x86 vs The World...

2014-06-05 01:08, rickman skrev:
> Many years ago when dinosaurs roamed the earth, there was the x86 and
> then there were the viable contenders.  That list was headed by MIPS and
> PowerPC with others as "also ran".  With time none of these
> architectures kept up with the megalith Intel's forces and the x86 took
> over the world resulting in the extinction of the dinosaur.
>
> In more recent times the focus has shifted from raw power (so essential
>   in the times of dinosaurs) to low power as dictated by the smaller
> size and battery operation of new devices (think mammals).  Now the
> power guzzling of the x86 line is creating global warming and
> threatening our very existence.  But today's processor on the white
> charger is the ARM, not PowerPC or MIPS.
>
> Was it inevitable that these lesser architectures take second fiddle to
> the ARM or was it just a fluke of marketing or corporate mismanagement
> that led to their being relegated to the processing hall of fame?
>
> Is there any compelling technical reason for the emergence of the ARM
> over other non-x86 processors?
>

The ARM (Acorn RISC Machine) was initially design for the BBC 
Microcomputer TV Series.
Once available they tried to sell it to the embedded world,
with focus on laserprinters. This was a total failure.

The critical moment was the design-in at Apple for the Newton PDA.

The Newton was a total failure, but at its introduction, this was not 
known, and Semiconductor Companies had learned the need for binary 
compatability from the PC market, so many companies licensed the
ARM core for use in ASICs, which they hoped then to sell to Apple.

With the ARM core available, people started to design ASICs with the core.

When I worked at National Semiconductor, Ericsson Mobile Phones started
a study (IIRC around 1995) to replace the Z-80 they were using in the 
GSM phones, and the main contenders were the ARM-7, AVR and the CR-16.

There were some criteria:
* Low Power
* 4 MB address space
* Synthesizable HDL description of the core
* IAR C compiler
* ENEA OSE RTOS support.

It was good timing, because while the CR16 was real low power, it only 
had 128 kB address space, beeing a 16 bit machine.
it was to be redesigned with a larger address space.

While working in the National Research Labs, I ran a project to
make a synthesizable core, and together with my Manager, we made sure 
that National focused on IAR compilers for all parts.

ENEA could get an RTOS very quickly once needed.
My only problem was that they designed the core with 2 MB address
space,
When the "official" project started, they requested working samples 
within 2 months.
I still remember the expression on the VP of Microcontroller,
when he found out from the manager of the Architecture group
that they had been aware of the 4 MB address space requirement for 2 
years...

The ARM architecture was judged as almost, but not completely useless 
for mobile phones, and Ericsson selected the AVR.

ARM redesigned their core to include the Thumb instruction set.
At the same time, at National Semiconductor's request, (based on my 
studies :-) they also designed a synthesizable ARM core.
I do not think National ever used the core though.

Most other mobile phone vendors were using ARM-7.
Ericsson ran out of steam with the AVR around 2001 or so, and then
went for the ARM-9 and later as many others the Cortex-A series.

The first general purpose ARM microcontroller was the Atmel AT91M40400,
which I think was introduced around 1998.

BR
Ulf Samuelsson










BR
Ulf Samuelsson

On 7/1/2014 6:48 AM, Ulf Samuelsson wrote:
> 2014-06-05 01:08, rickman skrev:
>> Many years ago when dinosaurs roamed the earth, there was the x86 and
>> then there were the viable contenders.  That list was headed by MIPS and
>> PowerPC with others as "also ran".  With time none of these
>> architectures kept up with the megalith Intel's forces and the x86 took
>> over the world resulting in the extinction of the dinosaur.
>>
>> In more recent times the focus has shifted from raw power (so essential
>>   in the times of dinosaurs) to low power as dictated by the smaller
>> size and battery operation of new devices (think mammals).  Now the
>> power guzzling of the x86 line is creating global warming and
>> threatening our very existence.  But today's processor on the white
>> charger is the ARM, not PowerPC or MIPS.
>>
>> Was it inevitable that these lesser architectures take second fiddle to
>> the ARM or was it just a fluke of marketing or corporate mismanagement
>> that led to their being relegated to the processing hall of fame?
>>
>> Is there any compelling technical reason for the emergence of the ARM
>> over other non-x86 processors?
>>
>
> The ARM (Acorn RISC Machine) was initially design for the BBC
> Microcomputer TV Series.
> Once available they tried to sell it to the embedded world,
> with focus on laserprinters. This was a total failure.
>
> The critical moment was the design-in at Apple for the Newton PDA.
>
> The Newton was a total failure, but at its introduction, this was not
> known, and Semiconductor Companies had learned the need for binary
> compatability from the PC market, so many companies licensed the
> ARM core for use in ASICs, which they hoped then to sell to Apple.
>
> With the ARM core available, people started to design ASICs with the core.
>
> When I worked at National Semiconductor, Ericsson Mobile Phones started
> a study (IIRC around 1995) to replace the Z-80 they were using in the
> GSM phones, and the main contenders were the ARM-7, AVR and the CR-16.
>
> There were some criteria:
> * Low Power
> * 4 MB address space
> * Synthesizable HDL description of the core
> * IAR C compiler
> * ENEA OSE RTOS support.

So which of these features did the PowerPC or MIPS not have?  I guess 
PowerPC was not availble as HDL, but wasn't MIPS being used in ASICs 
already?


> It was good timing, because while the CR16 was real low power, it only
> had 128 kB address space, beeing a 16 bit machine.
> it was to be redesigned with a larger address space.
>
> While working in the National Research Labs, I ran a project to
> make a synthesizable core, and together with my Manager, we made sure
> that National focused on IAR compilers for all parts.
>
> ENEA could get an RTOS very quickly once needed.
> My only problem was that they designed the core with 2 MB address
> space,
> When the "official" project started, they requested working samples
> within 2 months.
> I still remember the expression on the VP of Microcontroller,
> when he found out from the manager of the Architecture group
> that they had been aware of the 4 MB address space requirement for 2
> years...
>
> The ARM architecture was judged as almost, but not completely useless
> for mobile phones, and Ericsson selected the AVR.

Any idea why the ARM was thought useless?  Was it too large and power 
hungry compared to the tiny AVR?


> ARM redesigned their core to include the Thumb instruction set.
> At the same time, at National Semiconductor's request, (based on my
> studies :-) they also designed a synthesizable ARM core.
> I do not think National ever used the core though.
>
> Most other mobile phone vendors were using ARM-7.
> Ericsson ran out of steam with the AVR around 2001 or so, and then
> went for the ARM-9 and later as many others the Cortex-A series.

I guess your window into the scene is limited to what could be seen from 
inside National Semi?

-- 

Rick

>Dimiter_Popoff wrote:
>
>> On 05.6.2014 г. 02:08, rickman wrote:
>> > Many years ago when dinosaurs roamed the earth, there was the x86 and
>> > then there were the viable contenders.  That list was headed by MIPS
and
>> > PowerPC with others as "also ran".  With time none of these
>> > architectures kept up with the megalith Intel's forces and the x86
took
>> > over the world resulting in the extinction of the dinosaur.
>> >
>> > In more recent times the focus has shifted from raw power (so
essential
>> >   in the times of dinosaurs) to low power as dictated by the smaller
>> > size and battery operation of new devices (think mammals).  Now the
>> > power guzzling of the x86 line is creating global warming and
>> > threatening our very existence.  But today's processor on the white
>> > charger is the ARM, not PowerPC or MIPS.
>> >
>> > Was it inevitable that these lesser architectures take second fiddle
to
>> > the ARM or was it just a fluke of marketing or corporate
mismanagement
>> > that led to their being relegated to the processing hall of fame?
>> >
>> > Is there any compelling technical reason for the emergence of the ARM
>> > over other non-x86 processors?
>> >
>>
>> Power architecture is very much alive, just check what Freescale are
>> making.
>
>PowerPC in quite a few of the Freescale parts is designed to work
>in bad environments both electrical and physical.  (automotive
>engine controllers, and bad environment applications like process
>control.
>
>w..
>
>
>

I work with a lot of safety-critical embedded customers, and most of the
ones I have seen prefer PowerPC.

All that being said, never underestimate the value of a good tooling
infrastructure to the success of a chip. There are a lot tools out there
for ARM chips making it easy to experiment with.


	   
					
---------------------------------------		
Posted through http://www.EmbeddedRelated.com