[Arm-netbook] Fwd: Allwinner A31 + Imation/PowerVR/SGX

mike.valk at gmail.com mike.valk at gmail.com
Thu Apr 4 13:38:25 BST 2013


Hi Luke,

I Didn't try to pry, bash or complain. Just trying to prevent it.

Ok alternative's/current state

For ARM linked GPU's there are a number of choices for which open source
drivers are well underway.

GPU's I think are worth pursuing:
Mali : Lima (Almost ready to start build drivers and mainlining)
Ardreno: Freedreno (already mainlining)
Vivante: Etna_viv (Coming along niceley)

GPU's which might get interesting GPU wise:
Nvidia Tegra: Open source drivers are being build. With backing of Nvidia.
No 3D but there is info already available at Nouveau
http://www.phoronix.com/scan.php?page=news_item&px=MTI5MTk
Intel ATOM: The current Atoms contain little 3D/Video capabilities or the
Infamous PowerVR SGX IP cores. But the upcoming line of Atom's have Ivy
bridge GPU's. And Intel is writing mainline open source drivers already.


For PowerVR SGX cores noting is happening concrete. Nor does there seem any
interest.

I know generating money is not the goal. It's setting a standard.

Setting a standard requires momentum. That's usually gained by money. Like
I see it money=>power. But knowledge can also lead to power thus... Lets be
smart.

If we would like to gain momentum, we need to have least friction as
possible. Binary drivers cause us to be depening on single suppliers.
Vendor lock-in, VLI.

If we choose to make example EOMA's we need to make them work in a variety
of platforms. Having open source drivers releases the vendor lock in.

Choosing A31 results in availablity of one platform only: Andriod. And
which versions of Andriod is at the mercy of "Imagination Technologies". If
they do it like they did with Poulsbo than it will be just one.

For the EOMA standard using the A31 is no problem. But the validity of such
a A31 Card may become obsolete to quick.

If an EOMA card can be used by the masses for all purposes without any VLI.
Than the demand will be bigger. Thus creating more momentum.

And I'd whish I had the money, there it is again, to invest in the CPU you
proposed. It really sounds awesome and liberating. It reminds me of the
possiblities of the PS2 CPU. The greatly modified PowerPC Processor.

SoC Alternatives:
QualCom's snapdragons, Vivante
Marvell's Aramda's, Vivante (CuBox heart)


2013/4/4 luke.leighton <luke.leighton at gmail.com>

> On Thu, Apr 4, 2013 at 11:58 AM, mike.valk at gmail.com
> <mike.valk at gmail.com> wrote:
> > Hi All,
> >
> > I've been passively been watching the list for quite a while now. But
> > could not keep my mouth shut on this one.
> >
> > Why bother with the A31? It has a PowerVR SGX GPU core and probably
> > the VPU resides in that IP part of the SoC as well.
> > http://linux-sunxi.org/A31
>
>  correct.  just as vivante does not provide information, just as ARM
> does not provide information, just as nvidia does not provide
> information, just as every single GPU hard macro company does not
> provide information.
>
> > Please, I urge you to put your time to better use an leave every
> > Imation tainted SoC to rot. Perhaps that will persuade Imation to be a
> > bit more Linux friendly.
>
>  and we must also urge freescale to leave vivante alone, we must urge
> TI to stop using PowerVR, we must urge allwinner to stop using MALI,
> we must urge samsung to stop using MALI and PowerVR...
>
> ... in fact, we must urge absolutely every single SoC company in the
> market to cease and desist from designing multi-billion-dollar
> products.
>
> and, if they don't listen, we must influence their customers and urge
> the customers not to buy their crap.
>
>  ... can you see how that would be a waste of everybody's time?
>
> why would it be a waste of everybody's time?
>
> because you are not offering an alternative, are you?
>
> why would they listen to you if you do not offer them an alternative
> GPU which satisfies your requirements as well as theirs?
>
> do you know what their requirements *are*?
>
> where is the open hardware GPU that can be recommended to these companies?
>
> has it been tested?
>
> does it have the test vectors that these proprietary companies can use
> to verify that their $50 millon non-recoverable investment stands a
> high probability of working?
>
> so the plan is to make money, and then use that money to fund projects
> which *don't* depend on these arseholes.
>
> it's no good complaining about these things - we have to actually be
> in a position of influence, and to be able to offer these SoC
> companies what they want.
>
> and, even then if they won't listen, bypass them [*1].
>
> l.
>
> [*1]
> http://tech.slashdot.org/story/12/12/04/1748232/toward-an-fsf-endorsable-embedded-processor
>
> _______________________________________________
> arm-netbook mailing list arm-netbook at lists.phcomp.co.uk
> http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
> Send large attachments to arm-netbook at files.phcomp.co.uk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.phcomp.co.uk/pipermail/arm-netbook/attachments/20130404/d89d30eb/attachment.html 


More information about the arm-netbook mailing list