Little bit of good news re intel/powervr

[hammerd]

Some classic misunderstandings of PowerVR and Intel's relationship with IMG (owner of POWERVR).

There are no "proper" drivers for this because Intel CHOSE TO DO THEIR OWN DRIVERS. It is not a restriction put on it by IMG who would have happily done the driver job for Intel if they had been asked - and of course paid.

Off the top of my head I can't remember 100% who they did get - Tungsten was it ??

But I've seen some amazing stuff done on Atom systems using PowerVR so it's Intel's fault, not IMG's that there is not the "support" for PowerVR SGX.

There is a massive reason why Intel don't want "the community" buggering about with the drivers and that's because they never wanted Z Atoms to go into anything resembling laptops. The SGX based chips were always intended to be used for more embedded markets (a la mobile phones, MIDs, STBs, etc) rather than the more open laptop/laptop like markets.

Intel, I think it's safe to say, were mightily P@@@ed off that the Z Series chips jailbroke and were used in laptops. That eats into their big profit chip markets.

And there are going to only be more, not less, units shipping with PowerVR inside it.

While Pinetrail is using Intel's own graphics core (someone remind me again why this is REMOTELY a good idea !!), this is the replacement for the Atom N series chips and will again be aimed at larger format units.

The Z series chips are being replaced by Oaktrail which will, again, have SGX inside and will, again, be pointed towards more embedded applications.

The reason ?

It's BETTER. Much, much better. It's smaller, more powerful per square inch, more power efficient, basically everything the mobile market needs.

[droidhacker]

Some classic misunderstandings of PowerVR and Intel's relationship with IMG (owner of POWERVR).

The particular relation is irrelevant. The problem is that intel doesn't support their hardware.

There are no "proper" drivers for this because Intel CHOSE TO DO THEIR OWN DRIVERS. It is not a restriction put on it by IMG who would have happily done the driver job for Intel if they had been asked - and of course paid.

That would depend on your definition of "proper". At this point, proper is anything that *actually works*. Binary if it has to be, but SOMETHING.

Off the top of my head I can't remember 100% who they did get - Tungsten was it ??

Something like that.

But I've seen some amazing stuff done on Atom systems using PowerVR so it's Intel's fault, not IMG's that there is not the "support" for PowerVR SGX.

That amazing stuff... was it open source? My guess is probably not. Which means that we are still stuck not only with the possibility that this problem is of intel's invention, but also the possibility that intel is TRYING, and either IMG or Tungsten isn't playing along with them.

There is a massive reason why Intel don't want "the community" buggering about with the drivers and that's because they never wanted Z Atoms to go into anything resembling laptops. The SGX based chips were always intended to be used for more embedded markets (a la mobile phones, MIDs, STBs, etc) rather than the more open laptop/laptop like markets.

If intel didn't want it in netbooks, then they shouldn't have made it x86. There is really NO REASON for this bloat since a PDA/phone has NO REASON to be compatible with windoze.

Intel, I think it's safe to say, were mightily P@@@ed off that the Z Series chips jailbroke and were used in laptops. That eats into their big profit chip markets.

You do realize that the chip maker has the right to demand that the customer sign the contract that says "I agree not to use the chip in anthing larger than X", don't you? There is no law that says intel has to sell their hardware to everyone to do anything they want with it.

And there are going to only be more, not less, units shipping with PowerVR inside it.

While Pinetrail is using Intel's own graphics core (someone remind me again why this is REMOTELY a good idea !!), this is the replacement for the Atom N series chips and will again be aimed at larger format units.

The Z series chips are being replaced by Oaktrail which will, again, have SGX inside and will, again, be pointed towards more embedded applications.

The reason ?

It's BETTER. Much, much better. It's smaller, more powerful per square inch, more power efficient, basically everything the mobile market needs.

... everything the luggable market needs as well. Think of having a 15 inch laptop that can run continually for 3 days straight. Rig up a typical big battery from a 15" laptop to a z-netbook and it'll surely go for a week! No more of this 45 minute nonsense!

[hammerd]

The particular relation is irrelevant. The problem is that intel doesn't support their hardware.

Exactly - Intel isn't supporting this hardware, not IMG which is my angle. For disclosure, I have a major financial stake in the company (major in my terms, not the companies !) and my gripe in this discussion is always that this issue is seen as a PowerVR problem, when it simply isn't. IMG design the IP, the licensees design/manufacture the silicon. It is up to licensees what they decide to do with the silicon, who they sell it to, for what purpose and what support to provide for it.

That would depend on your definition of "proper". At this point, proper is anything that *actually works*. Binary if it has to be, but SOMETHING.

I mean as in half a dozen on screen videos running and simultaneously playing HD video on an external monitor while the CPU load is about 15% sort of "proper". But, the reason there is nothing for the open source community is down to Intel's choice for their embedded processors and where they want them positioned.

Something like that.


That amazing stuff... was it open source? My guess is probably not. Which means that we are still stuck not only with the possibility that this problem is of intel's invention, but also the possibility that intel is TRYING, and either IMG or Tungsten isn't playing along with them.

Half right. It IS Intel's invention, but no, IMG are not not playing along. AFAIK they were never asked to provide driver support, Intel chose Tungsten.


If intel didn't want it in netbooks, then they shouldn't have made it x86. There is really NO REASON for this bloat since a PDA/phone has NO REASON to be compatible with windoze.

Intel didn't want it in anything with a larger LCD size. I think the Dell Mini 10 was the last straw for them. They then saw that the Z series could power reasonably large powerful formats.

There is a very important reason to make a portable processor x86 compatible. It's called applications. Make it run full Windows and you don't need special developer communities to make it usable.

Intel's roadmap is not just Atom v 1. There's Moorestown and then the next one at least that we know about already. This is all part of a process for Intel.


You do realize that the chip maker has the right to demand that the customer sign the contract that says "I agree not to use the chip in anthing larger than X", don't you? There is no law that says intel has to sell their hardware to everyone to do anything they want with it.

There's more than one way to skin a cat. I can buy chips from anywhere I choose, Intel haven't told me how I am allowed to use them...

... everything the luggable market needs as well. Think of having a 15 inch laptop that can run continually for 3 days straight. Rig up a typical big battery from a 15" laptop to a z-netbook and it'll surely go for a week! No more of this 45 minute nonsense!

The next gen/one after that will bring Nirvana ;-)

[chithanh]

There are no "proper" drivers for this because Intel CHOSE TO DO THEIR OWN DRIVERS. It is not a restriction put on it by IMG who would have happily done the driver job for Intel if they had been asked - and of course paid.

If ImgTec do their own Linux driver, it will end up like the one for PowerVR Kyro: Proprietary driver, no updates for new kernel versions / X servers, ie. totally useless with any modern distro.

There is a massive reason why Intel don't want "the community" buggering about with the drivers and that's because they never wanted Z Atoms to go into anything resembling laptops. The SGX based chips were always intended to be used for more embedded markets (a la mobile phones, MIDs, STBs, etc) rather than the more open laptop/laptop like markets.

Intel, I think it's safe to say, were mightily P@@@ed off that the Z Series chips jailbroke and were used in laptops. That eats into their big profit chip markets.

Not credible. Intel allowed the Intel brand to be used for using Atom Z-Series and Poulsbo chipsets in laptops. They could have just forbidden the use of their brand name for anything above certain screen size and memory, but they chose not to.
And Intel has the power over this, like they successfully stopped most vendors from producing Atom N270 Netbooks with more than 10" Screen, 160GB HDD, 1GB RAM until recently.

[whizse]

So what happended with the free DRM/DDX combined with a new proprietary Gallium 3D driver?

http://www.phoronix.com/scan.php?page=news_item&px=NzY2Mg

(I'm a little bit surprised no one has set out the reverse engineer this one yet. Not as much love for these as for the Nvidia stuff? )

[Svartalf]

If ImgTec do their own Linux driver, it will end up like the one for PowerVR Kyro: Proprietary driver, no updates for new kernel versions / X servers, ie. totally useless with any modern distro.

1) They're providing binaries like it
2) Define "modern" distro- the drivers are in any Android, WebOS, Maemo device shipping right at the moment as well as Angstrom based devices such as the BeagleBoard and the soon to be shipping Open Pandora Handheld.

Having said this, I don't know how long there will be support like that- probably as long as TI keeps paying for it or Intel pays for it for MeeGo. I'm none too happy about the situation and would love to see a FOSS driver in this space. Unfortunately, the suppliers, one and all, are more used to dealing with the VxWorks, WinMobile, Symbian type customers where they don't demand source code. However, there might be some change in the wind in that space- can't say how or how much there will be, but...

Not credible. Intel allowed the Intel brand to be used for using Atom Z-Series and Poulsbo chipsets in laptops. They could have just forbidden the use of their brand name for anything above certain screen size and memory, but they chose not to.

That'd gotten them in trouble with government agencies- there's no TECHNICAL reason for "forbidding" their use. If the customer wants to use it in an application that Intel doesn't like, but if they've got the money otherwise and it'll do the job- they've got to let it go.

And Intel has the power over this, like they successfully stopped most vendors from producing Atom N270 Netbooks with more than 10" Screen, 160GB HDD, 1GB RAM until recently.

It wasn't Intel that did that (the same source of info for the change in the wind also gave me other tidbits... ). Take a guess who might've been involved with that and, if it could be proven, would be drug back into Federal Court over that fun and games...

[chithanh]

modern = boots on current commodity hardware and still has security support

Restricting use of the brand is not forbidden in any jurisdiction I am aware of. No government can punish Intel for not allowing "Intel Inside" stickers on Atom Z-Series netbooks.

Regarding limited netbook specifications, Intel and Microsoft ganged up on consumers to accomplish even more than protecting the brand: they deprived consumers of choice.

[droidhacker]

2) Define "modern" distro- the drivers are in any Android, WebOS, Maemo device shipping right at the moment as well as Angstrom based devices such as the BeagleBoard and the soon to be shipping Open Pandora Handheld.

First off, maemo is DEAD, so that is NOT modern.
Second, Android? Most android devices I am aware of use AMD/Qualcomm GPUs.
Third, palmos doesn't count as anything since it is tivoized up the wahzoo.

Fourth, and most importantly, these things you're talking about are ALL ARM chips. No intel branding on any of them AT ALL. The drivers are also entirely closed like a steel trap.

And that is again, the source of the problem -- if intel is going to brand it as theirs, they should be willing to support it!

And the reason why what I said in the first post of this thread is GREAT NEWS is that meego is an INTEL product containing a TRULY MODERN XSERVER (1.8) and their intention appears to be BIG TIME along the direction of supporting THEIR OWN HARDWARE (x86+SGX, or more specifically, moorestown). Presumably, a driver ripped from a moorestown+xserver1.8 will be directly and trivially applicable to a Z-netbook.

Release for meego is supposed to be MAY, which coincides with F13 (bleeding bloody edge of death as modern as it gets), both of which ship with xserver 1.8, which almost guarantees that there WILL be drivers compatible with *just any* modern distro (possibly with the exception of those containing xserver 1.7 -- there was a message on the intel forum to the effect that they may just skip support for 1.7).

Only thing to do now is WAIT.