PDA

View Full Version : New Q6600 ?


yog
12-30-07, 01:23 AM
OK, so I finally took the plunge and got a q6600. Now I want to use it to fold to get the most folding power out of it. I installed the 4 core program, however no matter what I do it will only use 3 of the 4 cores. I've re-installed windows, and it only uses 3 of the 4 cores. The first time core 3 wouldn't do anything, now core 2 won't do anything (and I'm not seeing much being done by core 4 either). What have I done wrong?

Thanks

jmsanders2
12-30-07, 01:28 AM
if you installed the smp client, (from my experience) each time I boot up I have fah and mpiec (or mpiex or something like this) running behind the scenes. I have to close these two together and quickly to get all four cores running full blow. This may be your problem if you are running smp...

Adak
12-30-07, 01:47 AM
OK, so I finally took the plunge and got a q6600. Now I want to use it to fold to get the most folding power out of it. I installed the 4 core program, however no matter what I do it will only use 3 of the 4 cores. I've re-installed windows, and it only uses 3 of the 4 cores. The first time core 3 wouldn't do anything, now core 2 won't do anything (and I'm not seeing much being done by core 4 either). What have I done wrong?

Thanks

Yog, I'm not the best person to answer this, but it's quite late, so I'll mention a few things since you're still up.

First, tell us about the computer: what speed are you running it at, what version of Windows, and how much RAM does it recognize?

Did you install it with the install bat program? In Task Manager, do you have 4 Fahcore threads showing?

Please copy and paste what the FAH client writes into it's FAHlog.txt file, WHEN IT FIRST STARTS FOLDING, in a reply post. That will be a big clue for us, about what's going on.

If you can also check your temps while you're folding (cpu-z, is one free program that usually does a good job of this), that would also be great.

Lastly, you may have a "power saving feature" that is slowing down your folding, to save power. Check your BIOS.

That's a good start, but I'll have to leave the rest to other's with more quad experience, and that is likely not to happen until Sunday AM.

yog
12-30-07, 01:51 AM
OK, I'm not sure what I did, but I really screwed something up. I just downloaded FAHmon, and it says that Core 2 isn't doing anything, Core 4 isn't moving and that all WU are being done by...well it said anonymous for the longest time, but now it has my name in there.

jmsanders2: I have no idea if I installed the smp client. How can I tell?

Adak: I have my CPU up to 2.52 right now, it recognizes all of my 2 gigs of RAM, running XP. In Task Manager I have 2 Fahcore threads showing. This is the log of my core 4 fahlog file:

Launch directory: C:\Program Files\FAH\FAH4
Executable: C:\Program Files\FAH\FAH4\FAH502-Console.exe
Arguments: -local -service -forceasm -advmethods -verbosity 9

Warning:
By using the -forceasm flag, you are overriding
safeguards in the program. If you did not intend to
do this, please restart the program without -forceasm.
If work units are not completing fully (and particularly
if your machine is overclocked), then please discontinue
use of the flag.

[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

A potential conflict was detected:

Process 3964 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\Program Files\FAH\FAH3 before restarting.

and here is core 2:

Launch directory: C:\Program Files\FAH\FAH2
Executable: C:\Program Files\FAH\FAH2\FAH502-Console.exe
Arguments: -local -service -forceasm -advmethods -verbosity 9

Warning:
By using the -forceasm flag, you are overriding
safeguards in the program. If you did not intend to
do this, please restart the program without -forceasm.
If work units are not completing fully (and particularly
if your machine is overclocked), then please discontinue
use of the flag.

[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

A potential conflict was detected:

Process 1692 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\Program Files\FAH\FAH1 before restarting.

How can I fix these problems?

Adak
12-30-07, 03:22 AM
OK, I'm not sure what I did, but I really screwed something up. I just downloaded FAHmon, and it says that Core 2 isn't doing anything, Core 4 isn't moving and that all WU are being done by...well it said anonymous for the longest time, but now it has my name in there.

jmsanders2: I have no idea if I installed the smp client. How can I tell?

Adak: I have my CPU up to 2.52 right now, it recognizes all of my 2 gigs of RAM, running XP. In Task Manager I have 2 Fahcore threads showing. This is the log of my core 4 fahlog file:

Launch directory: C:\Program Files\FAH\FAH4
Executable: C:\Program Files\FAH\FAH4\FAH502-Console.exe
Arguments: -local -service -forceasm -advmethods -verbosity 9

Warning:
By using the -forceasm flag, you are overriding
safeguards in the program. If you did not intend to
do this, please restart the program without -forceasm.
If work units are not completing fully (and particularly
if your machine is overclocked), then please discontinue
use of the flag.

[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

A potential conflict was detected:

Process 3964 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\Program Files\FAH\FAH3 before restarting.

and here is core 2:

Launch directory: C:\Program Files\FAH\FAH2
Executable: C:\Program Files\FAH\FAH2\FAH502-Console.exe
Arguments: -local -service -forceasm -advmethods -verbosity 9

Warning:
By using the -forceasm flag, you are overriding
safeguards in the program. If you did not intend to
do this, please restart the program without -forceasm.
If work units are not completing fully (and particularly
if your machine is overclocked), then please discontinue
use of the flag.

[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

A potential conflict was detected:

Process 1692 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\Program Files\FAH\FAH1 before restarting.

How can I fix these problems?

You have installed the wrong client - and you're running two of them at the same time. (That's why you get those warnings).

This is what your start up should look like:


# SMP Client ################################################## ################
################################################## #############################

Folding@Home Client Version 5.91beta5

http://folding.stanford.edu

################################################## #############################
################################################## #############################

Launch directory: C:\Program Files\Folding@Home Windows SMP Client V1.01
Executable: C:\Program Files\Folding@Home Windows SMP Client V1.01\fah.exe
Arguments: -verbosity 9 -forceasm

Warning:
By using the -forceasm flag, you are overriding
safeguards in the program. If you did not intend to
do this, please restart the program without -forceasm.
If work units are not completing fully (and particularly
if your machine is overclocked), then please discontinue
use of the flag.

[22:19:08] - Ask before connecting: No
[22:19:08] - User name: Adak (Team 32)
[22:19:08] - User ID: 42B5CD637E404119
[22:19:08] - Machine ID: 1
[22:19:08]
[22:19:08] Loaded queue successfully.
[22:19:08]
[22:19:08] + Processing work unit
[22:19:08] - Autosending finished units...
[22:19:08] Core required: FahCore_a1.exe
[22:19:08] Trying to send all finished work units
[22:19:08] Core found.
[22:19:08] + No unsent completed units remaining.
[22:19:08] - Autosend completed
[22:19:08] Working on Unit 09 [December 29 22:19:08]
[22:19:08] + Working ...
[22:19:08] - Calling 'mpiexec -channel auto -np 4 FahCore_a1.exe -dir work/ -suffix 09 -checkpoint 30 -forceasm -verbose -lifeline 3340 -version 591'

[22:19:09]
[22:19:09] *------------------------------*
[22:19:09] Folding@Home Gromacs SMP Core
[22:19:09] Version 1.74 (March 10, 2007)
[22:19:09]
[22:19:09] Preparing to commence simulation
[22:19:09] - Ensuring status. Please wait.
[22:19:26] - Assembly optimizations manually forced on.
[22:19:26] - Not checking prior termination.
[22:19:34] - Expanded 2971382 -> 15199495 (decompressed 511.5 percent)
[22:19:35]
[22:19:35] Project: 2653 (Run 13, Clone 126, Gen 31)
[22:19:35]


You can tell it's the SMP client, it will show SMP Client on one of the first few lines of the FAHlog.txt file (or window, if you have one open for it).

I highlighted it in Red above, taken from my SMP Windows rig.

So first thing is to remove the old clients, and d/l the SMP client.

Stop all FAH programs, You have a service install according to the log you posted. So, Start>Control Panel>Administrative Tools>Services> and then select the FAH service. Right click on it, and select "Action">>"Off". Then go back to the FAH service before you right clicked on it, and click on Manual.

Now go into Windows Explorer, and right click on Fah.exe. program. Select "make shortcut". Now right click on that shortcut (there's the method to my madness) :D, and add this to the initialization string: -configonly. Just one space before the hyphen, and all one word.

Now click on apply and OK, and double click on the shortcut to start up FAH. It will ask you straightaway, if you want it to uninstall itself as a service? Enter Yes, and that one should be toast as a service.

Repeat for the other install of FAH, and then reboot to check that no FAH program is running, as a service or as a program in the foreground.

Now just delete the entire FAH directory of files, for each install.

Go to the Stanford D/L page and get the SMP client 5.91beta5, for Windows.

BEFORE you install it. Read and take notes on the blue question mark's, install directions. You MUST have netframe 2.0 or better installed on your rig, AND you must have an administrative account for yourself with FULL rights, and be signed into that account, before trying to run install.bat.

Then run the install bat program, and you should be aces.

You have to follow those install directions. We've all tried cutting lots of corners - and no luck on this, at all. :bang head:

You will install just ONE FAH SMP program, for right now. In a bit, if you want to try it, we'll show you how to run two of them, for slightly more points, if you want.

And I'm going to bed before the sandman makes me asleep sitting up. :)

Post any troubles, and you should get some help later this AM.

Hack30
12-30-07, 09:15 AM
It sounds like he installed the Wedo one-click for multiple cores and didn't set up each config file or change the machine ID's either.


core1
[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

core2
[07:32:25] - Ask before connecting: No
[07:32:25] - User ID: 4940104177CA1273
[07:32:25] - Machine ID: 1
[07:32:25]

I don't think its a Windows problem. I think its a folding setup problem.
both clients are trying to crunch the same data. you have to change the machineid for all four clients if you want to use the uni-core client x 4


SMP client all the way on a quad!:thup:

ChasR
12-30-07, 09:17 AM
To add to what Adak said, for the SMP client to work, the admin account running it must have a password. Also, installing the SMP client as a service isn't supported, so don't. It can be made to work, but it's best not to start off trying unsupported stuff. If you're running XP, the Win SMP client must have have a network connection 100% of the time. Lose the net connection and you stop folding. Not good on a flakey wireless connection. The SMP client will produce roughly twice the ppd of 4 uniprocessor clients. So this is the client you want.

Your original install of the uniprocessor client probably would have worked if you'd given each instance a unique machine ID. Per the log you posted, it appears you've borked the client.cfg files and have no username or team number on the two instances in the post. If you manually edit the .cfg file, you MUST NOT use Word or Wordpad. Notepad or or some of the third party apps like Notetab and Metatab will work. Don't delete the carriage return (end of line) symbols if using notepad.

yog
12-30-07, 10:58 AM
OK, now I'm really confused. I do want to run the SMP client? I'm all in favor of the more work units, but I do have a flakey wireless connection and if it goes down I don't want my folding to stop.

I got core 2 to work, but core 4 is still having a conflict with core 2. I rechecked the config files and they all have different machine Id's. I just started core 4 manually and now it is running happily.

Should I still install the SMP client? Should I re-install windows? I just installed it so I'm not opposed to doing that. Adak, I'm sorry but I didn't get your steps for uninstalling FAH client, could you be more specific?

deadlysyn
12-30-07, 11:35 AM
I think it is better to give that same client you are running a little time, finish up the WU's you are working on, maybe set up the -oneunit flag, and take some time to get used to that particular client before trying to jump into SMP. That is what seems to be recommended to a lot of people starting out this way. I think this way, you would be able to get the hang of things a lot easier, as a few minutes of experience is worth several hours of being shown IMO.

Edit: Also, not finishing the WU's or deleting them is bad for the science. It could be one of those WU's that solves the cure for the diseases that we are looking for.

Adak
12-30-07, 02:58 PM
Without a doubt at all, SMP is the client you want to use with a Quad. It's also more difficult because the client is new, and still in beta release. Yes, it has small bugs in it still.

With a flakey wireless connection, it would probably just be more problems than joy.

At this point I believe running the single core client while you learn more about FAH, and then stepping up to the SMP client, would be a better way to go.

And the Linux SMP client will KEEP FOLDING if the internet connection is lost! For a long term folder, you can't get better than Linux SMP. That's my preferred set-up for a folding rig - but you can't play a lot of games on Linux. :(

Let's get together on the forum at say 4 to 5pm Pacific time, (since you're in WA & I'm in CA), and get your rig, going.

Until then

yog
12-31-07, 01:21 PM
So I've got it working now, mostly. The only problem is that when I first start up the computer two sessions don't start up with the computer. Sessions 2 and 4 to be specific. Here is the message I get with session 2:

[18:06:42] - User ID: 4940104177CA1273
[18:06:42] - Machine ID: 1
[18:06:42]

A potential conflict was detected:

Process 1884 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\FAH_1 before restarting.

Please press any key to exit.

Here is what I get from session 4:

[18:06:43] - User ID: 4940104177CA1273
[18:06:43] - Machine ID: 1
[18:06:43]

A potential conflict was detected:

Process 1968 is currently running and may also be a client with Mach. ID 1.
Program will now exit. Upon restart, this check will not be done --
you may wish to check that no client is currently running in
C:\FAH_3 before restarting.

Please press any key to exit.

Any idea as what I can do to fix this problem? It really isn't much of a problem but I'd like to have it start up by itself without me having to tinker with it to get it to run correctly.

Thanks

ChasR
12-31-07, 01:42 PM
Here's the problem:
Instance 2:
[18:06:42] - User ID: 4940104177CA1273
[18:06:42] - Machine ID: 1
[18:06:42]

Instance 4:
[18:06:43] - User ID: 4940104177CA1273
[18:06:43] - Machine ID: 1
[18:06:43]
The two are identical and likely match Instance 1.

Run the config program to change the machine IDs or edit client.cfg so that the instance in the FAH1 folder is machine ID 1, in FAH2, edit client.cfg so that machine ID is 2, machine ID 3 for FAH3, and machine ID 4 for FAH4. All four instances will work. It still appears you have no username or team number.

To rebuild you client.cfg, I suggest you run services.msc and stop all four FAH services. Then run "C:\Program Files\FAH\FAH1\FAH502-Console.exe" -configonly and enter your user name and team number. Leave every thing else alone. Then run "C:\Program Files\FAH\FAH2\FAH502-Console.exe" -configonly and enter your username and team number, change nothing until you get to the advanced options question, to which you answer yes. In the advanced options section, change machine ID to 2. Repeat this process for FAH3, changing machine ID to 3, and for FAH 4, changing machine ID to 4. This will rebuild all four client.cfgs with no chance of borking the file. When you're done, go back to services.msc and restart the four services. It'll work fine after that.

yog
12-31-07, 01:55 PM
Then why does instance 3 not have any problems, since it is also machine ID 1?

EDIT: I just restarted and instance 1,2,and 4 started up just fine this time. I guess I'd better change 3 to machine ID 3

ChasR
12-31-07, 02:14 PM
I can't tell you that, but it's highly likely that, if instance 3 is working on anything, it's working on the identical WU to instance 1 and you'll only get points for the first one turned in. The Asignment Server (AS) keeps track of the user ID and machine ID of each assignment. It sees all 4 of your instances to be identical. As you have it set up, if instance one turns in a WU and requests another, the AS knows it made the assignment to User ID: 4940104177CA1273, Machine ID: 1. If instance 3 subsequently requests assignment the AS sees the exact same rig, User ID: 4940104177CA1273, Machine ID: 1, requesting a WU without having turned in its last assignment and so assigns the exact same WU again. You can fold away on it, but you won't get any points for it. The machine IDs MUST be different on each instance for the AS to recognize each as unique.