GMA3600 Linux Support.

[DDing4]

Hey,

I know that their is practically no Linux support for any PowerVR based chipset. But I have a netbook that came pre-installed with Linux, and because of Intel not being bothered to make drivers that have 3D Acceleration we are stuck using GMA500 2D drivers on outdated systems. It's bad enough that you will not develop drivers despite the community pleading for them, but you will not even release the source code to the existing drivers (like the GMA500 driver). If you made the driver opensource then people from the Linux Community and other developers would most likely be able to develop fully working drivers with 2D and 3D Acceleration.

PLEASE Intel listen to your community. You sell us a product with Linux then abandon us. You cannot tell us to go to the Linux Community because we cannot develop drivers from complete scratch without you opening the sourcecode. You cannot tell us to goto Imagination Technologies (The people who developed the GMA3600 and licensed the chip to Intel) as they say it is Intel's responsibilty.

I urge people in need or these drivers to share your thoughts and show Intel that we need these drivers to use our netbooks.

-Sam,26 interg

[agree11]

More info so that we can compare specs and hopefully prove this board will work, before we buy and blow anything up The following are approved mini PCIe boards for the Acer Aspire One D270. While searching it appears most of the other Cedarview netbooks are derivatives of this in particular Gateway, Packard Bell and one other which I can't remember now sorry....

HUAWEI EM770W-REV2 LC.21300.066 http://www.s-chip.net/upload/picfiles/20120413151339123.pdf http://www.s-chip.net/upload/picfiles/20120413151339123.pdf

HUAWEI EM820W-Generic LC.21300.068 http://www.s-chip.net/upload/picfiles/20120413154745474.pdf http://www.s-chip.net/upload/picfiles/20120413154745474.pdf

If anyone feels rich and is willing to support me over Skype I have an old Samsung NC10 which I am will to test the board out on, to destruction if necessary...

[DDing4]

Honestly, I dont think anything is in danger of boom if pins dont touch. As for just replacing the wifi card, should be absolulty no problems what so ever.

I wish I could send you the stuff to test it, But I cannot, sorry. More pressure on me however lol, if I mess this up, I have no netbook haha

-Sam

[agree11]

No probs So what's the general idea? Exchange the wireless for the Broadcom, and then install a further mini PCIe slot, or just use a USB slot for wireless.

What is your netbook? Would you be able to upload pictures of the motherboard? I don't really want to open mine again just to take pictures, but assuming yours works I'm tempted to do the same. I understand this is the Linux post, but would you be able to test this using Windows 7? Particularly if the card has any impact on graphics performance apart from video (I assume it won't, but we can dream! )...

Thanks, Alasdair

[DDing4]

Yeah basically thats the idea. I havnt performed the modifcation yet, but when I get the PCIe card and the header, Ill post performances in Linux and Windows. The card should handle Hardware Acceleration with the drivers installed.

-Sam

[agree11]

Good luck with that Here is a second option if that fails, but from the looks of it, it only handles 2D and not 3D, but then that's still an improvement on the current situation.... http://www.advantech.com/products/PCA-5650-00A1E/mod_507F89E3-E80A-4964-B6EA-64A2EE21F2B1.aspx http://www.advantech.com/products/PCA-5650-00A1E/mod_507F89E3-E80A-4964-B6EA-64A2EE21F2B1.aspx

In the mean time the guy on this forum might be worth talking to, just to make sure it works with the motherboard:

http://forum.xbmc.org/showthread.php?tid=133899 http://forum.xbmc.org/showthread.php?tid=133899

[DDing4]

The card shouldnt have any problems at all, its just a card

A problem might be getting the second PCIe to work

[DDing4]

On Sun, Feb 14, 2010 at 2:03 AM, Ken Mandelberg <<a href="http://lists.x.org/mailman/listinfo/xorg">km at mathcs.emory.edu> wrote:

> I know there is a Linux driver for the chip, and XBMC knows how to use 

> it. However, would it make any sense to have an xvideo driver that knows > about it, so that arbitrary xvideo apps could use it? 

That doesn't make sense, the chip just decodes video, it isn't a GPU.

So you still have to transfer the resulting frame to the GPU. Besides

xvideo doesn't do

any decoding so you gain nothing by letting a chip does colorspace

conversation and scaling when you can do the same on the GPU using

xvideo.

Just done some more research. The above is an extract from a mailing list[1]. Looks like that CrystalHD card wont help afterall. and we really do still need the Linux drivers.

-Sam

[agree11]

Hmm, and the same goes for Windows? I always did wonder how the data moved from the chip to the actual screen, it didn't seem like it would be completely automatic, especially with the board we have!

On a selfish note, have you ordered everything? Will you still be trying in any case?

[DDing4]

Havn't ordered anything, I dont intend on now, considering it wouldnt help.

Back to square one.

[agree11]

I'm not so sure you're right, there are plenty of sites giving instructions to use this on linux, ubuntu in particular, if it had no effect then surely it wouldn't be used?

http://forum.xbmc.org/showthread.php?tid=137176 http://forum.xbmc.org/showthread.php?tid=137176

http://phoronix.com/forums/showthread.php?50200-Broadcom-Crystal-HD-Improvements-Under-Linux/page2 Broadcom Crystal HD Improvements Under Linux - Page 2

http://knowledge.evot.biz/documentation/how-to-compile-and-install-the-broadcom-crystal-hd-hardware-decoder-bcm70012-70015-driver-on-ubuntu How to compile and install the Broadcom Crystal HD Hardware Decoder (BCM70012/70015) driver on Ubuntu. - Evo Technologies, Inc. : Knowledge Base

Equally I've seen it used on an NC10, I think under Windows...

[DDing4]

No, The BCM70015 is a HD Decoder, ie, it only decodes video, not graphics. I was mislead by this.

[agree11]

Ok yes I agree, sorry that's what I thought all along, I was just hoping/dreaming that it might help other things also

[DDing4]

I seriously don't understand why, if they can't opensource the drivers, update them. It's not like they dont have the resources, they have done it before.

-Sam

[agree11]

Who are you yelling at now? To be fair to Broadcom they say it is a video card and that's what it does. Intel said they provided us with a graphics card, when they actually created a graffiti card...

[DDing4]

The overall fact is that Intel havn't even learned from the GMA500. Which had the same exact freaking problems with driver support. This is utter bull**** and its starting to get on a lot of peoples nerves.

Intel, just listen for gods sake. And try and save some of your brand.

[agree11]

I agree entirely, they haven't learnt. But to be clear it is almost definitely a case of a poor business decision rather than any lack of capability. In particularly they haven't learnt that either they are unable to write decent driver for IM's IP or that IM's drivers are completely useless and they can not be used to write anything.

[DDing4]

That cannot be true though, The PowerVR SGX5xx series are included on some phones, Including one model of the iPhone. So these SDKs and DDKs do work. If Intel knew that they was going to have no support at all, after previously experiencing this, thats just criminal.

[agree11]

I've just gone through 30 pages of the IT's forum. It appears that Apple also has similar problems, they just put more effort into fixing them. Oh and to get the full potential out of the chip, you must 'discover' all of the processes, and that only happens if your producer spends the time to do that. So guess what we have got, Intel spent the minimum amount of time to provide the minimum spec, i.e. HD performance and then has decided to ignore any game or any other type of graphical performance. I'll be going through the other 30 pages this week I hope, maybe I'll find some more information. Suffice to say there have been more than a few posts of people complaining about Linux support, what a surprise...

[DDing4]

https://www.change.org/en-GB/petitions/intel-listen-to-the-community-and-develop-gma3600-drivers-for-linux Please don't forget about the petition. Please sign it.

[DDing4]

I can't believe this. How long will it take for you to listen to us?

Also, I recently contacted Imagination Technologies via email. I asked that was Intel not updating drivers due to them not updating the DDK, they said they update it often, and while they may not be able to release an Open Source driver due to licencing, they should have no excuse to make good drivers.

Do you understand what you are causing? Linux is unusable. Windows performance sucks ass.

We arn't going to forget about this. I bought something, with an Intel chipset knowing it would have good linux support. Clearly I was wrong. Fix your crap Intel, seriously.

Sam.

[bleo]

I couldn't agree more with you Sam.

The problem here is that Intel is not (no longer?) doing its work to deliver rated-to-specs driver performance and support. I do not believe that the PowerVR graphics chip should be avoided like a plague if it weren't for companies (that's you INTEL) that license the graphics IP from ImgTec doing short of destroying the brand's reputation by non-delivery of support of the hardware's rated capabilities and published specifications.

I do believe that ImgTec should be putting pressure on its graphics IP licensees to deliver the goods as part of protecting the brand name. I even wrote something about it in one of ImgTec's blog: http://withimagination.imgtec.com/index.php/powervr/understanding-powervr-series5xt-powervrs-hardware-is-nothing-without-software-part-9# comment-966642962 http://withimagination.imgtec.com/index.php/powervr/understanding-powervr-series5xt-powervrs-hardware-is-nothing-without-software-part-9# comment-966642962

Clearly, the problem is NOT in ImgTec, but is in INTEL that should have been doing the driver porting work, being the licensee of the proprietary technology. Sadly, now that INTEL realized their blunder, never are they to repeat again and now will be releasing next-gen Atoms with Intel HD graphics . Too bad they SHOULD have done this NOW with Atom Cedar Trails (Trials?). This is a great DISSERVICE to Intel's LOYAL customers, ESPECIALLY those who were LEAD TO BELIEVE that the underlying hardware is Intel but is only masquerading as something else, who BELIEVED that the INTEL brand meant GREAT SUPPORT.