1260258 - Intermittent graphics artifacting during normal browser usage

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

Attachment: screenshot 1

I've been seeing this for some time now, but haven't been able to nail down a firm STR. What I've been able to figure out:
* Seems to be easiest to hit when using context menus or opening/typing in the awesomebar. But of course, never on demand. Seems to be tied to longer-running sessions.
* Once the screen artifacts, resizing to force a repaint makes it go away. However, it tends to happen more readily once it's already occurred once.
* Seems to only happen in Optimus mode. At least, I don't recall any instances over a period of time when I was running only on the discrete GPU.
* I've tried pretty much every Intel driver imaginable at this point, up to and included the latest beta driver. The recent beta driver appeared to have made things better, but now I'm not so sure as I can still reproduce.

Relevant about:support information:
Adapter Description: Intel(R) HD Graphics P530
Adapter Description (GPU #2): NVIDIA Quadro M1000M
Adapter Drivers: igdumdim64 igd10iumd64 igd10iumd64 igd12umd64 igdumdim32 igd10iumd32 igd10iumd32 igd12umd32
Adapter Drivers (GPU #2): nvd3dumx,nvwgf2umx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um,nvwgf2um
Adapter RAM: Unknown
Adapter RAM (GPU #2): 2048
Asynchronous Pan/Zoom: wheel input enabled; touch input enabled
Device ID: 0x191d
Device ID (GPU #2): 0x13b1
Direct2D Enabled: true
DirectWrite Enabled: true (10.0.10586.0)
Driver Date: 3-3-2016
Driver Date (GPU #2): 2-8-2016
Driver Version: 20.19.15.4404
Driver Version (GPU #2): 10.18.13.6191
GPU #2 Active: false
GPU Accelerated Windows: 1/1 Direct3D 11 (OMTC)
Subsys ID: 06d91028
Subsys ID (GPU #2): 06d91028
Supports Hardware H264 Decoding: Yes; Using D3D11 API
Vendor ID: 0x8086
Vendor ID (GPU #2): 0x10de
WebGL Renderer: Google Inc. -- ANGLE (Intel(R) HD Graphics P530 Direct3D11 vs_5_0 ps_5_0)
windowLayerManagerRemote: true
AzureCanvasAccelerated: 0
AzureCanvasBackend: direct2d 1.1
AzureContentBackend: direct2d 1.1
AzureFallbackCanvasBackend: cairo

Dell Precision M7510 w/ Skylake Xeon, K1100M, 4K resolution. I haven't tried disabling e10s yet, will give that a try next. As you can see by the screenshots I'm attaching, it's pretty awful when it happens.

Comment 1

[Ryan VanderMeulen [:RyanVM]]

Attachment: screenshot 2

Comment 2

[Ryan VanderMeulen [:RyanVM]]

Attachment: screenshot 3

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Attachment: screenshot 4

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Still reproduces with e10s disabled. I'll try disabling HWA next.

Comment 5

[Ryan VanderMeulen [:RyanVM]]

No issues so far since disabling HWA. Bas, what can I do on my end to help provide actionable information when this problem happens?

Comment 6

[Bas Schouten (:bas.schouten)]

(In reply to Ryan VanderMeulen [:RyanVM] from comment #5)
> No issues so far since disabling HWA. Bas, what can I do on my end to help
> provide actionable information when this problem happens?

Not really, we've seen some slightly similar artifacts on hybrid intel/nvidia machines before. We know Microsoft is working on improving some things on their end. It's not the precise NVidia GPU type at least as I use a K1100M as well (although with an Intel HD 4600 I guess) and am not seeing this type of corruption.

Comment 7

[Bas Schouten (:bas.schouten)]

If you find any correlation with certain actions, memory usage, etc. that would be helpful.

Comment 8

[Ryan VanderMeulen [:RyanVM]]

Sorry, that should have been M1000M, but I think your general point still stands. If I run on the nVidia GPU only, I don't seem to have any issues.

So far the only correlations I've been able to make have been with menus (right-clicking, awesomebar dropdown, etc). But it doesn't reproduce on demand, of course.

Comment 10

[Bas Schouten (:bas.schouten)]

Seems more people with 530s are seeing this, do you have a 530 in Toronto lying around to test on Jeff?

Comment 11

[Ryan VanderMeulen [:RyanVM]]

FWIW, I recently installed driver version 20.19.15.4424 and so far, it's looking much better. I ran off the Intel GPU for an entire week while traveling and didn't have any artifacting at all. This version is available from Intel's website with a little hunting (it's the official version for some of their NUC products), or it's obtainable from Station Drivers. Anyway, hopefully that means a more-official release will be coming soon that alleviates the problem.

Comment 12

[Oriol Brufau [:Oriol]]

(In reply to Ryan VanderMeulen [:RyanVM] from comment #11)
> FWIW, I recently installed driver version 20.19.15.4424 and so far, it's
> looking much better.

I couldn't find version 20.19.15.4424. I tried 20.19.15.4404 beta but I still see the artifacts.

Comment 13

[Ryan VanderMeulen [:RyanVM]]

http://www.station-drivers.com/index.php?option=com_remository&Itemid=353&func=fileinfo&id=2161&lang=en

Comment 14

[Oriol Brufau [:Oriol]]

Thanks, I couldn't find it because I was searching 20.19.15.4424 instead of 15.40.22.4424.

The official download links are
https://downloadcenter.intel.com/download/25394/
https://downloadcenter.intel.com/download/25662/
https://downloadcenter.intel.com/download/25663/

All of these download the same file (SHA-1: 02a69c409c304dafc9cc846e3526771db7a3275b)

With this driver I have not seen any artifact.

Comment 15

[Milan Sreckovic [:milan] (needinfo for best results)]

Bas, Dees, do we have contact at Intel to have them somehow "push" these new versions out?

SurfaceBooks have been in bad shape, if these drivers are solving the problems, we'd really want them...

Comment 16

[Desigan Chinniah [:cyberdees] [:dees] [London - GMT]]

I have a few folks from Vincent to Joe to Tobie. Let me see if one of them can get us to the right folks. Bas may have others.

Comment 17

[Bas Schouten (:bas.schouten)]

(In reply to Milan Sreckovic [:milan] from comment #15)
> Bas, Dees, do we have contact at Intel to have them somehow "push" these new
> versions out?
> 
> SurfaceBooks have been in bad shape, if these drivers are solving the
> problems, we'd really want them...

I don't think we have anyone who has leverage on this, Joe would be the most obvious person to ask but I doubt there's much he can do outside the regular process.

Comment 18

[Desigan Chinniah [:cyberdees] [:dees] [London - GMT]]

Joe has confirmed that he is trying to get us to the driver team. He needs a few days.

Comment 19

[Oriol Brufau [:Oriol]]

The driver has been released for normal PC (not just Compute Stick and NUC).

https://downloadcenter.intel.com/download/25948/

Strangely, it says 15.40.22.4424 is old and recommends the problematic 15.40.18.4380 instead.

Comment 20

[Patrick Storz]

I'm afraid that the updated driver also does not fully resolve the issue...

After installing 15.40.22.4424 (aka 20.19.15.4424) I did not see the corruption while browsing anymore, but sporadically I still see it in menus (bookmark menu to be precise).

Comment 21

[Patrick Storz]

Driver version 15.40.23.4444 (aka 20.19.15.4444) seems to finally solve the issue.
It's currently available as beta from Intel [1].

[1] https://downloadcenter.intel.com/de/download/25957/Intel-Beta-Grafik-Treiber-f-r-Windows-10-und-Windows-7-8-1-15-40-?product=88355

Comment 22

[Andrew Overholt [:overholt]]

Let's just re-open this if it happens again. I was experiencing similar problems that went away with a driver update.