Hello again everyone. Following in the same spirit as the last Intel® Graphics Media Accelerator 3600 driver feedback thread, I'm making a new one to announce our latest driver release. Version 1091 just went live today, and I would again like to ask everyone to provide constructive feedback on it.
The link to the download is here; http://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&ProdId=3468&DwnldID=22633&ProductFamily=Graph... Download Center
The same rules apply for this thread as did apply for the last one. 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 requests/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.
Along with the feedback, if there are any new issues that resulted from the new driver, please provide detailed information on where you saw them, how it happened, and what steps to reproduce.
Does it crash every time you try to open the video? MPC-HC crashed the first time I tried to play the VC-1 file, but started up normally from there on out. I don't think my initial crash had anything to do with the GMA 3600 or Atom, I think it was just a software hiccup.
For MPC I had to go into the settings and change everything and then it worked, no idea why and I have now unselected everything and it is still working? :S For vlc the crucial change was enabling GPU acceleration, if that is on then it 'works' if it is off it fails. I've used MPC yesterday and today and had the same problem, I can only guess that somehow all the changes enabled GPU acceleration and that this has remained selected after I deselected everything again. I hope this is somewhat useful
Further video errors, using this video I am getting a green flash at the end of the video.http://www.bbc.co.uk/news/world-us-canada-22481629 http://www.bbc.co.uk/news/world-us-canada-22481629
I am not getting this effect on an Intel i7 3610QM.
If the driver can't even handle basic 2D performance... :-
...it's difficult to have any hope for any 3D performance improvements, in any case, I'm pretty much done with this chipset, and any subsequent PowerVR based hardware Intel chooses to release.
Powerarmour, we are all frustrated with the situation, but the purpose of this post is to try and fix these problems. If you could be more specific, as Kory and others are being then at the very least nic has some constructive criticism to work on. Telling him 2D is crap, doesn't necessarily help fix the problem. I have to say I have no problem with 2D graphics, but please name the application and the problem and I am sure it can go on the long list of problems that plague this piece of silicon.
May I suggest these locations for more direct criticism?
http://withimagination.imgtec.com/index.php/powervr/understanding-powervr-series5xt-powervrs-hardwar... Understanding PowerVR Series5XT: PowerVR's hardware is nothing without software optimization - With Imagination
They've even reinstated my comment which they deleted previously after I emailed, Intel also have a post on their facebook account today about the origin of a computer bug, another worthwhile post to comment on.
I wouldn't say the Intel Facebook post is a good place to comment, anything you say will get buried in other non-GMA related comments.
Halo 2/Half-Life video is now public, hopefully it will attract more attention to our situation.
On the subject of 2D performance, I get mixed results. On some pinball games I have (Addiction Pinball, Full Tilt! 1 & 2, etc.), 2D performance is fine. On games like Worms Armageddon/World Party and Icy Tower, there are slowdowns at times. Cave Story was completely unplayable until I switched to the NXEngine. On some graphics benchmarks, 2D scores are significantly lower than 3D scores. 2D performance could really be improved, but I haven't seen anything that's screamingly bad (that can't be fixed by switching engines).
How much work would it take to enable OpenCL 1.0 support for Cedar Trail? I use GIMP and HandBrake on a regular basis, and I'd love to have a boost (both support OpenCL to a certain extent). I could see myself switching to VLC if support were good enough, as stabilization and denoising are offloaded to the GPU. I'm sure Photoshop and WinZip fans would appreciate it, too. (info based on http://openclnews.com/apps/ http://openclnews.com/apps/)
I've been dabbling in Android development lately, and I've been having some trouble with OpenGL ES emulation in an Android virtual machine. I have Android 1.6 & 4.2 images installed; the former doesn't support hardware acceleration, so no problem there, but the 4.2 machine is very unstable with OGL emulation enabled. Whether it's because my netbook is simply too slow or there are driver issues (maybe both), I don't know.
I tried Extreme Tux Racer the other day, and found that it's great proof that 2D acceleration is broken. When snow is enabled, either in a race or in the menu, the framerate plummets. Disable it, and everything is smooth. I thought 2D OpenGL was fine on the GMA 3600, but apparently not.
So I've got a little bit of positive news, I tried Warzone 2100 today http://sourceforge.net/projects/warzone2100/?source=dlp (Warzone 2100 | Free Games software downloads at SourceForge.net), and it worked perfectly, it's also freeware so it might be of use to you (Nic) as it is the first 3D game that has worked as well as my previous Atom. For everyone else maybe it will remove some of our frustration
So there seems to be plenty of freeware games on this site (sourceforge) I've just downloaded FreeOrion and I'm sorry to say the graphics in this, even though they are only 2D are a complete mess. Random polygons appearing and the cursor constantly has a square box around it, similar to my experience in Kerbal. Would it be possible for someone to help catalogue the issues with all Sourceforge games?
I think it would be too complicated, time-consuming to check on all sourceforge games to look for bugs. The driver has no 2D acceleration so random 2D rendering is to be expected while Intel does not offer a properly working driver.
Nic, could you please tell us if a driver update can be expected anytime or are we just losing our time here ?
I think that's a little defeatist Lio, but I agree I think it would help if we could have some feedback on progress and or whether you need any extra help nic.
This is, of course - still a work in progress. As always, I will reply back here when I have some updated information.
Thank you for the update, even if it is just to say things are still taking time. I unfortunately have some bad news, today I was browsing the internet using Firefox version 22, I had 4 tabs open and no other programs running.
I started to see significant corruption on the screen, see an example in the attachment, this continued to increase until one quarter of the screen went black momentarily, I ignored this thinking possibly the website was poorly written. However a couple of minutes later after continuing corruption of both the pictures and text the computer stopped completely and restarted with a Windows notification that the graphics driver had crashed. When I continued to use the computer it then proceeded to go into a full BSOD. I then had to do a hard shutdown and restart, I then tried viewing the same website again and I have again had significant corruption, but so far no further crash.
Given that these graphics problems are now even affecting basic web usage of this hardware I think there needs to be a much higher priority given to this issue. If you require any further information on this please let me know. Might be useful if someone else can replicate this, I've been unable to on my desktop, which also is using integrated graphics.
Here is what Windows said after the BSOD:
Problem Event Name: BlueScreen
OS Version: 6.1.7600.2.1.0.768.11
Locale ID: 2057
Additional information about the problem:
OS Version: 6_1_7600
Service Pack: 1_0
What is your configuration alasdair555 ?
And after how much time you noticed the problem ?
Because I browsed trough the site for about 10 minutes with Firefox and I did not had absolutely any problem during my test, everything worked perfectly for me.
Maybe its the browser version.
Here again my configuration.
Cpu > Atom 2600
Video > Gma 3600
Sound > Realtek HD
Ram Memory > 2 GB
Operating System > Windows 7 Starter SP1
Browser Used during the test > Mozilla Firefox 19.0
I don't see any problem with the page either after scrolling through the page several times, but I'm using Chrome 30 (the dev version) with 5 other tabs open. Perhaps the way Firefox 19 handles hardware acceleration is different from 22, which would explain why Alasdair gets corruption and Viktor doesn't.
Me neither, I don't see any kind of corruption using Chromium 30. Have you tried with a different browser ?
As a general principle, I would recommend using Chrome or Chromium-based browsers as they are much more efficient under Atom CPUs. They are heavily multithreaded unlike Firefox which runs on a single-thread and causes a lot of system hangs.
Unless you only have a single gigabyte of RAM, that would be a bottleneck for both browsers, firefox uses less RAM in comparison to Chrome.
So thanks everyone for the replies, I tried to respond last night but the forum was refusing to let me log in.
So down to the questions:
My configuration is an Acer Aspire One with:
Processor: N2800 (yes not N2600) with GMA 3650
Memory: 4GB RAM, 3GB usable
OS: Windows 7 Starter
Browser: Firefox version 22
I noticed the corruption of everything when I moved down the page to the photo section.
I tried after Kory's comment with Internet Explorer and I agree there are no issues with that at all. So who knows what is happening, it's no issue to change to Chrome, I use the two interchangeably on my desktop, but it still raises the question what is going on with the driver and firefox that destroys the GPU so effectively?
Now Sery your question, it sounds like you are still using the older driver, just to be sure have you upgraded to the latest driver?
GPU acceleration under firefox is not really working great, regardless of the GPU used. I've read somewhere that disabling it in firefox would make the browser faster, which is a non sense when you talk about acceleration.
To see this in real use, you can try to run a 1080p video on on youtube under firefox and then run the same video under chrome/chromium.You will see the difference between the GPU acceleration of both browsers.