[Arm-netbook] A33

[Luke Kenneth Casson Leighton]

On Tue, Jun 3, 2014 at 10:58 PM, Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:
> http://webcache.googleusercontent.com/search?q=cache:prYbhMDZblEJ:www.allwinnertech.com/plus/view.php%3Faid%3D398&hl=en&gl=uk&strip=1
>
> ok it's pin-compatible with the A23.  with the removal of SATA and
> replacement with USB it can be done.  i'll ask for some PCB CAD files.
>
> l.
>
> On Tue, Jun 3, 2014 at 10:55 PM, Luke Kenneth Casson Leighton
> <lkcl at lkcl.net> wrote:
>> On Tue, Jun 3, 2014 at 10:14 PM, Paul NeoStormer
>> <paulneostormer at gmail.com> wrote:
>>> So I can't find the page where this was discussed but I was told that the
>>> reason we couldn't go to a Quad-Core processor A31 was not because of pin
>>> compatibility but because it utilized a different graphics processor which
>>> we didn't really have open source for, a PowerVR instead of the Mali-400. We
>>> today they announced the A33 which uses the same Mali-400MP2 that the A20
>>> uses, so is it possible to go to quad-core now?
>>
>>  ooooo yes!  hooray!  if it's pin-compatible with the A20 we can use
>> it straight away.

ok just checked the A23 datasheet: it's only a 16-bit DDR3 RAM
interface, one USB2 and one USB-OTG.  so something like the 9514
USB-Eth hub would be needed (the 9512 apparently gets veery hot).
also apparently A23 only addresses 1gbyte of RAM, we'll see if the A33
goes up to 2.

 about powervr, you want luc's talk about the state of software libre
reverse-engineering of GPUs.  it turns out he's worked (in a
proprietary environment) with powervr, so he knows that it is an
absolute god-awful mess (hardware and software wise).  his
recommendation is to let imgtec stew, and to not even attempt the
reverse-engineering *at all*, it's that bad.

l.