SMP not utizeing all four cores at 100%

[science man]

It's utilizing all four cores but not at 100%. I tired altering the avanced options to tell to usage but it still won't do it. Does anybody know what do to?

[benbaked]

Your thread title says it's not utilizing all four cores at 100%, but your post says that it is. Which is it and which are you trying to accomplish? Do you want it to use all four cores at 100% or do you want to limit the cpu usage? If the latter, how are you changing your advanced options? Are you editing the config file, or running the client with the "-configonly" option to go through the setup questions again? I think using the -configonly option is the preferred way to do it.

[blakehenry00]

I think he means that it's using all of the cores, just not at 100%. I am having the same problem with the 6.22 Client.

[science man]

Oh sorry there I edited the post.

[science man]

I think he means that it's using all of the cores, just not at 100%. I am having the same problem with the 6.22 Client.

Yea that's it. Does anyone know what to do?

[Hack30]

So just how much of the cores are utilized?

50%
30%
80%
?%

[blakehenry00]

I just went back to the 5.xx one for now.

[science man]

Your thread title says it's not utilizing all four cores at 100%, but your post says that it is. Which is it and which are you trying to accomplish? Do you want it to use all four cores at 100% or do you want to limit the cpu usage? If the latter, how are you changing your advanced options? Are you editing the config file, or running the client with the "-configonly" option to go through the setup questions again? I think using the -configonly option is the preferred way to do it.

I only know how to change the advancedd options when first installing the client because when in that process it asks you if you want to change them.

[science man]

So just how much of the cores are utilized?

50%
30%
80%
?%

around 25%

[benbaked]

Hmm, is this in Windows or Linux? With the linux client you have to use the "-smp" switch when starting the client to instruct it to download SMP units, otherwise it will only work on single processor clients, which would explain 25% on each core. Can you post a snippet of your FAHlog.txt file so we can see what unit it's working on? Looks like it was working fine on SMP a couple of days ago according to your stats, so hopefully the problem is something minor. :)

[science man]

It's Windows and It's not that its not working at all because it is just not at 100%

Here's the log




--- Opening Log file [August 4 20:25:14 UTC]


# Windows CPU Console Edition #################################################
################################################## #############################

Folding@Home Client Version 6.22 SMP Beta2

http://folding.stanford.edu

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

Launch directory: C:\Users\my UN\Documents\SMP
Executable: C:\Users\my UN\Documents\SMP\Folding@home-Win32-x86.exe


[20:25:14] Configuring Folding@Home...


[20:30:36] - Ask before connecting: No
[20:30:36] - User name: Science_Man (Team 32)
[20:30:36] - User ID: 6BDBA49A7D0F3C47
[20:30:36] - Machine ID: 1
[20:30:36]
[20:30:36] Work directory not found. Creating...
[20:30:36] Could not open work queue, generating new queue...
[20:30:36] - Preparing to get new work unit...
[20:30:36] + Attempting to get work packet
[20:30:36] - Connecting to assignment server
[20:30:39] - Successful: assigned to (169.230.26.30).
[20:30:39] + News From Folding@Home: Welcome to Folding@Home
[20:30:39] Loaded queue successfully.
[20:30:39] + Closed connections
[20:30:39]
[20:30:39] + Processing work unit
[20:30:39] Core required: FahCore_82.exe
[20:30:39] Core not found.
[20:30:39] - Core is not present or corrupted.
[20:30:39] - Attempting to download new core...
[20:30:39] + Downloading new core: FahCore_82.exe
[20:30:40] + 10240 bytes downloaded
[20:30:40] + 20480 bytes downloaded
[20:30:41] + 30720 bytes downloaded
[20:30:41] + 40960 bytes downloaded
[20:30:41] + 51200 bytes downloaded
[20:30:41] + 61440 bytes downloaded
[20:30:41] + 71680 bytes downloaded
[20:30:41] + 81920 bytes downloaded
[20:30:41] + 92160 bytes downloaded
[20:30:41] + 102400 bytes downloaded
[20:30:41] + 112640 bytes downloaded
[20:30:41] + 122880 bytes downloaded
[20:30:41] + 133120 bytes downloaded
[20:30:41] + 143360 bytes downloaded
[20:30:41] + 153600 bytes downloaded
[20:30:41] + 163840 bytes downloaded
[20:30:41] + 174080 bytes downloaded
[20:30:41] + 184320 bytes downloaded
[20:30:41] + 194560 bytes downloaded
[20:30:41] + 204800 bytes downloaded
[20:30:41] + 215040 bytes downloaded
[20:30:41] + 225280 bytes downloaded
[20:30:41] + 235520 bytes downloaded
[20:30:41] + 245760 bytes downloaded
[20:30:41] + 256000 bytes downloaded
[20:30:41] + 266240 bytes downloaded
[20:30:41] + 276480 bytes downloaded
[20:30:41] + 286720 bytes downloaded
[20:30:41] + 296960 bytes downloaded
[20:30:41] + 307200 bytes downloaded
[20:30:41] + 317440 bytes downloaded
[20:30:41] + 327680 bytes downloaded
[20:30:41] + 337920 bytes downloaded
[20:30:41] + 348160 bytes downloaded
[20:30:41] + 358400 bytes downloaded
[20:30:41] + 368640 bytes downloaded
[20:30:41] + 378880 bytes downloaded
[20:30:41] + 389120 bytes downloaded
[20:30:41] + 399360 bytes downloaded
[20:30:41] + 409600 bytes downloaded
[20:30:41] + 419840 bytes downloaded
[20:30:41] + 430080 bytes downloaded
[20:30:41] + 440320 bytes downloaded
[20:30:41] + 450560 bytes downloaded
[20:30:41] + 460800 bytes downloaded
[20:30:41] + 471040 bytes downloaded
[20:30:41] + 481280 bytes downloaded
[20:30:41] + 491520 bytes downloaded
[20:30:41] + 501760 bytes downloaded
[20:30:41] + 512000 bytes downloaded
[20:30:41] + 522240 bytes downloaded
[20:30:41] + 532480 bytes downloaded
[20:30:41] + 542720 bytes downloaded
[20:30:41] + 552960 bytes downloaded
[20:30:41] + 563200 bytes downloaded
[20:30:41] + 573440 bytes downloaded
[20:30:41] + 583680 bytes downloaded
[20:30:41] + 593920 bytes downloaded
[20:30:41] + 604160 bytes downloaded
[20:30:41] + 614400 bytes downloaded
[20:30:41] + 624640 bytes downloaded
[20:30:41] + 634880 bytes downloaded
[20:30:41] + 645120 bytes downloaded
[20:30:41] + 655360 bytes downloaded
[20:30:41] + 665600 bytes downloaded
[20:30:41] + 675840 bytes downloaded
[20:30:41] + 686080 bytes downloaded
[20:30:41] + 696320 bytes downloaded
[20:30:41] + 706560 bytes downloaded
[20:30:41] + 716479 bytes downloaded
[20:30:41] Verifying core Core_82.fah...
[20:30:41] Signature is VALID
[20:30:41]
[20:30:41] Trying to unzip core FahCore_82.exe
[20:30:41] Decompressed FahCore_82.exe (1683456 bytes) successfully
[20:30:46] + Core successfully engaged
[20:30:51]
[20:30:51] + Processing work unit
[20:30:51] Core required: FahCore_82.exe
[20:30:51] Core found.
[20:30:51] Working on queue slot 01 [August 4 20:30:51 UTC]
[20:30:51] + Working ...
[20:30:51]
[20:30:51] *------------------------------*
[20:30:51] Folding@Home PMD Core
[20:30:51] Version 1.03 (September 7, 2005)
[20:30:51]
[20:30:51] Preparing to commence simulation
[20:30:51] - Looking at optimizations...
[20:30:51] - Created dyn
[20:30:51] - Files status OK
[20:30:51] - Expanded 16941 -> 106875 (decompressed 630.8 percent)
[20:30:51]
[20:30:51] Project: 4596 (Run 28, Clone 110, Gen 8)
[20:30:51]
[20:30:51] Assembly optimizations on if available.
[20:30:51] Entering M.D.
[20:30:58] Protein: p4596_T0498-12_minout
[20:30:58]
[20:30:58] Completed 0 out of 2500000 steps (0%)
[20:36:01] Writing local files
[20:36:01] Completed 25000 out of 2500000 steps (1%)
[20:41:03] Writing local files
[20:41:03] Completed 50000 out of 2500000 steps (2%)
[20:41:03] Writing checkpoint files
[20:46:05] Writing local files
[20:46:05] Completed 75000 out of 2500000 steps (3%)
[20:51:13] Writing local files
[20:51:13] Completed 100000 out of 2500000 steps (4%)
[20:51:13] Writing checkpoint files
[20:56:18] Writing local files
[20:56:18] Completed 125000 out of 2500000 steps (5%)
[21:01:25] Writing local files
[21:01:25] Completed 150000 out of 2500000 steps (6%)
[21:01:25] Writing checkpoint files
[21:06:30] Writing local files
[21:06:30] Completed 175000 out of 2500000 steps (7%)
[21:11:38] Writing local files
[21:11:38] Completed 200000 out of 2500000 steps (8%)
[21:11:38] Writing checkpoint files
[21:16:42] Writing local files
[21:16:42] Completed 225000 out of 2500000 steps (9%)
[21:21:50] Writing local files
[21:21:50] Completed 250000 out of 2500000 steps (10%)
[21:21:50] Writing checkpoint files
[21:26:58] Writing local files
[21:26:58] Completed 275000 out of 2500000 steps (11%)
[21:32:21] Writing local files
[21:32:21] Completed 300000 out of 2500000 steps (12%)
[21:32:21] Writing checkpoint files
[21:37:24] Writing local files
[21:37:24] Completed 325000 out of 2500000 steps (13%)
[21:42:31] Writing local files
[21:42:31] Completed 350000 out of 2500000 steps (14%)
[21:42:31] Writing checkpoint files
[21:47:35] Writing local files
[21:47:35] Completed 375000 out of 2500000 steps (15%)
[21:52:43] Writing local files
[21:52:43] Completed 400000 out of 2500000 steps (16%)
[21:52:43] Writing checkpoint files
[21:57:51] Writing local files
[21:57:51] Completed 425000 out of 2500000 steps (17%)
[22:03:02] Writing local files
[22:03:02] Completed 450000 out of 2500000 steps (18%)
[22:03:02] Writing checkpoint files
[22:08:12] Writing local files
[22:08:12] Completed 475000 out of 2500000 steps (19%)
[22:13:20] Writing local files
[22:13:20] Completed 500000 out of 2500000 steps (20%)
[22:13:20] Writing checkpoint files
[22:18:26] Writing local files
[22:18:26] Completed 525000 out of 2500000 steps (21%)

[MARCI]

you need to set the -smp 4 flag in your config. you can erase your client.cfg file and run the f@H client. In the advanced options when it asks if you need to use any options type -smp 4

the fahcore_82 is a single core process.. won't run on more than 1 core

[MARCI]

[settings]
username=Science_Man
team=32
passkey=
asknet=no
machineid=1
extra_parms=-smp 4

[http]
active=no
host=localhost
port=8080
usereg=no

[clienttype]
type=3

[core]
addr=


that's what your client.cfg should look like.. the extra parms is the the part you're missing

[science man]

How do I get to settings? FahCore_82 is what downloaded when I download the client. Should go into that?

[MARCI]

in the same directory where your client is installed, there is a file called client.cfg. use wordpad/notepad to edit that

[benbaked]

According to your posted fahlog, try checking C:\Users\my UN\Documents\SMP

[MARCI]

nice catch ben, can't believe i caught core82, but not his working dir 'doh

[science man]

here are my settings just tell me what I should edit to get my computer running at full power. I'm doing this because it doesn't exactly match what you posted MARCI.

[settings]
username=Science_Man
team=32
passkey=
asknet=no
machineid=1
bigpackets=big

[http]
active=no
host=localhost
port=8080
usereg=no

[core]
priority=96
disableassembly=yes
checkpoint=4
nocpulock=1
addr=

[clienttype]
type=3

[MARCI]

okay.. add the "extra_parms=-smp 4" after the "bigpackets=big"

[MARCI]

you're gonna have to delete the work directory too, since core82 won't run on smp. so you'll lose a couple hours work. have to get a new WU

[MARCI]

on my p4 ht i was getting 200ppd with core84, with smp i'm around 500ppd

[science man]

when I added that it now says


Note: Please read the license agreement (Folding@home-Win32-x86.exe -license). F
urther
use of this software requires that you have read and accepted this agreement.

4 cores detected
Unable to connect to 'FrankGreico-PC:8676',
sock error: generic socket failure, error stack:
MPIDU_Sock_post_connect(1228): unable to connect to FrankGreico-PC on port 8676,
exhausted all endpoints (errno -1)
MPIDU_Sock_post_connect(1275): unable to connect to FrankGreico-PC on port 8676,
No connection could be made because the target machine actively refused it. (er
rno 10061)
[00:27:07] Error running mpiexec. Please check your configuration and re-run in
stall.bat if necessary

[MARCI]

hmmmm.. okay.. new plan.. erase client.cfg. then re-run the client.. it should allow you to reconfig. in the advanced area where it allows you to add parameters, add -smp 4

[science man]

Ah man I did that and now it's runing but no difference. What a head banger. :bang head

[MARCI]

did you add the extraparameters in the advanced config?

[science man]

yea I typed -smp 4.

[science man]

Wow I looked at my fahlog and the server it says my computer is connected to isn't in the list of servers.

[science man]

I gotta tell yea my conclusion is that It's a problem with the client unless, do you run the same client and not have this problem?

[MARCI]

i do.. what does your client.cfg say now?







editted to add, WOOT, got my 2nd star :P

[orion456]

I assume you are trying to install the Deino version of 6.22b2 when you previously were using MPICH.

1. stop FAH

2. RUN "smpd -uninstall"

3. Delete the contents of the FahSMP folder including the work folder (you will loose your data so you might want to wait until the current WU is done)

4. reboot

5. make sure smpd isn't listed by Task Manager as a process.

6. uncompress 6.22b2 deino version to the FahSMP folder

7. run install - answer the question according to the the install instructions

8. Go here (http://www.stanford.edu/~kasson/folding/FAH6.22beta2-SMP-shm.exe) and copy the file into your FahSMP folder (it comes from this thread http://foldingforum.org/viewtopic.php?f=46&t=4411&st=0&sk=t&sd=a)

9. Create a short cut to the above file on your desktop

10. Edit the short cut to include "-smp -deino -forceasm" at the end of TARGET

11. execute the short cut and answer the questions.

It should run at 100% now.

[Adak]

Wow I looked at my fahlog and the server it says my computer is connected to isn't in the list of servers.

Oh, don't let that bother you. We do that to make those that need a conspiracy theory to be happy, happy. :)

Seriously, if you haven't updated your version of Windows to fix the "pfishing" bug, that was exposed last month, you should do that, right away.

Also, you may be directed elsewhere if your system is infected with any number of other virii. What server # are you connecting with?

[science man]

same and btw when I edited the target of the shortcut it wouldn't let me and said that it was not a vaild path.

P.S. I'm gonna go back to the way I had it runnng the simple smp even though that issue still remains I'll live.

[science man]

same and btw when I edited the target of the shortcut it wouldn't let me and said that it was not a vaild path.

P.S. I'm gonna go back to the way I had it runnng the simple smp even though that issue still remains I'll live.

the server it's connected to is 171.64.65.111 go here to see that it's not an existing server. (http://fah-web.stanford.edu/serverstat.html)

EDIT something must be right because I got credit for the WU that just finished.

[Adak]

the server it's connected to is 171.64.65.111 go here to see that it's not an existing server. (http://fah-web.stanford.edu/serverstat.html)

EDIT something must be right because I got credit for the WU that just finished.

Whoa there, runaway conspiracy lover! :)

171.64.65.111 *is* a FAH server at Stanford:

171.64.65.111 vspg5v2 vvoelz full Accepting 0.64 88 11 23836 5328 1 35.15 73 73 29653


Which is why your points were credited on your returned WU. :D

If your shortcut tells you that the path you have in it is invalid, then you have made a typo somewhere. I just did this, in fact, myself. :eek:

Because of the proportional font it's displayed in, it's tough to see whether you have added one space, or two, and the " (double quotes), are pretty small, as well.

But you can believe your OS. If it says your path is invalid - your path is invalid.

No conspiracy needed. :)

[science man]

there were no typos because I cut and pasted it right at the end of the path. No spaces. Also, although I get credit of a WU it's very little the last WU my computer did got 104 points, Whensmp WU's are soppused to get 2,000 points of credit that what I got on the old beta and yes I did set the WU size to big.

[Adak]

there were no typos because I cut and pasted it right at the end of the path. No spaces. Also, although I get credit of a WU it's very little the last WU my computer did got 104 points, Whensmp WU's are soppused to get 2,000 points of credit that what I got on the old beta and yes I did set the WU size to big.

Science man, I'm sorry, but if your computer tells you that your path or filename is invalid in a shortcut - then it's invalid. Doesn't matter if you cut and paste or stood on your head for 10 minutes - it's still invalid to the OS.

Obviously, you're not folding SMP work units, for some reason.

Here's what I need:

1) the first 50 lines of your fahlog.txt file when you first re-start the folding client.

2) A copy of your target line that the OS says is invalid.

3) A copy of your client.cfg file.

Then I can start sorting this out, in more detail. I know it's frustrating, but we'll get there.

If you'll follow the directions, it will take a lot less time than if you want to argue. I know it can be fun, but it doesn't get much actually done.

[science man]

Here's what I can give you out of thoose



Note: Please read the license agreement (Folding@home-Win32-x86.exe -license). F
urther
use of this software requires that you have read and accepted this agreement.



--- Opening Log file [August 6 04:41:25 UTC]


# Windows CPU Console Edition #################################################
################################################## #############################

Folding@Home Client Version 6.22 SMP Beta2

http://folding.stanford.edu

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

Launch directory: C:\Users\my name\Documents\SMP
Executable: C:\Users\my name\Documents\SMP\Folding@home-Win32-x86.exe


[04:41:25] - Ask before connecting: No
[04:41:25] - User name: Science_Man (Team 32)
[04:41:25] - User ID: 6BDBA49A7D0F3C47
[04:41:25] - Machine ID: 1
[04:41:25]
[04:41:26] Loaded queue successfully.
[04:41:26]
[04:41:26] + Processing work unit
[04:41:26] Core required: FahCore_82.exe
[04:41:26] Core found.
[04:41:26] Working on queue slot 02 [August 6 04:41:26 UTC]
[04:41:26] + Working ...
[04:41:26]
[04:41:26] *------------------------------*
[04:41:26] Folding@Home PMD Core
[04:41:26] Version 1.03 (September 7, 2005)
[04:41:26]
[04:41:26] Preparing to commence simulation
[04:41:26] - Looking at optimizations...
[04:41:26] - Files status OK
[04:41:26] - Expanded 93122 -> 599777 (decompressed 644.0 percent)
[04:41:26]
[04:41:26] Project: 2170 (Run 52, Clone 210, Gen 7)
[04:41:26]
[04:41:26] Assembly optimizations on if available.
[04:41:26] Entering M.D.
[04:43:15] Protein: p2170_lambda_obc_300K
[04:43:15]
[04:43:15] Completed 203025 out of 500000 steps (40%)



[settings]
username=Science_Man
team=32
passkey=
asknet=no
machineid=1
bigpackets=big
local=1

[http]
active=no
host=localhost
port=8080
usereg=no


I can't give you the traget because I don't run denio and the text to insert that orion456 wrote was for the denio version.

btw when I said that I cut and pasted that text II was just trying to prove that their was no typo.

EDIT: Since the WU's are worth such little credit I've decided to not run it until this is fixed.

[orion456]

I can't give you the traget because I don't run denio and the text to insert that orion456 wrote was for the denio version.

btw when I said that I cut and pasted that text II was just trying to prove that their was no typo.


I don't see the "-smp" switch listed. You have to include that switch on the Target line of the shortcut even for MP. The switch is placed outside the quotation marks at the end.

You should switch to deino because it won't quit when the net goes down momentarily.

[science man]

I unstalled the client I had and was go to install the denio one but I don't see it here (http://folding.stanford.edu/English/DownloadWinOther)

[orion456]

I unstalled the client I had and was go to install the denio one but I don't see it here (http://folding.stanford.edu/English/DownloadWinOther)

2nd last one is the deino new 6.22b2 client...no?

Last one is the MPICH version.

[Leonardo]

I don't see the "-smp" switch listed. Run the config setup again for the client. Select yes or y for advanced features. When you get the line that asks for additional parameters, simply type in -smp. If you don't enter that, the client will be a uniprocessor Folder. That's for the MPI~ version - don't know if the deino version has that configuration entry or not.

[science man]

Run the config setup again for the client. Select yes or y for advanced features. When you get the line that asks for additional parameters, simply type in -smp. If you don't enter that, the client will be a uniprocessor Folder. That's for the MPI~ version - don't know if the deino version has that configuration entry or not.

No that doesn't work and I unziped denio but there's no run file like there is on the other client. Can somebody link me to one or should I just use the one from the other client?, which to me doesn't make sense.

[orion456]

No that doesn't work and I unziped denio but there's no run file like there is on the other client. Can somebody link me to one or should I just use the one from the other client?, which to me doesn't make sense.

Unzip to your FahSMP folder. Run the install program in the FahSMP folder, not from the unzipper application. The unzipping produces the client "Folding@home-win32-x86.exe" in the FahSMP folder. Create a short cut to it and then open the properties and add -smp after the quotation marks on the target line.

[Adak]

You need to add the <space>hyphensmp flag. That's one space, then a hyphen -, and then smp with no space between the hyphen and the smp.

Only after you add that as a starting flag, will we be able to see if there is anything else that needs to be sorted out to get you up and folding smp WU's.

[science man]

Unzip to your FahSMP folder. Run the install program in the FahSMP folder, not from the unzipper application. The unzipping produces the client "Folding@home-win32-x86.exe" in the FahSMP folder. Create a short cut to it and then open the properties and add -smp after the quotation marks on the target line.

Ok this time it allows me to edit the target which I did and then when I run that shortcut it still says this


Note: Please read the license agreement (Folding@home-Win32-x86.exe -license). F
urther
use of this software requires that you have read and accepted this agreement.

4 cores detected

job aborted:
process: node: exit code[: error message]
0: FRANKGREICO-PC: -1: Unable to establish a secure connection to host 'FRANKGRE
ICO-PC' for rank 0 using user account 'FrankGreico-PC\Frank Greico'
Error: No connection could be made because the target machine actively refused i
t 192.168.1.104:62740

1: FRANKGREICO-PC: -1: Unable to establish a secure connection to host 'FRANKGRE
ICO-PC' for rank 1 using user account 'FrankGreico-PC\Frank Greico'
Error: No connection could be made because the target machine actively refused i
t 192.168.1.104:62740

[00:41:55] Error running mpiexec. Please check your configuration and re-run in
stall.bat if necessary



EDIT: You guys think I should just ebnch big computer out myself by running four of those clients at once?

DUOBLE EDIT: I'm gonna try it.

[Adak]

You are not installed correctly, Science Man.

This could be caused by:

1) You don't have an account with a password, with full administrative rights.

2) You aren't signed into that account, using the same password you used when you installed mpiexec.

3) You don't have .net framework version 2.0 installed. You may have higher versions of .net framework, but this install requires parts of the 2.0 version to be installed. (Other versions can also be installed, but the 2.0 version has to be there, they say).

4) Somebody with Vista needs to finish this off, but you may need to turn off your "UAR?", (User Access Rights, I believe), which is a security feature that can interfere with installing or running FAH.


Read your error message at 00:41:55, to confirm, as well as the first two messages. It is your own machine that is refusing the connection that mpiexec needs.

If you are running Vista, you can forget #3 above, your system will already have .net framework 2.0.

[orion456]

Ok this time it allows me to edit the target which I did and then when I run that shortcut it still says this


It does look like you didn't enter the right password for that user. Just run install again and pick repair.

[science man]

It does look like you didn't enter the right password for that user. Just run install again and pick repair.

Idk what you mean but it doesn't matter because I just decided to put three clients on my computer, which is better idea considering much much a big WU is wroth with these clients. I only manage to get three which cracked it up to 75% in total cpu usage.

[Mr.Guvernment]

I assume you are trying to install the Deino version of 6.22b2 when you previously were using MPICH.

1. stop FAH

2. RUN "smpd -uninstall"

3. Delete the contents of the FahSMP folder including the work folder (you will loose your data so you might want to wait until the current WU is done)

4. reboot

5. make sure smpd isn't listed by Task Manager as a process.

6. uncompress 6.22b2 deino version to the FahSMP folder

7. run install - answer the question according to the the install instructions

8. Go here (http://www.stanford.edu/~kasson/folding/FAH6.22beta2-SMP-shm.exe) and copy the file into your FahSMP folder (it comes from this thread http://foldingforum.org/viewtopic.php?f=46&t=4411&st=0&sk=t&sd=a)

9. Create a short cut to the above file on your desktop

10. Edit the short cut to include "-smp -deino -forceasm" at the end of TARGET

11. execute the short cut and answer the questions.

It should run at 100% now.

Trying this one myself...

using the installer.exe version for 32/64bit, it will only download 82 cores and run one, even with the SMP flag in my config file and doing it via advance options...

the deino was doing the same, would only download 82 cores and run on of them with the smp flag in it.

for a project they want help with, they sure could make it easier for people to actually help for the cause....

[science man]

for a project they want help with, they sure could make it easier for people to actually help for the cause....

Yea I will admit it is ashame that you have to figure out what to do to help the best you can if you know what I mean.

[Mr.Guvernment]

It does look like you didn't enter the right password for that user. Just run install again and pick repair.

i did this, and i know i entered the right user and pass and i get an error Denoi crap is starting..... denoi cant start... no error reported.

trying again. one more time before i say screw it.

[orion456]

the deino was doing the same, would only download 82 cores and run on of them with the smp flag in it.

Yes, I had the problem too. Deleting the folder, rebooting and reinstalling fixed it for me.

[Mr.Guvernment]

argh!!

i am doing this under server 2003, followed the guides to a T including the on eon F@H forums, same crap, keeps downloading the 82 core even with the SMP flag in the config


[settings]
username=MrGuvernment
team=32
passkey=
asknet=no
machineid=1
bigpackets=big
extra_parms=-smp 4

[http]
active=no
host=localhost
port=8080
usereg=no

[core]
priority=0
cpuusage=100
disableassembly=no
checkpoint=15
ignoredeadlines=no
nocpulock=0
addr=

[power]
battery=no

[clienttype]
memory=4093
type=0


i'm going to bed, screw this for now, i better things to do . it is nice when this works, because it works, then they make version expire and make the newer one 10x harder to install!

at least the GPU one works fine.



you're gonna have to delete the work directory too, since core82 won't run on smp. so you'll lose a couple hours work. have to get a new WU

Yes, I had the problem too. Deleting the folder, rebooting and reinstalling fixed it for me.

have done that, same crap over and over.

[orion456]

argh!!

i am doing this under server 2003, followed the guides to a T including the on eon F@H forums, same crap, keeps downloading the 82 core even with the SMP flag in the config


You need both the "-smp -deino" no 4 needed but the -deino is needed if you are using that version. Include -verbosity 9 as well to see details about what is happening.

[Mr.Guvernment]

add that in the config or on the shortcut?

[Adak]

Either should be fine, but I'd prefer the shortcut.

[orion456]

Shortcut is easier to edit.

[Mr.Guvernment]

the Demino service wont start for me, on either installed the .exe version or the zip file, i run the install.bat file (this is after uninstalling and rebooting) and the Denio setup starts and it gets to starting the service, and says it couldn't start

try starting it under services and says it started and then stopped......

[Mr.Guvernment]

So install the 32/64bit mphic version again


SO when i enter in -smp 4 - it downloads core 82!!


[settings]
username=MrGuvernment
team=32
passkey=
asknet=no
machineid=1
extra_parms=-smp 4

[http]
active=no
host=localhost
port=8080
usereg=no

[core]
addr=



[edit]

FAINLLLLY!!

delete the work folder, RE_ran the config, added -smp 4 to the confi in advance options - it still went - download core 82, so i said mmmmmmmmmmmmmmm


so i deleted queue.dat file and cleared my FAHlog.txr file and BAM!!! got an _a1 core!


So it seems the queue.dat file was storing some info to grab the 82 core for some reason.

So if you keep getting 82 files with the -smp 4 line - delete your queue.dat file!

[MARCI]

queue.dat is the actual work unit, so if that isn't deleted, it just re downloads the core to work for that unit

[ChasR]

The actual work unit results and working files are stored in the work folder. queue.dat can be thought of as the link index of the work folder. The queue.dat file contains WU information that determines which core is used so you must delete along with the work folder to affect a change.

[Mr.Guvernment]

see, that was something i never knew, in the past everyone just said delete the core and the work folder...

[science man]

Hey guys sorry to be a big pian but it happened again and what worked before isn't now. It was working but then when a WU expired it kept running into an error and shutting down so I uninstalled and redownload it, made the shortcut edited it at the end of the traget with a space and then -smp. and it says this


Note: Please read the license agreement (Folding@home-Win32-x86.exe -license). F
urther
use of this software requires that you have read and accepted this agreement.

4 cores detected
Unable to connect to 'FrankGreico-PC:8676',
sock error: generic socket failure, error stack:
MPIDU_Sock_post_connect(1228): unable to connect to FrankGreico-PC on port 8676,
exhausted all endpoints (errno -1)
MPIDU_Sock_post_connect(1275): unable to connect to FrankGreico-PC on port 8676,
No connection could be made because the target machine actively refused it. (er
rno 10061)
[20:21:50] Error running mpiexec. Please check your configuration and re-run in
stall.bat if necessary

and then closes. I even tried installing it first using the orignal exe file before going into the shortcut but still no luck.

Keep in mind that what I said I did worked before and now what's happening now makes it seem like a miricle that I ever got it to work correctly that one time I did.

[AlabamaCajun]

I've been running on 2 CPUs for a day or so finished the unit then complained about instability. I decided to figure it out once I had the other 2 CPUs in. I'm getting an Error right off the bat about having an internal error of some sort and can't continue. I ran OCCT all 16 cores for an hour and not an issue so I'm sure 4 Opterons and ECC RAM are good but still have a little doubt. I'm going to try what I've seen in this thread and may have to turn off the affinity switch but not sure.

[harlam357]

Which v6.22 install? Deino or MPICH?

I just installed the MPICH tonight and it's just like the old v5.91 client. Went off without a hitch. Nothing is really "different" except for the main .exe name.

[science man]

MPICH

[Mr.Guvernment]

you didnt enter the right user/password when it asked. delete and reinstall, make sure when it asks for the User and Password for your windows machine , enter it correctly.

[MARCI]

That is my thought as well, but, since it was working and suddenly stopped I don't know. Unless he tried to reinstall again after everything was working?

[science man]

It doesn't ask for the username and password of my computer only the username of my folding account and the update is that I almost got it running, but now it says this

[17:51:16]
[17:51:16] Loaded queue successfully.
[17:51:16]
[17:51:16] + Processing work unit
[17:51:16] Work type 78 not eligible for variable processors
[17:51:16] Core required: FahCore_78.exe
[17:51:16] Core found.
[17:51:16] Using generic mpiexec calls
[17:51:16] Working on queue slot 01 [August 25 17:51:16 UTC]
[17:51:16] + Working ...
[17:51:17]
[17:51:17] *------------------------------*
[17:51:17] Folding@Home Gromacs Core
[17:51:17] Version 1.90 (March 8, 2006)
[17:51:17]
[17:51:17] Preparing to commence simulation
[17:51:17] - Ensuring status. Please wait.
[17:51:17] - Looking at optimizations...
[17:51:17] - Working with standard loops on this execution.
[17:51:17] - Previous termination of core was improper.
[17:51:17] - Going to use sta- Expanded 1647603 -> 8381617 (Starting from checkp
oint)
[17:51:17] Protein: p2452_Fragment-Receptor HDS
[17:51:17]
[17:51:17] Writing local files
[17:51:17] E(Starting from checkpoint)
[17:51:17] Prot(Starting from checkpoint)
[17:51:17] Protein: p2452_Fragment-Receptor HDS
[17:51:17]
[17:51:17] Writing local files
[17:51:17] Size of work/wudata_01.bed not what saved.
[17:51:17]
[17:51:17] Folding@home Core Shutdown: FILE_IO_ERROR
[17:53:42] IO_ERROR

When what I did was installed it created the shortcut and eedited the traget with a space -smp. I restarted it just now and this it said this

[18:08:53] Loaded queue successfully.
[18:08:53]
[18:08:53] + Processing work unit
[18:08:53] Work type 78 not eligible for variable processors
[18:08:53] Core required: FahCore_78.exe
[18:08:53] Core found.
[18:08:53] Using generic mpiexec calls
[18:08:53] Working on queue slot 01 [August 25 18:08:53 UTC]
[18:08:53] + Working ...
[18:08:53]
[18:08:53] *------------------------------*
[18:08:53] Folding@Home Gromacs Core
[18:08:53] Version 1.90 (March 8, 2006)
[18:08:53]
[18:08:53] Preparing to commence simulation
[18:08:53] - Ensuring status. Please wait.
[18:08:53] - Looking at optimizations...
[18:08:53] - Working with standard loops on- Expanded 1647603 -> 8381617 (decomp
ressed 508.7 percent)
[18:08:53] - Starting from initial work packet
[18:08:53]
[18:08:53] Project: 2452 (Run 185, Clone 2, Gen 17)
[18:08:53]
[18:08:53] Assembly optimizations on if available.
[18:08:53] Entering M.D.
[18:08:53] Protein: p2452_Fragment-Receptor HDS
[18:08:53]
[18:08:53] WrProtein: p2452_FraExtra SSE boost OK.
[18:08:53] Writing local files
[18:08:53] Completed 0 out of 400000 steps (0%)
[18:08:53] Writing local files
[18:08:53] Completed 4000 out of 400000 steps (1%)
[18:08:53] wn: FILE_IO_ERROR
[18:08:55] etected. Shutting down core
[18:08:55]
[18:08:55] Folding@home Core Shutdown: CLIENT_DIED

[Mr.Guvernment]

uninstall and run install.bat first, it should ask you for your windows accounts like [DOMAIN\Account]:

AFTEr you run install.bat then you run the fah*.exe file and set up your account info

[ChasR]

You should be installing FAH from the account you are going to run it under in which case, install.bat defaults to the current account. You should be hitting enter at that point to accept the default and then enter your password for the current account. If you enter something other than the default for the account name, I doubt it's going to work. In any event, running foo will tell you if you have MPI set up correctly.

[science man]

you mean I should run command promopt and direct it to the folder where it is than run install..bat just like we had to do with 5.91 and click the exe file itself?

EDIT: trid that and there's no difference both through shortcut and through orignal exe file.

[harlam357]

Did install.bat end with...

"If you see this twice, MPI is working"
"If you see this twice, MPI is working"

[science man]

yep

[ChasR]

The last log wasn't from the smp client you're running regular Gromacs. Forget the -smp flag? What happens when you double click on the file named foo? Post the beginning of the log from a restart of the client.

[science man]

I unstalled and reinstalled and goit it to say MP1 is working, made the traget edited it with....oh u know, ran it and now it and it says it runs into a serious error and shuts down.

EDIT: I quit folding until the next version comes out because is one is a downgrade from v5.91 and is impossilbe it get working.