• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Just started folding again...could use a hand.

coffaholic

Gawd
Joined
Dec 24, 2011
Messages
565
Hi Everyone,

I just started folding again since the weather is cooling down and wanted to get my CPU and GPU going for your big push to pass EVGA.

I was looking for the easiest way to get things set up and I found mention of FAH GPU Tracker V2 so I downloaded it and set things up. Upon starting both clients, I noticed that my GPU was pegged at 99-100%, but my CPU wouldn't budge over 50%.

Here are my settings:


What it looks like when everything is running:


My SMP FAHlog:
Code:
# Windows SMP Console Edition #################################################
###############################################################################

                       Folding@Home Client Version 6.34

                          http://folding.stanford.edu

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

Launch directory: D:\apps\fahgputrackerv2\FAH GPU Tracker V2\SMP
Executable: D:\apps\fahgputrackerv2\FAH GPU Tracker V2\FAH.exe
Arguments: -oneunit -forceasm -smp -advmethods -verbosity 9 

[23:24:22] - Ask before connecting: No
[23:24:22] - User name: coffaholic (Team 33)
[23:24:22] - User ID: 1482DDF2059B4D77
[23:24:22] - Machine ID: 2
[23:24:22] 
[23:24:22] Work directory not found. Creating...
[23:24:22] Could not open work queue, generating new queue...
[23:24:22] - Preparing to get new work unit...
[23:24:22] - Autosending finished units... [September 19 23:24:22 UTC]
[23:24:22] Cleaning up work directory
[23:24:22] Trying to send all finished work units
[23:24:22] + No unsent completed units remaining.
[23:24:22] - Autosend completed
[23:24:22] + Attempting to get work packet
[23:24:22] Passkey found
[23:24:22] - Will indicate memory of 16288 MB
[23:24:22] - Detect CPU. Vendor: GenuineIntel, Family: 6, Model: 10, Stepping: 7
[23:24:22] - Connecting to assignment server
[23:24:22] Connecting to http://assign.stanford.edu:8080/
[23:24:22] Posted data.
[23:24:22] Initial: 8F80; - Successful: assigned to (128.143.199.96).
[23:24:22] + News From Folding@Home: Welcome to Folding@Home
[23:24:22] Loaded queue successfully.
[23:24:22] Sent data
[23:24:22] Connecting to http://128.143.199.96:8080/
[23:24:23] Posted data.
[23:24:23] Initial: 0000; - Receiving payload (expected size: 1764192)
[23:24:27] - Downloaded at ~430 kB/s
[23:24:27] - Averaged speed for that direction ~430 kB/s
[23:24:27] + Received work.
[23:24:27] + Closed connections
[23:24:27] 
[23:24:27] + Processing work unit
[23:24:27] Core required: FahCore_a3.exe
[23:24:27] Core found.
[23:24:27] Working on queue slot 01 [September 19 23:24:27 UTC]
[23:24:27] + Working ...
[23:24:27] - Calling '.\FahCore_a3.exe -dir work/ -nice 19 -suffix 01 -np 8 -nocpulock -checkpoint 3 -forceasm -verbose -lifeline 4748 -version 634'

[23:24:28] 
[23:24:28] *------------------------------*
[23:24:28] Folding@Home Gromacs SMP Core
[23:24:28] Version 2.27 (Dec. 15, 2010)
[23:24:28] 
[23:24:28] Preparing to commence simulation
[23:24:28] - Assembly optimizations manually forced on.
[23:24:28] - Not checking prior termination.
[23:24:28] - Expanded 1763680 -> 1944296 (decompressed 110.2 percent)
[23:24:28] Called DecompressByteArray: compressed_data_size=1763680 data_size=1944296, decompressed_data_size=1944296 diff=0
[23:24:28] - Digital signature verified
[23:24:28] 
[23:24:28] Project: 6947 (Run 0, Clone 193, Gen 46)
[23:24:28] 
[23:24:28] Assembly optimizations on if available.
[23:24:28] Entering M.D.
[23:24:34] Mapping NT from 8 to 8 
[23:24:34] Completed 0 out of 500000 steps  (0%)
[23:30:36] Completed 5000 out of 500000 steps  (1%)
[23:36:26] Completed 10000 out of 500000 steps  (2%)
[23:41:50] Completed 15000 out of 500000 steps  (3%)
[23:47:14] Completed 20000 out of 500000 steps  (4%)
[23:52:43] Completed 25000 out of 500000 steps  (5%)
[23:58:03] Completed 30000 out of 500000 steps  (6%)
[00:03:38] Completed 35000 out of 500000 steps  (7%)
[00:10:36] Completed 40000 out of 500000 steps  (8%)
[00:16:05] Completed 45000 out of 500000 steps  (9%)
[00:21:29] Completed 50000 out of 500000 steps  (10%)
[00:26:42] Completed 55000 out of 500000 steps  (11%)

I cant seem to figure out why it wont use 100% of my CPU. I can load it to 100% using prime or any other stress test.

I've tried updating FAH Clients from the FAH GPU Tracker's Setup menu, manually replacing the FAH.exe with the most recent one, changing the SMP CPU settings in configuration to "Use specified number of CPU cores" and set it to 8, and running just the SMP client without the GPU client in case they weren't playing nice together.

Any ideas would be appreciated.

Thanks.
 
Last edited:
in the cpu cores to use put -1 once or put the cpu cores you want to use.

Haven't used the gpu client in a while, check out the new v7, it does gpu in it as well as cpu.
 
So I tried popping the -1 figure in the number of cores to use, as well as 8 to try to use all 8 threads on my CPU but both failed to load the CPU to 100%. I also tried downloading and configuring the V7 client which left me in the same place.

Since every stress test that I throw at my CPU will load it to 100%, but the folding client will not, could it be possible that the work units I've been getting are for quads? (don't know if they segregate WU's to specific CPU core values) The WU's I've gotten since I started folding have been 6940, 6947, 6973, 6980, 6993, 8064, 8065, 8066, 8067 with a range of about 11k-16k PPD. I assume they vary because of the WU and the fact I have had the GPU client running at the same time on occasion.
 
No, those are all SMP units and will fold on 2 to whatever threads... in fact some of those units are new A4 projects that are also eligible for single threaded systems. There aren't any projects that are limited to just for threads. If there were, the assignment server just wouldn't assign them to you.

Code:
[23:24:27] - Calling '.\FahCore_a3.exe -dir work/ -nice 19 -suffix 01 -np 8 -nocpulock -checkpoint 3 -forceasm -verbose -lifeline 4748 -version 634'

The -np 8 flag in the above log entry is the client telling the core to use all 8 threads. Umm..

*scratches head*
 
Odd. When you switch GPU folding off, do you get 100% load on the CPU?
 
additionally, are you using the modified GPU Tracker program to work with the 6xx series GPUs correctly?
 
Negative. It will still only use 4 cores without the GPU client running.

At first I thought it was odd that it was just using every second core, as if i had turned off hyperthreading, but I confirmed first by running stress tests, then by double checking in the BIOS and everything is set to use 8 threads.

I've given myself a bit of a headache thinking about this for too long. I just wanted to see if i could fix it before going down the "format to see if that fixes it" route.

edit
additionally, are you using the modified GPU Tracker program to work with the 6xx series GPUs correctly?

Yes, that is the one I downloaded. I believe I found the link in a thread around here. It does indeed say "kepler support enabled (-betateam)" when you start it up
 
Right click on the FahCore and click Properties. "Threads" tab is of most interest...
 
I see you have a lot of things running on the box, can you kill each and every unrelated to FAH process, including little gadgets, and see if anything changes? It looks like something is messing with affinity big time. The core is running 8 worker threads and 4 private threads like it should but each are using very little CPU. *scratches head*
 
Additionally, right click on FahCore_a3.exe in the task manager and click set affinity to make sure all cores are selected by default.

2r3h93d.jpg
 
Well, I nuked every process I figured I could and the problem still occurs. Looks like its time to format. Come to think of it, it has been quite some time since the last one, but everything has been running great since this happened...I really had no reason to consider it.

I need to back up a few things so I probably wont get to it till late in the weekend so if you come up with anything else I may be able to try in the meantime I would be happy to.

edit
Additionally, right click on FahCore_a3.exe in the task manager and click set affinity to make sure all cores are selected by default.

Just checked. All cores are selected.
 
For giggles, all of your Windows updates are up to date? Otherwise, I really don't know. It's not a FAH problem. 8 threads have been initiated by the core. Something on the Windows CPU scheduling end isn't working correctly for some reason. I really don't know.
 
Before you format I'd suggest doing two more things:
EDIT:0. Bumping FahCore priority in task manager (one or two notches up)

1. (For kicks) selecting only odd-numbered cores in there and checking if they get loaded;
-- e.g. we want to see 0-1-0-1-0-1-0-1 instead of current 1-0-1-0-1-0-1-0.

2. Actually installing all available system updates... :-o

EDIT: 3. Not really warranted ... but... you could see if BIOS update changes anything.
 
Last edited:
Actually, there are a few recent updates I haven't installed yet. Let me give those a go.

I will also see if i can load the other cores via affinity.
 
Also, via taskman, how about right clicking on the core and manually raising the priority?
 
Ok, progress I think, or a place to go from here anyways.

Before installing the updates I played with the affinity, starting with 0-1-0-1-0-1-0-1, which flipped the cores that were working in TaskMan. Then I tried turning more on, starting with 7, then, 5, then 3....I notice the CPU usage increase with every enabled core, until I enable core 1 and it drops back to 50%. So right now, with 7 cores enabled I'm folding at 87%. Just waiting to finish a few frames to see the tpf decrease to confirm.

I think the next place to explore is my overclock. Seems as mild as it is, it may be time for a tweak. It was prime stable way back when, but maybe core 1 can't hack it anymore. I will lower things down and see if that changes the situation.
 
EDIT: nevermind, heresy

How about doing symmetric exercise? 1-3-5 (and leaving 7 out) ?
 
Heh, that's what I did. TaskMan in my first pictures show core 1,3,5,7 working. I flipped the affinith to 2,4,6,8 and it worked. Then I enabled 7,5,3 one by one and it got me to 87% usage, but enabling core 1 to finish it off brings me back to 50%.

Hope that makes sense, I am a bit tired.
 
Sorry, I misread your other post.

How about doing same exercise (7 cores total) but using 1-3-5 sequence (and leaving 7 out)?
I wonder if that also gives 87%...
 
Yes, it seems as though as long as I leave 1 core out (doesn't matter which one) it will run at 87%. As soon as I add the 8th, it slows to 50%.
 
Hmm, this is very odd but glad we are making progress. I guess looking at the overclock next would be a good thing to do.
 
Well, I dialed it back 100mhz and now FAH pegs the CPU at 100% so I guess over time my chip decided to get moody. Gonna have to play around with it a bit more but its safe to say that a now unstable overclock caused my issue.

Thanks for your help, much appreciated ;)
 
Well, I dialed it back 100mhz and now FAH pegs the CPU at 100% so I guess over time my chip decided to get moody. Gonna have to play around with it a bit more but its safe to say that a now unstable overclock caused my issue.

Thanks for your help, much appreciated ;)

some tdp settings in the bios perhaps...
 
Back
Top