Showing results for 
Search instead for 
Do you mean 
Reply
Punch Card
Posts: 61
Registered: ‎10-21-2009
Location: NY
Message 1 of 18 (8,489 Views)

W510 checkered pixel patterns while running on battery [workaround]

[ Edited ]

Hey all, got my W510 in the mail yesterday and am quite pleased with it so far, but I have one issue that I've determined to be related to the power profiles.  When using any energy-saving power profile while running off battery power, when I have a flash website (e.g. Nvidia's homepage) open in IE or Chrome or if I move any window around the screen enough, some kind of strange pixel checkering occurs constantly when the screen has too much activity.  It's as if every other pixel gets shifted over by one making what looks like a checkerboard out of all the pixels on the screen.  I can still read the screen, but it makes my head spin.

 

If I choose the standard 'Max Performance' power profile, the issue doesn't manifest when running on the battery.  If I choose any of the others or create my own profile to mimic the Max Performance profile, I still get the checkers.  It still manifests if I go through and turn off all power-saving options or increase the brightness of the backlight.  If I start with the Max Performance profile and adjust it to my needs, the problem doesn't occur.  Something tells me that the performance profile has something subtly different than all the others and it's not a configurable setting.

 

Has anyone else experienced this problem?  Should I upgrade my Nvidia drivers from version 188.25, which came with the machine?  I've read that other people have had issues upgrading their Nvidia driver.  The way the system behaves, this seems like a power management issues more than an Nvidia driver issue.

 

Edit: I should also add that I don't have the issue while running on the AC adapter.  Also if I take a screenshot while the pixel shifting occurs, the screenshot is crisp and clear with no signs of the pixel problem.  This seems purely hardware/power management related.

 

Edit: see my later post on adjusting the power settings to work around this issue.

802.11n
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
Message 2 of 18 (8,468 Views)

Re: W510 checkered pixel patterns while running on battery

What you may be seeing is the graphics card turning off pixel-interpolation in the lower power modes.  So instead of smoothly magnifying certain images for display, it's using a rougher "pixellated" type of magnification. Just another way to save power.

 

Could be tied to a setting in the NVIDIA Control Panel.

 

 

Punch Card
Posts: 61
Registered: ‎10-21-2009
Location: NY
Message 3 of 18 (8,401 Views)

Re: W510 checkered pixel patterns while running on battery

[ Edited ]

I can't find any settings to adjust pixel interpolation or the like.  There are also no settings to turn off powermizer, but even with the max performance profile the powermizer is enabled, so I can't point the finger at powermizer, but I can't turn it off either.  I can't find any correlation in GPU-Z between the pixel issue and GPU clock speeds or GPU loads.

 

I can't get the behavior to occur while using the "Maximum Performance" profile (regardless of whether I adjust that profile to the same settings as any other profile).  The core/memory clocks drop to the 400mhz and then to the 135mhz range far more often and far sooner with other profiles, but that's the only thing I notice.  I do find that the GPU load is frequently higher while on other profiles.  Another thing I see is that the Nvidia 188.25 drivers have crashed a couple times on me.  The event message I get is Event 4101 Source "Display" - "Display driver nvlddmkm stopped responding and has successfully recovered."  This has happened at least once at the same time that the Power Manager stopped responding and crashed.

 

Another thing I do notice is that while playing video games the GPU temperature rises to as much as 66C (150F).  Is this a reasonable temperature to expect while running heavy graphics applications?

 

For now I'm just going to live with the max performance profile customized to the more energy efficient settings.  I'm hoping, since the 188.25 drivers are still not mature that all this gets resolved in a driver update later on.

 

I should add, here is a demonstration of what I'm seeing.  I can't screenshot the problem because it seems to be at the hardware/driver level.  Keep in mind that when it does occur, the pixel shifting is across the whole screen and is not bias towards text or graphics.

 

Pixel shifting issue.

802.11n
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
Message 4 of 18 (8,343 Views)

Re: W510 checkered pixel patterns while running on battery

[ Edited ]

It's reasonably likely that whatever issue you're seeing is linked to a setting in the NVIDIA Control Panel. So the best diagnostic you can do is, in high-performance mode (where the problem DOESN'T occur), write down the value of EVERY single setting that's available in that control panel.

Then, put the machine in a power mode in which the problem DOES occur. Then again go into the NVIDIA Control Panel and check every single available setting to see if anything has changed. There's a reasonable chance that you'll then find out what setting is causing the problem.


A picture is worth a thousand words, and a video probably ten times that. Taking a close-up shot of the screen while the problem is occuring, with either a still camera or preferably a video camera (or both), may help siginifcantly in diagnosing the problem. A simulated picture may differ in some subtle way from what's really happening, and those subtleties may be important. For instance, does the effect only break up vertical lines, and not horizontal lines? And does it only happen to content that is moving on the screen? And you're running your display at its native resolution, and not some larger or smaller resolution, right? Is the horizontal offset in the effect always just one pixel-width, or is it more than one pixel-width if you move things fast enough?

The simulated picture that you provide looks like it suffers from an artifact that appears WITHIN moving items that are displayed on an interlaced-scan display. The artifact goes away when the item ceases to move. But these aren't supposed to be interlaced displays. And I assume that you're talking about the built-in LCD in the W510, and not some external display, right? If somehow your display card is in interlaced display mode, then the problem probably won't appear in any screenshot the W510 takes of its own display. So at least that part of such a problem would agree with what you're describing.

So maybe the card is in an interlaced display mode when it's not supposed to be. Such a setting would normally be contained under some sort of "Video" menu in the NVIDIA Control Panel.


And from nVIDIA's website:

http://nvidia.custhelp.com/cgi-bin/nvidia.cfg/php/enduser/std_adp.php?p_faqid=2343&p_created=1231528...!!&p_li=&p_topview=1

"Graphics card temperatures typically range from 40°C to 90°C."


The only mention of interlacing that I see on their site for that graphics card is in the "Features" section of:

http://www.nvidia.com/object/product_quadro_fx_880_m_us.html

"PureVideo HD Technology 
NVIDIA® PureVideo® technology is the combination of high-definition video processors and software that delivers unprecedented picture clarity, smooth video, accurate color, and precise image scaling for SD and HD video content. Features include, high-quality scaling, spatial temporal de-interlacing, inverse telecine, and high quality HD video playback from DVD."

Plus there are many references to interlace features in the "NVIDIA Accelerated Linux Graphics Driver README and Installation Guide" which covers your video card. And those features are also presumably present in the Windows driver as well:

http://us.download.nvidia.com/XFree86/Linux-x86_64/195.36.08/README/README.txt

One such reference is:

"While [it] is possible to apply de-interlacing algorithms to progressive streams using the techniques outlined in the VDPAU documentation, NVIDIA does not recommend doing so. One is likely to introduce more artifacts due to the inverse telecine process than are removed by detection of bad edits etc."

So maybe it's trying to do spatial temporal de-interlacing when it doesn't need to (I once wrote software to do spatial temporal de-interlacing for a thermal camera, and you don't want that to be turned on when it's not supposed to be turned on). It mentions the de-interlacing in connection with a video mixer feature. So that's another possibility.


Anyway, that should keep you busy, if you want to fix it. ;-)  Good luck!



Punch Card
Posts: 61
Registered: ‎10-21-2009
Location: NY
Message 5 of 18 (8,325 Views)

Re: W510 checkered pixel patterns while running on battery

Problem Solved!  Well, with a workaround.  Still would like to see a driver fix for this issue and would like to see this option available in the Lenovo Power Management program.

 

It might very well be a bug in some sort of Nvidia interlacing feature, but I found the power setting that causes this issue.  I also read somewhere that Windos Vista/7 may have some kind of supersaver power feature that does this interlacing stuff.

 

I went to "Control Panel > Hardware and Sound > Power Options" and chose the custom plan (which the lenovo power management program apparently creates) and then I went to "Change advanced power settings" under that plan, and dug down under Display in the tree listing and found what is called "NVIDIA Display Power Saving technology".  I set that to Disabled when on battery (it was set to Enabled) and my problem went away.

 

So you know, the pixel shifting only happens on horizontal lines and is not selective on what it shifts.  The entire screen shifts as if the interlaced mode causes the display to shift each alternate line over one pixel with exception of the very left and very right edge pixels where I can't see the effect happening (probably because the display doesn't rewrite that pixel memory when the shifting occurs.

 

Thanks for your help Jimbo!

802.11n
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
Message 6 of 18 (8,268 Views)

Re: W510 checkered pixel patterns while running on battery

Good work, nanovox!

It's not clear what they were trying to do with the alternate rows of the display. But whatever it was, it has a bug in it. Probably at the driver level as you mention. Too bad you have to turn off all power-saving for the video card, however.

I have a ThinkPad W700 with the NVIDIA Quadro FX 3700M, which is a bit beefier, but the only setting that I have in the NVIDIA Control Panel that seems related is under Video & Television -> Adjust video image settings -> Use inverse telecine. It's turned on by default, no matter what my power settings.

 

Anyway, that's one more issue now in the rear-view mirror.

802.11n
Posts: 323
Registered: ‎10-26-2008
Location: Los Angeles
Message 7 of 18 (8,229 Views)

Re: W510 checkered pixel patterns while running on battery

Punch Card
Posts: 61
Registered: ‎10-21-2009
Location: NY
Message 8 of 18 (7,353 Views)

Re: W510 checkered pixel patterns while running on battery [workaround]

Hey all, I finally have camera shots of the interlaced pixel shifting bug.  I upgraded to the latest released Lenovo drivers for the FX 880, but the bug is still here.  Here are links to a couple screenshots for proof.

 

Pixel Shift 1

 

Pixel Shift 2

Punch Card
Posts: 13
Registered: ‎06-03-2010
Location: Australia
Message 9 of 18 (7,230 Views)

Re: W510 checkered pixel patterns while running on battery [workaround]

Thanks nanovox, I have the same problem as yours, see my post:

 

http://forums.lenovo.com/t5/W-Series-ThinkPad-Laptops/W510-display-problem/td-p/237166

 

I followed your instructions and fixed it. Thanks!

 

But I wonder is it normal or just in our cases.

Punch Card
Posts: 13
Registered: ‎06-03-2010
Location: Australia
Message 10 of 18 (7,228 Views)

Re: W510 checkered pixel patterns while running on battery [workaround]

another question:

 

is your windows 7 64 bit?

 

because I noticed that in the link posted by Jimbo, the two mates are all using 64 bit system, so am I.

 

Maybe it's a problem with the Nvidia driver under 64 bit system.

 

Cheers!

Top kudoed Authors