirmed working with others) even though I already had plenty of mismatched resisters. It doesn't care that my second monitor shows up in the display properties, nooooooo... it just refuses to work after the first application restart.

Every time I attempt to optimize my PPD I'm confronted with yet another roadblock (my last VMWARE excursion ended in failure >.>). I've simply had enough of all this extra work and additional work-arounds. Until they start making this stuff more user friendly (and Windows friendly for efficient A2 cores---Linux and Mac get them, c'mon!), I'm simply not prepared to give this my time as WELL as my money.

Keep on folding, but don't break your back over it!

Did you change your clientID and use the following parameters?
-gpu x (x; 0,1,2...)
-forcegpu nvidia_g80/g92 (optional)

Edit:
http://www.stanford.edu/group/pan...inGPUGuide
I'm not sure if SLI mode works yet for folding.

Indeed I did. For my two clients I used ids 1/2 and parameters -gpu 0/1 respectively. My dummy adapter appears to kick the default monitor on the second card. I'm on the very latest CUDA-approved NVIDIA drivers. SLI was turned off, and the bridge removed. I followed 3 separate guides to the letter, and no joy. I considered pulling out another monitor to test without the dummy plug, but I'd had enough---every man has his limits.

I'm not sure what -forcegpu nvidia_g92 would do exactly, as I'm not finding references online. I'll give it a go in a few days, but I'm not holding out hope there.

Hey MOkoFOko, what OS are you running for your multi GPU setup?

Vista x64. I haven't bothered trying with XPx64 because it has such lousy general support, even compared to Vistax64. It always runs fine on the first start after the settings configuration, but as soon as the program is shut down for the first time, I'm in trouble. I can keep deleting the config file and re-entering the data, but that's not a permanent solution.

wonder what the problem was, i turned UAC back on and ran it as an administrator and still ran into the same problem -_-. do i need to uninstall the CPU edition i have already before installing the new one?

edit: also my girlfriend wants an explanation of what exactly this is doing. i tried to explain it to her and had her read the website but she still doesnt understand. anyone have a way to explain this to her at like a 5th grade level haha.

You're using your unused computer resources to unlock the mysteries of the human anatomy, more specifically to find a cure for cancer. Remember the NASA project for searching the cosmos for alien life that was all the rage back in the year 2000? This is like that, but closer to home. You're given a tiny portion of the near-limitless workload, and you return the results when you're done. Honestly, this should be a tax write-off :P

I'm not a buff on your actual problem, but depending on what client you're using, you're supposed to uninstall older versions of software when you install any new ones. I go with the manual client versions (rather than tray) because I can simply copy/paste new folders and decide exactly when it's running.

I've used that excuse with my girlfriend too. Never for f@h though... :lmao:

Well I'm running my multi-gpu folder under XP and when I first booted it up, it gave some of the same errors. I ended up having to change the machine IDs around a few times before it stopped giving me an error on start up. I know in XP, you still have to extend your desktop to the other video cards, do you need to do the same with Vista with the dummy plugs?

Isn't that what they built the internet for in the first place? And isn't that exactly why hard drive capacity continues to advance? If your GF has the gall to question you again, tell her to blame the tech sector! While you're at it, be sure to ask why she keeps buying shoes. One person simply does not need more than 5 pair!!

In short, yes. I decided to play with my computer after installing my new Phenom II 920---I tried to extend the monitor on the 3 "default" monitors (2-4) and every time I hit apply, it just reverted to original settings. That was obviously wrong, and the dummy monitor wasn't showing up in Nvidia Control Panel (didn't know whether it was supposed to at first). After a reboot, I tried adjusting the Display Adapter settings again... this time I got the 3rd monitor to stick (all of my "guides" showed the second monitor being the required one), which makes sense, as my second GPU is on the 3rd PCI-E slot. My second adapter then showed up in the Nvidia control panel.

In short, it would appear that the dummy adapter has to be plugged in from the very start, as attaching it while within Windows resulted in absolutely jack squat. Anyhow, I'll check that more later---I'm more interested in ramping up the OC on this phenom--I'm currently sitting at 3.2ghz on air with straight frequency tweaking--this is going to take quite a bit of breaking in : \

Hooray. The adventure continues.

I did a reboot to up my cpu frequency after several hours of folding/testing, and while the second GPU client started right up (without the drivers error), my client immediately started spitting out UNSTABLE_MACHINE errors, trashed my WU, and started an endless loop of restarts, leading to an EUE error.

Checked my display settings (Nvidia/Display) and my second monitor no longer shows up in either, even though I made NO changes following my previous successful boot. HOORAY!! This also means I can't set the second false monitor as an extended desktop, because it no longer exists. There's nothing wrong with my files--I know it's failing simply because I don't have the backup monitor showing up. Why the client even allows itself to start when a second monitor is missing is beyond me.

Isn't folding fun?

***EDIT***
After multiple tries, it appears that you need to extend the dummy monitor BEFORE you attempt to run the multi-gpu client upon every subsequent OS login (at least for Vista). Unless you manually attempt to extend the monitor, it won't show up in the NVIDIA Control Panel, and it becomes completely unusable until your next restart...

Anyone know of a way to automate this so I don't need to keep manually setting/extending the monitor?

The extended monitor goes away on reboot if you don't have a monitor or a VGA dummy plug in the card. At least that's how it worked on my card. You can get three 75 ohm resistors and stick them into a DVI-VGA adapter to trick your video card into thinking there is a monitor plugged in.

This Link [soerennielsen.dk] has a guide to making the VGA dummy plug. For mine, I just plugged the resistors into the VGA end of a DVI-VGA adapter and then coated the thing in hot glue so it wouldn't come apart. Other option is to use a KVM switch, which will do the same thing and make your 2nd GPU think a monitor is plugged in. But a 10 pack of 75 ohm resistors is only like $1.50 at Radio Shack and we've all got extra DVI-VGA adapters. :P

You may not have seen my previous rants, but I have over 100 spare resisters laying about :P I have a dummy plug connected to my second card at all times, and I'm certain I've done it right. I even went out and bought a pack of matching 100 ohm's from radioshack several days ago.

Any idea what could be funking this process up? This whole thing has got me flummoxed! I've tried with two separate dvi-->vga adapters, and Vista simply isn't recognizing the adapter at startup. I really do not like having to extend the monitor every single time I reboot...

Using a KVM switch was the first thing I tried. That didn't do squat either.

What two video cards are you trying to use and on what mobo? Also have you tried switching the video cards into each other's PCI Express slots? Or tried running the new video card as a stand alone to see if it is stable?

I've been having some hardware problems myself. I finally got the last two video cards in for my dedicated folding rig on Tuesday. The very next day my 9800 GTX died on my main PC. I'm in the process of doing an RMA with XFX and I'm not for sure how long that will take. On the brighter side of things, I hit 500k points this week, and almost to 1k WU's!


Edit, just found something that may help as well:
Might want to give the -forcegpu nvidia_g80 flag a go. Wouldn't hurt at this point.

From what I've read the forcegpu flag doesn't work in Vista.



Woo! No. 8 spot finally taken back. Suckit Red!

Matching PNY Verto 768mb Geforce 9600 GSO's... on pci-e ports 1/3. I'm using an Asus M3N-HT Deluxe motherboard. Way ahead of ya--I swapped card positions yesterday when I added my Phenom 920 (currently pulling 2600ppd @3.3ghz). This wouldn't be SO bad if I could guarantee that I could start the extended monitor upon every restart, but sometimes an extra default monitor shows up on the display panel... in which case it refuses to save the settings for extending the monitor.

I'm considering switching to XP PRO x64... you aren't required to have a dummy adapter in XP :(

This probably goes without saying, but if your card was within the 30 days window of purchase, you should have just returned to the merchant for a replacement card. There's no reason for you to have to deal with an RMA through the manufacturer yet--they're all SOOOOOOOO slow.

LOL, within 12 hours of this post, my most recent GPU went *poof*. I called FRYs today and they're sending me a replacement (within 30 day window). How's that for irony? I'm supposed to send back the fried one once the new one arrives, and I'll get refunded on the new card. They're also providing compensation for shipping... I have to fax them a copy of the shipping receipt though.

Well, that makes sense I guess. I gave it a try the other night, and it didn't do anything (not from what I could tell anyhow)

Woo, almost going to be in the top 100 using a 3 year old desktop and 3 year old laptop. When I get some funds I'm going to get a 4830. How many ppd you think I can get out of that?

I would say around 2600 PPD. ATI cards are not as efficient at Folding compared to Nvidia cards. The equivalent of the 4830 would be the 9600 GT on the Nvidia side and it would probably net around 3500 PPD. The 9600 GSO is a rebadged 8800 GS. The 9600 GSO would probably run a bit slower for gaming (compared to the 9600 GT) but better for folding, and will probably net you in the 4k PPD range. If you got a 9600 GSO, make sure to get one with 96 stream processors.

My 9600s were pulling in 3000-3300 ppd each with stock settings. I haven't gotten to OCing them yet. Maybe in a few days once my replacement shows up

The 8800gs cards for me are doing 3600 ppd right now with shaders overclocked to 1732 (from 1375 stock speed):dontknow: It tends to vary by work unit, but these newer 1680 and 1888 point WUs are pretty consistent.

for gpu folding is it better to OC shader independent of the core clock speed, as in linked and unlinked, or it doesnt really matter?

because it seems like having the core clock linked or unlinked does not change the temperature i am getting while folding and itll just save e the hassle of changing the OC for when i game.

but if having the core clock raised will effect the stability of my folding then ill not have them linked.

You will see increases in PPD primarily from overclocking the shaders. The best thing to do when OC'ing a card is to unlink your core and shaders. If you have them linked, it will be hard to find the max clock for each setting. Download EVGA's Precision tool, RivaTuner, or ATITool. I usually use EVGA Precision for overclocking and ATITool to test the oc. Start with your core clock and slowly begin to raise it from the default speeds a little at a time. Everytime you adjust it, use ATITool to scan for artifacts. If it can scan return no errors for 10 minutes, then you are usually good to go. Keep raising your core clock until you start to get artifacts on ATITool and then back down a bit. Rinse and repeat with your shader clock.

Unlink the core and memory from the shaders and just OC the shaders.
OCing the core and memory does very little for folding and will just cause the card to overheat more quickly.

Good luck!

lately it seems that i a receiving more unstable machine on my gpu client though.

i would like to keep my OC cpu setting even though i know the gpu produces more ppd.

i have a stock shader of about 1300 and i have found that i can OC it to around 1900. it has completed 100% at 1900 before but lately it just seems to unstable left and right.

what do you guys think? is a OC of 600 more worth it or i should not risk unstable machine error? i feel that if i am going to error anyway i might as well do it at a faster speed? i mean doesnt part of the WU get sent back anyway?

There's no telling just how much credit you're getting--you could be getting pennies on the dollar. If you get enough errors within a short period, your client will lock up for a 24 hour period--and if you're not watching carefully, it could really disrupt your folding. Downclock to the point where you no longer get errors. No one ever said you had to lose your OC entirely. If your hardware is becoming unstable where it once WAS stable, that could actually be worrisome. Worry less about your PPD, and more about your wallet---F@H will not compensate you for your damaged hardware.

I agree with MOkoFOko completely. Not to mention while we all love getting as many PPD as possible, you aren't helping the cause if the work you submitted is full of errors and cannot be used.

I agree with his agreement concerning what I previously said. Addition--this thread is WAY too quiet.

I agree with his agreement of his agreement.
I also agree that this thread is too quiet.
I will also note that Slickdeals happens to be kicking total a$$ right now with folding.
That is all :)

Just wanting to test my signature

:lol: at all the agreements. Hopefully no disagreements.

How about everyone post a pic of their folding machines & specs. Always love seeing what other people have done. I'll take some snapshots and put mine up when I get home from work.

Always good to have the sig and links readily available---the more people see it, the more likely it is that members will join up.


On that note, I just got done sending a PM to the head honcho concerning improving the Team SLickdeals Folding visibility on the main site... the gist of what I sent is outlined below:
Comment if you see something you like, or if there's an issue that should be raised. Things simply won't happen unless we continue to "bump" this idea. And before anyone asks, my interest in this is purely for my own benefit :P I need more people scanning these threads to answer my stupid and inane queries!!!

By the way, I disagree with your last statement. I don't love seeing what everyone else has, because it's usually better than mine! :)

7x8800gs 384MB
1x9600gso 384MB
1x8800GT 512MB

After seeing Speedfreak post, I totally agree with your disagreement. :lol:

:nod: No disagreement there!... unless you take that to mean the disagreement which we have already agreed upon--if that is the case, then I wholeheartedly agree!

Mostly.

Anyhow, here's what I've got in my main rig:

1. Antec Three Hundred case w/6 fans
2. Phenom II 920 cpu (OC @ 3.5ghz, capped at 4.05ghz)
3. Zalmon CNPS9700 LED cpu cooler
4. 8gb ddr2 Patriot Viper ram with 4-4-4-12 timings @ 800mhz
5. M3N-HT Deluxe triple-SLI motherboard
6. 2x 768mb Nvidia Geforce 9600 GSO
7. 600W OCZ PSU
8. 1TB drive + 2x 640gb drives
9. Plus extras, etc.

Total system cost comes to $701 USD after rebates and coupons and whatnot.

Well I forgot I let my folks borrow my camera while they went on vacation so here are the specs on all the machines I have folding. I'll try and get some pics when I get my camera back. :)


Dedicated Folding Server:
-ARK 4U-500-CA Rackmount Case
-MSI K9A2 Platinum
-AMD Phenom 9600 @ 2.3GHZ
-G.Skill 2GB DDR2 SDRAM
-2x 8800GS 384MB Video Cards
-2x 8800GT 256MB Video Cards
(Running 4 GPU clients + 1 Linux VM w/SMP Client)

2008 Server
-Chieftec Bravo ATX Full Tower
-Gigabyte GA-EP45-UD3P
-Intel Q9550 @ 3.4GHZ
-OCZ Platinum 8GB DDR2 SDRAM
-Geforce GTX 260
(Running 1 GPU Client + 1 Linux VM w/SMP Client)

Main PC
-Cooler Master Centurion 5 Case (I love these cases)
-MSI P6N SLI Platinum
-Intel E6320 @ 2.OGHZ
-Corsair XMS2 2GB DDR2 SDRAM
-Geforce 9800GTX+ (Currently being RMA'd so I bought another one and will be installing tomorrow)
(Running 1 GPU Client, eventually 2)

HTPC
-Silverstone LC17-B HTPC Case
-Gigabyte GA-MA69GM-S2H
-AMD Athlon 64 X2 5000 @ 3.0GHZ
-G.Skill 2GB DDR2 SDRAM
(Running 1 Linux VM w/ SMP Client)

You built a server around the Phenom 9600? When I think of the step B2 phenoms... stability is not the thing that comes to mind. More like explosions, derailed trains, murdered puppies... that sort of thing.

:lmao:

I gave about $80 for it and I haven't had any stability issues with it. It has the worst OC potential of any processor I've ever owned, but it gets the job done. If a slick deal for a Phenom II came around, I might jump.

Well... obviously you're not running Vista x64. Good luck trying to pair those two together.

A deal for a phenomII did come around--just last week the 920 was $135.

http://vasko.biz/pix/79kPPD.jpg
(Problems with heat means different over or underclocking for the 9800gx2's which explains the varied performance.)

I've got two question/comments concerning that:

1. Three of your cpu smp clients pull in under 300 ppd. I'm curious, is it worth the extra energy and heat in your systems to raise your PPD a grand total of 500? If money is no object, then I guess the point is moot.
2. I noticed you've got a Phenom II 940 running under two SMP clients (pulling in about 3200ppd)---my Phenom II 920 OC'd gets 2800 ppd on a single client... can you confirm how much you get done on a single client? I simply wasn't seeing any noticeable benefit from running multiple SMP clients on the same processor. If you're seeing something I'm not, I'd most definitely like to know. Also, is your 940 OC or stock?