Quadro 4000 and Afterburner strange behaviour after OS resumes display

Hi All,

I have a Dell T3500 with Quadro 4000 on Win7 x64. I am using the latest (current) Nvidia driver. I run dual monitors. I have the OS power management set to turn off the displays after X mins, all other power is maintained. I use Nvidia control panel to set the GPU power management to adaptive. I use afterburner to control the gpu fan and act as the hardware monitor for clocks / temps etc. i

While the PC is in use, Afterburner shows 3 GPU clocks, 475MHz on full load( 3D CAD, Games etc) ,reducing to 405MHz partial load, and 50MHz for most 2D applications ( Web, office docs etc). This indicates that the adaptive power setting is working as expected and helps to minimize power and temps.

If I do not use the PC for X mins and the OS turns off the displays, when I resume the PC, afterburner shows that the GPU is under load at 405MHz and will not throttle back to 50MHz afterwards. I have to close afterburner and reopen it for the GPU to regain its NVidia adaptive power setting.:bang:

Is there a reason / solution to this?
Has anyone else experienced this?

This behaviour means that after resuming the display / PC, I see the temps rise significantly by 20C ( from mid 40C to high 60C) at idle.

 

settings>general tab, Read the context help for "Restore settings after suspend mode". My not help if just turning off monitors though.

 

Thanks for your reply,
I saw that setting in the user interface, and have tried with this option checked and unchecked. Doesn't seem to change the behaviour on resuming display.

I have also tried CPU-ID, to monitor the clocks and temps with afterburner turned off and found, the same behaviour on resuming the display, This program also needs to be turned off and back on again to initiate the NVidia adaptive power management.

I cannot confirm without third party monitoring tools, that the Nvidia adaptive power is re-initiated after the OS has resumed after the displays are off.

Any suggestions are welcome on how to determine if the NVidia driver works after resumed displays without third party monitoring apps. This would help to determine if its an Nvidia issue or the monitoring tools.

Any help would be greatly appreciated.

 

BTW, check to see if you have any active 3D processes running just for giggles going in or out of when monitor shuts off.

 

BTW : " restore settings after suspend mode" option is checked.

Afterburner (v4.3.0.8352)System information text
Before display off NVidia adaptive power
GPU clock @ 50MHz
Memory clock @ 135MHz

CPU
Model : Intel(R) Xeon(R) CPU W3580 @ 3.33GHz
Logical cores : 8
CPUID : GenuineIntel family 6, model 1A, stepping 5
Tjmax : 100°C

GPU1
Display device : Quadro 4000 on GF100GL GPU
Display driver : 368.39
BIOS : 70.00.56.00.07
GUID : VEN_10DE&DEV_06DD&SUBSYS_0780103C&REV_A3&BUS_2&DEV_0&FN_0
Registry key : \Registry\Machine\System\CurrentControlSet\Control\Video\{BC18A977-79F9-47C2-93A6-D5BE5C190A6A}\0000

On-Screen Display server
Path : C:\Program Files (x86)\RivaTuner Statistics Server
Version : v6.4.1
Active client(s) : On-Screen Display module
Active 3D process : NS.exe
: sldworks_fs.exe

Assigned hotkeys
Toggle OSD : CTRL + G



Resumed after display off.
GPU Clock @ 405MHz
Memory clock 1404 MHz
Active 3D process : NS.exe
: sldworks_fs.exe

After afterburner restart
GPU clock @ 50MHz
Memory clock @ 135MHz
Active 3D process : NS.exe
: sldworks_fs.exe

NS.exe = Norton security
sldworks_fs.exe = solidworks

Ending active 3D process solidworks = same behavior
Unable to stop / disable NS.exe in task manager.

 

Set RTSS detection to none for sldworks_fs.exe.Unless you need it for solidworks.

 

^^. This

 

Set RTSS for sldworks_fs.exe to none > reboot pc = same behaviour on resume.
Set ns.exe to none = same behaviour on resume, GPU clock stuck on 405MHz until restart of afterburner.

 

So there are no 3D processes now? Right?

 

That's correct.
Afterburner shows 3d processes- not detected.

 

Well after a little bit of searching, I found NVidia system monitor tool to show the gpu activity. without afterburner running, this monitoring tool shows the same behaviour. After resuming display the gpu clock is stuck at 405MHz, doesn't react to full load requirement or throttle back to save power.
Interestingly if you start another system monitoring tool, whilst the clock speed is stuck, the NVidia adaptive setting is re-initiated. Therefore I am assuming at this point that the odd behaviour I am experiencing is not directly due to a misconfiguration or conflict with Afterburner. It would appear that the NVidia driver needs a nudge ( in my case toggling hardware monitoring apps) to reinstate the adaptive power setting once the displays are woken.

My next step will be to contact NVidia support and see what they have to say, will keep you posted.

Thanks for your comments re: afterburner configurations, much appreciated.
Obviously any further suggestions on how to get the driver to behave properly would be appreciated. If all else fails, do you think its possible to run a script on display resume to initiate and close a hardware monitoring tool to force the driver to behave.

Thanks again.