Intel Gma 3600

[Enon Zoberman]

Irecently purchased an Intel D2700 MUD motherboard and I cannot find drivers for the Win7 x64 integrated graphics (Intel GMA 3650 aka PowerVR sgs545). The accompanying CD contains Win7 x32 version only. When I run it I got an error:

I tried to use online utility Intel Driver Update Utility Graphics. I used Chrome, Firefox, Internet Explorer without success. First, UAC prompt appear, and then endlessly spinning progress-bar with text "Analyzing computer...". The text in UAC prompt is:

Hardware acceleration is either disabled or not supported by your video card driver, which could slow game performance. Make sure you have the latest video card driver installed and that hardware acceleration is turned on.

While Intel Corporation makes its own graphics programming documentation freely available ( ), Imagination Technologies considers its own programming documentation "Confidential Information" (a.k.a. military "Top Secret"). Imagination Technologies refuses to provide 64-bit drivers for its PowerVR graphics cores and is jealous of keeping its own documentation strictly under lock and key and of employing the long arm of the law to insist that its documentation remain Top Secret.

Plain and simple, folks. This is why Intel CANNOT produce drivers for the GMA3600 for ANY 64-bit operating system platform, whether it be Windows XP, Vista, 7, 8, open-source Linux, you name it. Period.

It's not really an Intel GPU but an integrated/rebadged PowerVR SGX545, so Intel doesn't seem to have much control over it. They do not make 64-bit drivers for this GPU, and haven't made any plans to do so public.

I'm in the same boat. I'm using a Shuttle XS36V with that CPU, now as a low-power server. The same GPU (albeit slower clocked) was available in an older Atom, and there are still the same issues today with that 32-bit driver. I don't expect much here.

Hello everyone. We've just released a new driver for this product. It's for Windows* 7 (32-bit) only, and should have significant performance improvements over the previous version. The download link is below. If you are running a GMA 3600 device on Windows* 7, please upgrade to this driver and provide your feedback in this thread. (Relevant complaints, performance decreases, etc). *EDIT*: Additionally, if there are pre-existing issues that you are still experiencing with the new driver, please provide that feedback as well!

However, please be aware that this thread is dedicated specifically to this driver and the feedback for this driver. This thread is not a discussion place for Windows* 8 GMA 3600 driver questions/complaints, nor is it for Windows* 7 GMA (64-bit) driver questions/complaints. Any such comments will be promptly deleted. However, please feel free to use any of the already-existing threads of that topic to discuss and provide feedback there.

Nic, it would be useful to know if the performance increases are supposed to be all round, or whether we're talking about fixes and optimizations for specific applications/games. If there is supposed to be an all round "significant performance improvement" then unfortunately I haven't seen it. As this is just a point release (originally for an OEM) I'd assumed it was bug fixes rather than any real performance boost.

They look like bug fixes mostly, but a lot of people were looking for those. I'll make an amendment to the original post, stating that we also want feedback on any pre-existing issues that were not resolved with the driver update.

Using driver version 8.14.8.1089, I am experiencing graphic artifacts glitching on certain Flash-based games, specifically on GlobalPoker from Facebook. The glitch will appear at random times, but would disappear when the game redraws the graphic on the affected area. If memory serves me right, I have not observed such behavior with driver version 8.14.8.1075.

Also, setting the Power Plans from the Intel Graphics and Media Control Panel doesn't persist upon reboot. Graphic Power Plan Settings for both Plugged in and On battery default to Balance, but if I set On battery plan to Maximum Battery Life, then restart the computer, the setting reverts to default. Is this an intended behavior?

*Less OpenGL glitching. TORCS is now playable without all the flashing junk polygons all over the screen. Glest no longer crashes when clicking on a unit. Minecraft is now glitch-free (albeit at 15 fps).

I'm noticing an overall improvement in performance, especially scrolling and opening things quicker too. Not sure if it's to do with B-3 stepping support in the drivers but all of the bugfixes and performance issues are welcome. Thank You Intel. Hopefully you will continue to improve the Drivers, but there are still some issues that I how you will fix. Thank You!

I've experienced much of the same. Fortunately, I've been able to find a few fixes. For Unreal Engine 1 games, alternate renderers, especially OpenGL ones, have boosted my framerates from 15 to 75 in Unreal Tournament(yes, I'm exaggerating a bit). As for Quake 3, it'll crash if you used 1024x768 or 1152x864 (with the Display1_DownScalingSupported hack) before you updated. The registry value resets with every driver update, so you'll need to re-enable it. If that's not the cause...I don't know what to tell you. My Q3 performance is largely unaffected, but it was satisfactory in the first place.

True. And maybe if we offer some help (that would be bug reporting) Intel will show some effort and update the drivers. Blaming will surely turn unproductive, as Intel already know that their drivers are poor.

Now that I am writing, I should also report that the system could use some more GPU time, as when I monitor GPU usage with GPU-Z its load never goes above 20% even under pressure (scrolling up and down quickly in browsers, for example) .

There is something broken on this SGX545 GPU in regards to DirectX 3D/2D acceleration in general, the only thing it'll do somewhat properly is to decode HD video. I'd rather they tried to bypass the broken parts of the hardware and run what they can on the CPU, at least then we'd get a better performing driver (at least on the desktop)

It's especially annoying while the diplomacy window is up (as shown). During regular play the corruption contains itself to the minimap in the lower left corner. This problem doesn't appear in the core game or the first two expansions.

I was trying to make the point that a CPU only driver actually performs better on the desktop for a lot of Atom/PowerVR users, if the GPU driver is that bad you can't even move Windows fast across the screen without tearing, then those acceleration (deceleration...) calls should be disabled.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

I tested HDMI out and it works great. I had to specify the device for HDMI in .asoundrc for sound out but it worked fine after I did that. I'm on linux 3.5.2-1. 2D acceleration is very sluggish still but I just restart xorg with fbdev and everything is smooth. Is there any way to restart X but keep your apps open? I'm going to write a small script to swap from pvr to fbdev and vice versa when I need one or the other. On the brightside I was able to watch 720p extremely smoothly connected over hdmi at 1920 x 1200. This driver is awesome thanks man.

Hello guys. I recently bought a laptop with this piece of crap instead of normal video card. Is there possible way to achieve 2d/3d acceleration with that proprietary driver. I need it to smooth running

And some updates: I added packages for what I think are 1.0.2 userspace libs from the package thomas001 was referring to.

Ubuntu has them, but I can't find them anywhere on the Intel site (?).

Not sure whether they will work any better; the files only slightly differ in size, dependencies are apparently the same (xorg 1.11 etc)

1.0.1_bee packages are still available for downgrading.

Is 3D acceleration (like libgl, not only 1080p thing) working?

If yes, is 3D acceleration working also on newer kernels or it does only on 3.2?

If yes, could you please write down few instructions to get the whole thing working (I know you already have, but I found everything a bit confusing since there have been lots of updates in a short period of time)?

smilzoboboz, I dont know what in theory, but in practice - glxgears show black screen, wine dont start 3d apps.

As for me - currently driver is unusable. You can watch 1080p, but have very poor perfomance in 2D (trought Xlib) and no 3D.

smilzoboboz: I'm not even sure OpenGL is supposed to work on PVR.

Hardware support is only stated for OpenGL ES, which is not a drop-in replacement for regular OpenGL.

MeeGo, where these drivers come from, has some kind of hardware-accelerated game working, but chances are it's linked against libGLES.

VA-API, on the other hand, should work. I'm not sure why it crashes, most likely library version mismatch somewhere but I can't tell for sure.

This seems to be Arch-specific since Ubuntu apparently has it working.

Thanks to axs I've successfully installed and am running the pvr driver on my Aspire One. My results seem to be the same as others such as terrible 2d performance, glxgears shows nothing, but videos play much more smoothly than with fbdev. I am going to get this crap off of my system until hopefully this becomes usable.

I ask because I have a D2500/GMA3600 system that just won't stabilize at native resolution using the gma500_gfx driver. If this code base represents a different path to providing a working driver I wouldn;t mind giving it a try, but it's rather difficult to get some of the pieces I'd need to use it unless there is more recent work done. For instance, I can;t find the xorg 1.11 package anywhere, although I was able to grab the linux-3.5.1 package(s). I see from the github link that work was done recently to build for linux-3.7.2, but there doesn;t seem to be any discussion about it anywhere - especially how to to build it, or if packages have been made available anywhere.

3a8082e126