Follow-up

[Shurman292]

Some of you may remember this thread where I increased the IOH voltage and tried new drivers to remedy the problem where my GPU doesn't ramp up to the 3D clock speeds.

 

Well, I can safely say that the problem is not the GPU, because I am now having the exact same issues with the GTX 480 I just put in my system. I will mess around with it a little more tonight when I get home, but for now suffice it to say I've tried just about everything to keep it stable. I say "just about" because the only thing I have not messed with yet is the IOH PCIE voltage in my BIOS. My current overclock is nothing outrageous: 800 core, 1900 memory at stock voltage.

 

For completeness sake, I tried:

• 
   
  
• Cleanly re-installing the newest drivers
  
• Switching from MSI Afterburner to EVGA Precision X for overclocking (Afterburner would not change the core voltage per GPU-Z)
  
• Many restarts
  
• Increasing IOH voltage to 1.26 from 1.2 (I will go up to 1.3 tonight)
  
• Changed from adaptive to "prefer performance" in Nvidia control panel
  
• Added my overclock settings to the "3D profile"
  
• Backing down the overclock to near stock settings
  
• Increasing GPU core voltage
  

 

Any words of wisdom before I have at it again tonight? I'm at a loss.

[wevsspot]

Well, that's a bummer. I thought for sure that you had this problem whipped. Sorry to hear that it's still acting up. You've tried everything I can think of. As an experiment, have you re-ran your tests with the video card in the other x16 pcie slot to see if it exhibits the same symptoms there too?

[Shurman292]

That's the funny thing wevs, it wouldn't fit in the original slot with my small case and the HDDs in the way so I ended up putting it into the second x16 slot. Theoretically it should be the same, but you never know. I'm looking to get a larger case, so I may move it up to the first slot and see if it does the same thing.

[Shurman292]

Ran it with increased voltages, no changes. Performed clean driver installation, uninstalled all overclocking tools, rebooted, and it runs stock just fine. Clocks up to 700 mhz when necessary

 

Re-installed latest Afterburner, clocked it up to 750mhz/1900mhz no problem (was too excited to confirm stability for more than 5 minutes). Anything over that and it downclocked itself.

 

So now I'm thinking software, not hardware. Some sort of poor communication between the drivers and overclocking software.

 

Edit: By turning off voltage control in Afterburner, I am able to get it to stay at the higher clocks. Why would turning on the voltage control be doing this? I am glad I finally isolated the issue though.

[wevsspot]

Must have something to do with a faulty API call between Afterburner and the nVidia drivers. That's the only thing I can think of.

 

I'd be curious if it does the same thing using other overclocking tools like nTune or eVGA precision.

[Shurman292]

That's probably it, I just wish there were something I could do about it. I've tried Precision X and it did the exact same thing. I will have to give nTune a shot tonight and see how it goes.

[wevsspot]

This is interesting........... line 4 from the revision log on MSI Afterburner version 2.2.0 (2012-04-19);

 

• Dynamic clock frequency and voltage monitoring is now performed via new NVIDIA Kepler compatible API on branch 295 and newer NVIDIA display drivers

[Shurman292]

Forgive my lack of knowledge on the topic, but what does that mean? Only Keplar GPUs will get voltage monitoring on 295 and newer drivers?