Guide: Setting up notfred in VMWare Player

[capreppy]

It's in the Horde Guides sticky. First post near the bottom

 

[Nessus85]

Wow, i was soooo close to dropping this method seeing as how I had to keep changing the affinitys and priorities for the clients. And then i stumble upon the post saying you have to have winafc 0.9.2. Well crap, i had .9. Thank you thank you thank you!

 

[Nessus85]

Ok, I'm still in love with this method but I have a question. I cant go two wu's in a row without some sort of error. The most common error that I am getting is this one.

Im currently running two gpu clients and two notfred vmwares. Any ideas as to what I screwed up?

 

[capreppy]

I'll need a little time to research this one. The Segment Fault Error I've seen before in Native Linux (not me specifically, but on the Linux SMP forums). I'll see what the error relates to and see if there is something that causing it.

I myself have never seen that error so I'm not sure how to resolve it.

 

[Nessus85]

Thanks for the fast response! I would appreciate any info you have on it. Here is the log according to Fahmon if it helps...

[09:07:58] Completed 32500 out of 250000 steps (13%)
[09:18:48] Completed 35000 out of 250000 steps (14%)
[09:29:37] Completed 37500 out of 250000 steps (15%)
[09:40:29] Completed 40000 out of 250000 steps (16%)
[09:51:19] Completed 42500 out of 250000 steps (17%)
[09:52:52]
[09:52:52] Folding@home Core Shutdown: INTERRUPTED
[09:52:56] CoreStatus = 66 (102)
[09:52:56] + Shutdown requested by user. Exiting.
Folding@Home Client Shutdown.

 

[capreppy]

Could you provide (assuming you are able to) the Run-Clone-Gen of the WU in question. It seems that there might be some bad WUs out in the wild and you may have gotten one of them.

An option is to start with a fresh install of the notfred instance and see if you get that same WU. By setting up a new instance, it creates a new identity and hopefully (keeping fingers crossed) you don't get that bad WU again

 

[Nessus85]

Thats actually what I have tried several times. It will usually work for one wu, then back to this.

Here is the entire Strand after it loads the content...

[07:29:29] Verifying core Core_a2.fah...
[07:29:29] Signature is VALID
[07:29:29]
[07:29:29] Trying to unzip core FahCore_a2.exe
[07:29:30] Decompressed FahCore_a2.exe (4341288 bytes) successfully
[07:29:30] + Core successfully engaged
[07:29:35]
[07:29:35] + Processing work unit
[07:29:35] At least 4 processors must be requested.Core required: FahCore_a2.exe
[07:29:35] Core found.
[07:29:35] Working on Unit 03 [May 1 07:29:35]
[07:29:35] + Working ...
[07:29:35]
[07:29:35] *------------------------------*
[07:29:35] Folding@Home Gromacs SMP Core
[07:29:35] Version 2.07 (Sun Apr 19 14:51:09 PDT 2009)
[07:29:35]
[07:29:35] Preparing to commence simulation
[07:29:35] - Ensuring status. Please wait.
[07:29:45] - Assembly optimizations manually forced on.
[07:29:45] - Not checking prior termination.
[07:29:46] - Expanded 4843045 -> 23991465 (decompressed 495.3 percent)
[07:29:47] Called DecompressByteArray: compressed_data_size=4843045 data_size=23991465, decompressed_data_size=23991465 diff=0
[07:29:47] - Digital signature verified
[07:29:47]
[07:29:47] Project: 2669 (Run 4, Clone 114, Gen 56)
[07:29:47]
[07:29:47] Assembly optimizations on if available.
[07:29:47] Entering M.D.
[07:29:53] Using Gromacs checkpoints
[07:29:55] Multi-core optimizations on
[07:29:57] Resuming from checkpoint
[07:29:58] Verified work/wudata_03.log
[07:29:58] Verified work/wudata_03.trr
[07:29:58] Verified work/wudata_03.xtc
[07:29:58] Verified work/wudata_03.edr
[07:30:04] Completed 10010 out of 250000 steps (4%)
[07:41:22] Completed 12500 out of 250000 steps (5%)
[07:52:12] Completed 15000 out of 250000 steps (6%)
[08:03:01] Completed 17500 out of 250000 steps (7%)
[08:13:51] Completed 20000 out of 250000 steps (8%)
[08:24:41] Completed 22500 out of 250000 steps (9%)
[08:35:32] Completed 25000 out of 250000 steps (10%)
[08:46:20] Completed 27500 out of 250000 steps (11%)
[08:57:09] Completed 30000 out of 250000 steps (12%)
[09:07:58] Completed 32500 out of 250000 steps (13%)
[09:18:48] Completed 35000 out of 250000 steps (14%)
[09:29:37] Completed 37500 out of 250000 steps (15%)
[09:40:29] Completed 40000 out of 250000 steps (16%)
[09:51:19] Completed 42500 out of 250000 steps (17%)
[09:52:52]
[09:52:52] Folding@home Core Shutdown: INTERRUPTED
[09:52:56] CoreStatus = 66 (102)
[09:52:56] + Shutdown requested by user. Exiting.
Folding@Home Client Shutdown.

 

[Nessus85]

I just reset that vmware. Il let you know if I get another error

*crosses fingers

 

[jfb9301]

I think I figgured out some of my problem with WinAFC.

"Run as Administrator"

after I did that things seemed to run like I thought they should

 

[Nessus85]

I just instaled (yes I know its an older version), Windows 7 (7057) last night and set up two vmwares and two gpu clients. It turns out that I dont even need to setup winafc in 7 as it recognizes the priorities without fault automatically so my gpu's get all the power they can dream of. Also, the vmwares went all night without a hiccup. Im thinking the errors I had before were probably memory errors and now that I have 64bit, they are gone.

 

[capreppy]

This is GREAT news!!!

 

[Nessus85]

Mini Windows 7 Update -

The Vmwares are going just fine

The priorities on the gpu clients however did not stick like I though they were. When a second client loaded, the settings had changed. Per Vaulter98c's advice, I downloaded the 64bit winafc(didnt realize i had the 32bit version, doh!) and everything is great now.

 

[brokenarrow03]

Ok, so Im setting this up and I am at the WinAFC64 part and a little cofused as to what PAIR0 and PAIR1 need to be replaced with. If anyone has a Q6600, could you post what your WinAFC "Application Profile" looks like

thanks

 

[Nessus85]

Pair0+Pair1 is what you want like I have below. Are you running one gpu session or two?

I am running two and this is how mine is setup...

*\vmware-vmx.exe := PAIR0+PAIR1 [assign=2,priority=Idle]
*\Folding@home-gpu\FahCore_11.exe := PAIR0+PAIR1 [priority=High]
*\Folding@home-gpu\FahCore_14.exe := PAIR0+PAIR1 [priority=High]
*\Folding@home-gpu2\FahCore_11.exe := PAIR0+PAIR1 [priority=High]
*\Folding@home-gpu2\FahCore_14.exe := PAIR0+PAIR1 [priority=High]

If you are running one gpu session then get rid of the last two lines. The only part of yours that might be different is where mine says "*\Folding@home-gpu2" or "*\Folding@home-gpu," yours should say the name of the folder where you have installed your gpu client.

Now that I think about it, are the last two lines even neccasary?

 

[84gthatch]

hey brokenarrow how are you setting up your pairs together? pair0+pair1 should work for that

*\vmware-vmx.exe := PAIR0+PAIR1+PAIR2+PAIR3 [assign=2,priority=Idle]
*\FahCore_11.exe := PAIR0+PAIR1+PAIR2+PAIR3 [priority=High]
*\FahCore_14.exe := PAIR0+PAIR1+PAIR2+PAIR3 [priority=High]
*\vlc.exe := PAIR0+PAIR1+PAIR2+PAIR3 [priority=High]

Heres how i have mine set up currently thats for the i7 for you it would probably be like this

*\vmware-vmx.exe := PAIR0+PAIR1 [assign=2,priority=Idle]
*\FahCore_11.exe := PAIR0+PAIR1 [priority=High]
*\FahCore_14.exe := PAIR0+PAIR1 [priority=High]

I tossed the vlc line in because I couldn't get a video to play to save my life but when i added that tada no more issues

 

[84gthatch]

hey puppeteer from what i can tell having the line for each gpu client is not hurting anything but is kind of redundant.

 

[Nessus85]

Yeah, thats what I realized when I typed it out. thanx for confirming

 

[84gthatch]

lol np win afc took me awhile to get figured out and cap was super patient with me so i'm glad i can pass that on!

 

[sirmonkey1985]

well i finally got everything working..

also im not using the win AFC since you really dont need it.. vmware player only allows 2 cores to be used per instance anyways.. so you can either leave it on default or go into windows task manager and manually set it that way..

 

[Nessus85]

On mine if i dont use winafc, vmware will reset its affinities each time there is a new session(wu) in 64bit 7.

 

[sirmonkey1985]

On mine if i dont use winafc, vmware will reset its affinities each time there is a new session(wu) in 64bit 7.

so far i havent completed any yet.. so we will see later tonight..

 

[84gthatch]

well you might not need winAFC what does matter is making sure the core priorities stay the right way.

 

[capreppy]

Well I am very glad to see that this has become a team effort. Sorry to the delayed response. Church & Celebrating my in-laws 30th anniversary.

A couple of comments: If running, just VMs, you don't really need WinAFC. The VMs priority is Normal which is fine. WinAFC is used specifically to elevate the GPU2 clients whose priority is idle. This is bad in that the GPU2 client will starve for CPU cycles against VMs.

I'd like to get Win7 up and running so I can see what others are seeing. Seems to be a better Vista than Vista is

 

[sirmonkey1985]

Well I am very glad to see that this has become a team effort. Sorry to the delayed response. Church & Celebrating my in-laws 30th anniversary.

A couple of comments: If running, just VMs, you don't really need WinAFC. The VMs priority is Normal which is fine. WinAFC is used specifically to elevate the GPU2 clients whose priority is idle. This is bad in that the GPU2 client will starve for CPU cycles against VMs.

I'd like to get Win7 up and running so I can see what others are seeing. Seems to be a better Vista than Vista is

yeah i found that out the hard way.. but fixed it..

 

[brokenarrow03]

If you are running one gpu session then get rid of the last two lines. The only part of yours that might be different is where mine says "*\Folding@home-gpu2" or "*\Folding@home-gpu," yours should say the name of the folder where you have installed your gpu client.

excellent, thanks for the help, if its worth anything I currently have 1 gpu running, but I do plan to get up to 4 I hope...

 

[Nessus85]

excellent, thanks for the help, if its worth anything I currently have 1 gpu running, but I do plan to get up to 4 I hope...

No problem man! After all, there's no "I" in "FOLDING!"

err....wait...

 

[brokenarrow03]

ok last question...how long do the steps take for each SMP client? FAHmon says that both of mine have sitting at 0% for a solid 10-15 minutes, both status says running, is this normal?

 

[capreppy]

Core i7 920 @ stock (waiting on RMA MoBo to put WC back on) is approx 16 to 24 mins per step

Intel Q6600 @ 3.2 is approx 10 mins per step
Intel QX9300 @ 2.53 is approx 13 mins per step
AMD 940BE @ 3.0 is approx 10 mins per step

I don't have any C2D's, but they should be pretty equivalent. You should be able to extrapolate based on your CPU type & CPU speed.

 

[Nessus85]

My q9550 oc'd to 3.7ghz takes 8:23 per step

- stock is 11 min i believe

 

[sirmonkey1985]

my phenom II 940 @ 3.5ghz takes between 9-10 minutes per step.. but its also running at 1.6ghz HT so its a friggin neutered 3.5ghz..

 

[brokenarrow03]

so my GTX 260 was folding about 7kppd before the vm's were installed, any reason why it is only at 3300ppd now? same WU and everything...

 

[Nessus85]

Is winafc running? When the console comes up after loading the cpu cores, does anything inside it turn yellow? If so, then you have a problem with the way that you have entered in the core information.

 

[84gthatch]

also make sure to turn off test mode in winAFC that makes a big difference to

 

[flogge]

I have an ATI HD4890 running at about 900 Mhz and then a single VM on my 720 triple core at 3.5Ghz(no change to the HT speed, it is at stock). I am getting about 13.5 minutes per step. Windows Task Manager says I am running all 3 cores at 100%.

I am seeing other people with 940BE at 3.5 Ghz getting better 9-10 minutes per step and I am taking 50% more time to run 2 of my 3 cores in a VM. I left a core unused intentionally because I know how needy the ATI GPU client can be for CPU cycles. It almost looks like the GPU is using more than 1 core to keep happy and is robbing cycles from the VM. I'll do some more testing and try to get some more info on the table.

 

[sirmonkey1985]

yeah id venture to say it is.. since the 720 on 2 cores should out perform the 940 clock for clock because of the higher HT.. make sure the gpu client is set to a single core.. and that you specify which cores the vm is using.. otherwise winblows will constantly shift programs back and forth between cores.. since thats what its designed to do..

 

[sirmonkey1985]

so my GTX 260 was folding about 7kppd before the vm's were installed, any reason why it is only at 3300ppd now? same WU and everything...

its because by default the vmware clients run on normal and the gpu2 client is set to low.. so the gpu client is being starved.. use winAFC to force set the gpu client to run on normal..

 

[sirmonkey1985]

arg.. this winAFC craps giving me a friggin headache.. theres absolutely 0 examples of setting this pos software up for amd processors..


ok.. disaster is over thanks to william..

 

[brokenarrow03]

Im pretty sure that Ive got WinAFC set up, and not in test mode, there are no yellow messages in the console. I will post what is there when I go home for lunch

 

[sirmonkey1985]

Im pretty sure that Ive got WinAFC set up, and not in test mode, there are no yellow messages in the console. I will post what is there when I go home for lunch

do you have the FAHcore_11 and fahcore_14 set to a higher priority then the smp client?


btw capreppy can you put this in the guide please for amd users..

*\vmware-vmx.exe := CPU0+CPU1+CPU2+CPU3 [assign=2,priority=Idle]
*\FahCore_11.exe := QUAD0 [priority=High]
*\FahCore_14.exe := QUAD0 [priority=High]

thats how winAFC needs to be set for the gpu client and the smp clients to run correctly on an amd quad core..

only way i found that out was from William..

 

[brokenarrow03]

hmm the only part of that I don't have is the QUAD0 part, I think it says PAIR0+PAIR1 still...is that because its an Intel Q6600?