having issues with my new GTX980

Message boards : Number crunching : having issues with my new GTX980
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile cliff west

Send message
Joined: 7 May 01
Posts: 211
Credit: 16,180,728
RAC: 15
United States
Message 1661326 - Posted: 4 Apr 2015, 5:52:03 UTC

Put my 980 in tonight. wanted to maximize the card but so far it is only working on one work unit.

Help please.


4/3/2015 10:46:58 PM | | cc_config.xml not found - using defaults
4/3/2015 10:46:58 PM | | Starting BOINC client version 7.4.36 for windows_x86_64
4/3/2015 10:46:58 PM | | log flags: file_xfer, sched_ops, task
4/3/2015 10:46:58 PM | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.8
4/3/2015 10:46:58 PM | | Data directory: C:\ProgramData\BOINC
4/3/2015 10:46:58 PM | | Running under account Cliff
4/3/2015 10:46:58 PM | | CUDA: NVIDIA GPU 0: GeForce GTX 980 (driver version 347.88, CUDA version 7.0, compute capability 5.2, 4096MB, 3876MB available, 5599 GFLOPS peak)
4/3/2015 10:46:58 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 980 (driver version 347.88, device version OpenCL 1.1 CUDA, 4096MB, 3876MB available, 5599 GFLOPS peak)
4/3/2015 10:46:58 PM | | Host name: Cliff-PC
4/3/2015 10:46:58 PM | | Processor: 6 AuthenticAMD AMD Phenom(tm) II X6 1100T Processor [Family 16 Model 10 Stepping 0]
4/3/2015 10:46:58 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni cx16 popcnt syscall nx lm svm sse4a osvw ibs skinit wdt page1gb rdtscp 3dnowext 3dnow
4/3/2015 10:46:58 PM | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
4/3/2015 10:46:58 PM | | Memory: 8.00 GB physical, 16.00 GB virtual
4/3/2015 10:46:58 PM | | Disk: 465.66 GB total, 216.04 GB free
4/3/2015 10:46:58 PM | | Local time is UTC -7 hours
4/3/2015 10:46:58 PM | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 9686752; resource share 100
4/3/2015 10:46:58 PM | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 552848; resource share 100
4/3/2015 10:46:58 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 5728999; resource share 100
4/3/2015 10:46:58 PM | SETI@home | General prefs: from SETI@home (last modified 31-Aug-2014 23:52:56)
4/3/2015 10:46:58 PM | SETI@home | Computer location: home
4/3/2015 10:46:58 PM | | General prefs: using separate prefs for home
4/3/2015 10:46:58 PM | | Reading preferences override file
4/3/2015 10:46:58 PM | | Preferences:
4/3/2015 10:46:58 PM | | max memory usage when active: 8109.27MB
4/3/2015 10:46:58 PM | | max memory usage when idle: 8109.27MB
4/3/2015 10:46:58 PM | | max disk usage: 100.00GB
4/3/2015 10:46:58 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
4/3/2015 10:46:58 PM | | Not using a proxy
ID: 1661326 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1661336 - Posted: 4 Apr 2015, 6:09:53 UTC - in response to Message 1661326.  
Last modified: 4 Apr 2015, 6:13:41 UTC

Current status quo with MB provides a growing list of things you could address (each on their own terms, and many will help with appropriate tools/advice for each):

Broadly each optional 'at own risk' before OCing:
- Force p2 power state memory clock to full rate
- up the fan speeds/fan-curve to disengage turbo boost temperature limits
- use Lunatics to fix the application to the Cuda50 build
- use the mbcuda.cfg to raise process priority and pulsefind settings (abovenormal,16,400)
- check system DPC latencies while at full crunch (which will say something about quality of other drivers in the system that can interfere)
- increase instances from 1 to probably 3

Then for OCing:
- familiarise with an artefact scanner like OCCT
- practice dropping to factoiry settings (except for raised fan curve) then raising one setting at a time until artefacts are produced, then backing off 2 or more 'notches' (The size of a 'notch' is determined by granularity of application sliders or your own 'feel')
- possible core voltage increases & OC/hardware-limitations of the feeding system can be considered, rinsing and repeating artefact scans.

did I miss anything anyone ? AP has other totally different settings of course, which I'm not familiar with.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1661336 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1661338 - Posted: 4 Apr 2015, 6:13:05 UTC - in response to Message 1661336.  
Last modified: 4 Apr 2015, 6:13:33 UTC

The response to your question in the "GTX 970 about as fast as a 670 for crunching" thread, from Keith Myers

Simple, create a text file called app_config.xml with Notepad and put it into the Seti directory. The contents would have something like this:



<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>.10</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v6</name>
<gpu_versions>
<gpu_usage>.5</gpu_usage>
<cpu_usage>.50</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v7</name>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>0.50</cpu_usage>
</gpu_versions>
</app>
</app_config>

But in your case, I would reduce the GPU tasks to two per card with a 0.5 entry for gpu_usage.

Grant
Darwin NT
ID: 1661338 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1661343 - Posted: 4 Apr 2015, 6:19:37 UTC - in response to Message 1661338.  

Looking at your tasks, you're running CUDA42, where as you want CUDA50 for Maxwell GPUs.
Grant
Darwin NT
ID: 1661343 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1661345 - Posted: 4 Apr 2015, 6:23:22 UTC - in response to Message 1661343.  

Looking at your tasks, you're running CUDA42, where as you want CUDA50 for Maxwell GPUs.

If I am correct, it would appear he is running stock apps? So he would not have that choice unless he runs the Lunatics installer?
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1661345 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1661348 - Posted: 4 Apr 2015, 6:40:10 UTC - in response to Message 1661345.  

Looking at your tasks, you're running CUDA42, where as you want CUDA50 for Maxwell GPUs.

If I am correct, it would appear he is running stock apps? So he would not have that choice unless he runs the Lunatics installer?

The stock app is meant to try each application till it finds the fastest one, but it appears to be running only CUDA42, although it was running CUDA50 with his old card.
Grant
Darwin NT
ID: 1661348 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1661349 - Posted: 4 Apr 2015, 6:47:56 UTC - in response to Message 1661348.  

Looking at your tasks, you're running CUDA42, where as you want CUDA50 for Maxwell GPUs.

If I am correct, it would appear he is running stock apps? So he would not have that choice unless he runs the Lunatics installer?

The stock app is meant to try each application till it finds the fastest one, but it appears to be running only CUDA42, although it was running CUDA50 with his old card.

I understand that...
Sooooooo.....Either he waits and hopes the servers test 50 on the new card, or runs the Lunatics installer and can then make the choice himself. Correct?
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1661349 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1661359 - Posted: 4 Apr 2015, 7:01:19 UTC

Once he Lunatics fine offering is running the scope for optimising the number of simultaneous tasks arrives.
First step is getting the Lunatics installer, the main source is:
http://lunatics.kwsn.net/
which has a few mirrors:
http://www.arkayn.us/forum/index.php
http://mikesworldnet.de/download.html

You need to choose the installer that matches your operating system, then select the appropriate options - DO NOT try to run the wrong GPU types because they will not work.

Once you've got the optimised application running its time to try running tow or three simultaneous tasks.

You need to create a TEXT file called app_config.xml Use a text editor like Notepad and make sure you save it as a plain text file, and NOT an xml file as this has caused problems.
And this is what the file needs to contain:
<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>.10</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v7</name>
<gpu_versions>
<gpu_usage>.5</gpu_usage>
<cpu_usage>0.50</cpu_usage>
</gpu_versions>
</app>
</app_config>


You will also need to free up one or more CPU cores - this is done from the BOINC manager, advanced view/tools/computer usage/computer usage = x%
where x is the percentage of CPU cores you want used by BOINC CPU tasks (e.g. For a four core system, running 3 cores for BOINC, x=75)

Next time BOINC reads the options files (start up, or you force it to read the options) magic will happen.

A little explanation of what the lines do:
<gpu_usage>.33</gpu_usage>
causes three tasks to run on the GPU
<gpu_usage>.5</gpu_usage>
causes two tasks to run on the GPU

<cpu_usage>.10</cpu_usage>
allows 0.1 of the "spare" core you liberated a few minutes ago to be used by these tasks
<cpu_usage>.50</cpu_usage>
allows 0.5 of the "spare" core you liberated a few minutes ago to be used by these tasks. (This is why I only run 2 Astopulse tasks, if I run more I find I have problems with the feeding CPU getting tied in knots.


A couple of things. First don't be greedy and try to fill the GPU memory with tasks, this is totally counter productive, and will lead to a frozen system or errors, neither of which is a good place to be. And second, there is no equivalent to this way of running multiple tasks on a single CPU core - remember the average GPU has hundred of very simple processors...
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1661359 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1661436 - Posted: 4 Apr 2015, 13:23:36 UTC - in response to Message 1661359.  

Well I'm late to this party.

Ok,

Yes.

Install the lunatics choosing the one for which ever version of windows you use.

Next install the the 5.x of cuda

The next is where I and rob differ. I DO save it as a XML file. Not sure why he doesn't. But you just can't put xml at the end of the name. You have to change how you save the file.

Let me explain. (also you are using an AMD chip so the setting in the file need to be change just a bit as what they listed is for an INTEL chip...Not and AMD chip) {I can hear it now, It doesn't MATTER..Yes it does}

Open Notepad and copy the all the following.

<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>.33</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v7</name>
<gpu_versions>
<gpu_usage>.33</gpu_usage>
<cpu_usage>0.33</cpu_usage>
</gpu_versions>
</app>
</app_config>

Ok, now to save it. click on file and click Save As

This will open up a new box.

In file name type in app_config.xml

This will cause the computer to ask you if you REALLY want to use it, yes you do.

Just below that is the "Save as type" Click on it and change it to All Files (*.*)

Encoding is the last part at the bottom. Make sure it says ANSI

now you are ready to save

Save it. Next you need to move this file to where it goes.

If you have ever modified seti before then you probably know where the Setiathome folder is.

If you haven't then we need to get you to there.

It should be located in the following unless you specified a different location

Local disk (C:) > ProgramData > Boinc > project > setiathome.berkeley.edu

each of the > means opening a folder in those places. If you don't see ProgramData then it is probably still hidden on your computer.

So we need to Unhide those folders.

1 Open Folder Options by clicking the Start button , clicking Control Panel, clicking Appearance and Personalization, and then clicking Folder Options.

2 Click the View tab.

3 Under Advanced settings, click Show hidden files and folders, and then click OK.

Now you should be able to see ProgramData.

I only run work on the GPU and not the CPU that way I don't have to worry about starving my GPU. IF you decide to run CPU work then keep an eye on it.

Once you get to this point you should start to see work on the GPU.

After that it become optimizing the program like Jason said.

First would be installing the commmand line for Astropulse. The next would be modifying the mbcuda.cfg for the multibeams.

I think for now, let just get it up and running multiple instance first. After that we can proceed with these other modifications.


Zalster
ID: 1661436 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1661477 - Posted: 4 Apr 2015, 15:23:28 UTC - in response to Message 1661336.  


- check system DPC latencies while at full crunch (which will say something about quality of other drivers in the system that can interfere)

did I miss anything anyone ? AP has other totally different settings of course, which I'm not familiar with.



Pretty succinct summarizing, Jason. The note about DPC latencies got me thinking. I do have the DPCLAT utility sitting in my utilities folder. Jason, what do you think is the maximum or typical latency you can get away with when crunching MB? Probably subjective but I was wondering what you typically see?

Cheers, Keith


P.S. I have AP setting suggestions when the OP gets around to it.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1661477 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1661485 - Posted: 4 Apr 2015, 16:17:10 UTC - in response to Message 1661477.  
Last modified: 4 Apr 2015, 16:33:40 UTC


- check system DPC latencies while at full crunch (which will say something about quality of other drivers in the system that can interfere)

did I miss anything anyone ? AP has other totally different settings of course, which I'm not familiar with.



Pretty succinct summarizing, Jason. The note about DPC latencies got me thinking. I do have the DPCLAT utility sitting in my utilities folder. Jason, what do you think is the maximum or typical latency you can get away with when crunching MB? Probably subjective but I was wondering what you typically see?

Cheers, Keith


P.S. I have AP setting suggestions when the OP gets around to it.


I like using my main dev system, a crappy old COre2Duo, as a baseline for that kindof thing. That's because while well maintained, over the years chipset and (wifi) network drivers have been problematic/finicky. The simplest tool for a general overview / health check is DPC Latency checker, which gives me this at full crunch while streaming a video:


That's after a lot of hair pulling some years back, which led to using customised wifi drivers and forcing Intel chipset driver updates.

If I get periodic red spikes, the I use LatencyMon to isolate the specific driver/hardware involved. I'm told it's normal for such tools to report an extra 1 millisecond offset in Windows 8/8.1, so I suspect it's the consistency of the latencies that's more important than the exact values on a given system.

Also I have other systems that just always indicated good quality drivers etc out of the box, so some systems require more finagling than others.

[Edit:] as an extra note, I just find characterising the driver, hardware & configuration quality this way tends to rule out or isolate a whole swathe of possible issues at hardware, firmware, OS, driver, and application level.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1661485 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1661507 - Posted: 4 Apr 2015, 18:05:13 UTC - in response to Message 1661485.  

That's the tool I was talking about. I see from your screen shot that mine is similar but with lesser latency primarily due to the difference in age of the hardware. I'd forgotten about this utility and will give it a try and see if I can catch the system in one of its 'siesta' moments.

Keith
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1661507 · Report as offensive
Profile cliff west

Send message
Joined: 7 May 01
Posts: 211
Credit: 16,180,728
RAC: 15
United States
Message 1661792 - Posted: 5 Apr 2015, 18:30:16 UTC - in response to Message 1661436.  

Zalster, that was perfect!!! thanks.. system is starting to work. now just have to work out a way to it to do the same for the other two projects I'm working on.
ID: 1661792 · Report as offensive

Message boards : Number crunching : having issues with my new GTX980


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.