On 04.3.2015 &#1075;. 10:05, David Brown wrote:> On 03/03/15 17:54, 
Dimiter_Popoff wrote:
 > ......
 >>
 >> So my worst fears have become reality after all. Time to think of a
 >> future as a taxi driver or as a milkman... Apparently the power
 >> family has become too powerful too be allowed for general public
 >> access - and I not a politburo member.
 >> The world had to go down the drain, hadn't it. I am not sure many
 >> people understand how the vast rift is between a power processor
 >> and any other, well, it is vast. True CPU technology is becoming
 >> classified - which will kill it altogether of course but
 >> try to tell that the people who are behind this, then they are
 >> not interested in what technology there _is_ but in what technology
 >> is out there not under their control.
 >>
 >
 > I think you are a bit paranoid about politics here - maybe you remember
 > too much Bulgarian history!  There is only one influential power to
 > consider here, especially when talking about an American company like
 > Freescale and a Dutch one NXP (if it is still Dutch) - money.  If there
 > is money to be made by making and selling PPC devices, FreeXP will
 > continue to make and sell PPC devices.

Hah, may be you are right and I am too paranoid - or may be I am right
and you are too naive, who knows :-).
The thing is, we can't possibly tell - all we know is what we are
told...

Obviously I would dearly want to be the wrong one.

Dimiter

On 03/03/15 15:51, Boudewijn Dijkstra wrote:
> Op Tue, 03 Mar 2015 15:29:08 +0100 schreef David Brown
> <david.brown@hesbynett.no>:
>> On 03/03/15 14:50, Dimiter_Popoff wrote:
>>> On 03.3.2015 &#1075;. 15:27, Boudewijn Dijkstra wrote:
>>>> Op Tue, 03 Mar 2015 12:21:59 +0100 schreef Dimiter_Popoff
>>>> <dp@tgi-sci.com>:
>>>>
>>>> ......
>>>> (QorIQ) are new (and IMHO significantly big) markets for NXP.
>>>>
>>>>> In this case we don't know anything yet, as Don suggested. Either
>>>>> NXP want to shut down the power line and replace it with ARM - which
>>>>> is by far an inferior architecture - or NXP want to have
>>>>> access/control
>>>>> to the power architecture.
>>>>
>>>> I don't think NXP/Freescale is able to shut down Power.  They might
>>>> stop offering it, but if there is enough demand, there should be other
>>>> power.org members.  Or worse, power.org attracts a competitor (like
>>>> STM) as a new member next to IBM.
>>>
>>> Well yes, the thing is there is no other maker of power processors
>>> available on the market for non-politburo members (that is, things
>>> we can buy and work with without being a top 100 company).
>>> Let us just hope NXP want to develop this market, if this is the case
>>> they will likely leave things open (how else can they possibly be
>>> successful) as long as it is up to them.
>>
>> Can't you buy PPC devices from ST?
>> <http://www.st.com/web/en/catalog/sense_power/FM2098/SC963?sc=SPC56>
> 
> The cores are licensed via Freescale, I believe.  ST isn't listed as a
> member on the power.org website.
> 

That makes sense, given the similarity of the devices (not just the
cores, but the peripherals too).

On 03/03/15 17:54, Dimiter_Popoff wrote:
> On 03.3.2015 &#1075;. 17:26, Dimiter_Popoff wrote:
>> On 03.3.2015 &#1075;. 17:00, David Brown wrote:
>>> ....
>>>
>>>  From the quick look at ST's site, it is mostly automotive PPC devices,
>>> and not multi-core devices at several hundred MHz.  But at least it is a
>>> second source for some of these sorts of chips, and if FreeXP (or
>>> whatever) cuts down on them, then perhaps ST will ramp up their
>>> selection.
>>>
>>
>> At that size I don't really care much about the core. I have never used
>> a power core on such a small device, come to think of it. I would be
>> fine with ARM as I am fine with Coldfire, TI DSP etc., there is no large
>> programming effort for these small devices to think long term about
>> (years and decades ahead).
>> Things begin to matter when the entire DPS and the environment I work
>> under need to be ported (yes I can live without a wintel or any other
>> machine here and keep things going); last time I did it from 68k to
>> power it took me about a year (mostly to write the first vpa
>> compiler, had to figure out a lot of things I had never done before
>> (assemble 68k sources for power object code had - has - not been done
>> and I wanted to do it such that it would last, which it did for nearly
>> 15 years now).
>> Doing it yet again would be a lot easier but it would still be a huge
>> effort.
>> Which is why I am so paranoid about what is going to happen, I really
>> want to have things run on a multicore 64 bit power machine... And there
>> are a few very good looking - though yet to mature - choices on the
>> QorIQ list.
>>
>> Dimiter
>>
> 
> Just read a post on a Bulgarian language forum - a guy based in Chicago
> has spoken to a Freescale rep who said it would take a year or two until
> things are clear, but may be they would sell the power architecture
> family. It is obvious what this "may be" said that early means.
> 
> So my worst fears have become reality after all. Time to think of a
> future as a taxi driver or as a milkman... Apparently the power
> family has become too powerful too be allowed for general public
> access - and I not a politburo member.
> The world had to go down the drain, hadn't it. I am not sure many
> people understand how the vast rift is between a power processor
> and any other, well, it is vast. True CPU technology is becoming
> classified - which will kill it altogether of course but
> try to tell that the people who are behind this, then they are
> not interested in what technology there _is_ but in what technology
> is out there not under their control.
> 

I think you are a bit paranoid about politics here - maybe you remember
too much Bulgarian history!  There is only one influential power to
consider here, especially when talking about an American company like
Freescale and a Dutch one NXP (if it is still Dutch) - money.  If there
is money to be made by making and selling PPC devices, FreeXP will
continue to make and sell PPC devices.

Your biggest risk, I think, is that the PPC line will stagnate - if the
key customers for the chips (automotive for the small and mid size
devices, including some multicore, and networking for the big ones) move
to other cores (ARM, MIPS, x86), then the development of new devices
will fade away.  But historically, Freescale has been very good at
continuing to produce devices long after they have gone out of fashion.
 They still make 68332 chips, a good ten years after they tried to move
customers over to Coldfire or MPC so that they could shut down those
production lines.  They still have plenty of Coldfire devices, and still
have good support for them, even though they are mostly gone from new
developments.

You may also find that new PPC devices don't make it into the
mainstream, but will only be available for a few special customers.
It's not politics - the customers will be automotive and networking
companies.  But releasing a component to general use means a good deal
more documentation, sample code, development tools, marketing, etc.,
than releasing it to a specific high-volume customer.  I once worked on
a board that would use a brand new PPC microcontroller - my FAE got me
samples long before they were publicly available because the same device
had existed in automotive version for a couple of years.  (If you don't
have a good Freescale FAE from a good distributor, get one - they can
keep you informed here.)

On 03.3.2015 &#1075;. 19:51, rickman wrote:
> On 3/3/2015 11:54 AM, Dimiter_Popoff wrote:
>> On 03.3.2015 &#1075;. 17:26, Dimiter_Popoff wrote:
>>> On 03.3.2015 &#1075;. 17:00, David Brown wrote:
>>>> ....
>>>>
>>>>  From the quick look at ST's site, it is mostly automotive PPC devices,
>>>> and not multi-core devices at several hundred MHz.  But at least it
>>>> is a
>>>> second source for some of these sorts of chips, and if FreeXP (or
>>>> whatever) cuts down on them, then perhaps ST will ramp up their
>>>> selection.
>>>>
>>>
>>> At that size I don't really care much about the core. I have never used
>>> a power core on such a small device, come to think of it. I would be
>>> fine with ARM as I am fine with Coldfire, TI DSP etc., there is no large
>>> programming effort for these small devices to think long term about
>>> (years and decades ahead).
>>> Things begin to matter when the entire DPS and the environment I work
>>> under need to be ported (yes I can live without a wintel or any other
>>> machine here and keep things going); last time I did it from 68k to
>>> power it took me about a year (mostly to write the first vpa
>>> compiler, had to figure out a lot of things I had never done before
>>> (assemble 68k sources for power object code had - has - not been done
>>> and I wanted to do it such that it would last, which it did for nearly
>>> 15 years now).
>>> Doing it yet again would be a lot easier but it would still be a huge
>>> effort.
>>> Which is why I am so paranoid about what is going to happen, I really
>>> want to have things run on a multicore 64 bit power machine... And there
>>> are a few very good looking - though yet to mature - choices on the
>>> QorIQ list.
>>>
>>> Dimiter
>>>
>>
>> Just read a post on a Bulgarian language forum - a guy based in Chicago
>> has spoken to a Freescale rep who said it would take a year or two until
>> things are clear, but may be they would sell the power architecture
>> family. It is obvious what this "may be" said that early means.
>>
>> So my worst fears have become reality after all. Time to think of a
>> future as a taxi driver or as a milkman... Apparently the power
>> family has become too powerful too be allowed for general public
>> access - and I not a politburo member.
>> The world had to go down the drain, hadn't it. I am not sure many
>> people understand how the vast rift is between a power processor
>> and any other, well, it is vast. True CPU technology is becoming
>> classified - which will kill it altogether of course but
>> try to tell that the people who are behind this, then they are
>> not interested in what technology there _is_ but in what technology
>> is out there not under their control.
>
> Contrary to what seems to be a common belief I don't think it is very
> often that a company is bought just to shut down competing operations.
> It may be that the customers are more valuable than the technology, but
> clearly that does not apply in this case.
>
> What is feasible is for operations to continue with minor changes to
> combine operations for efficiency or for competing product lines to be
> sold off.  It will just be too expensive to buy a product line only to
> axe it.
>
> So don't worry that the power line will be shut down.  Instead consider
> that this may be an opportunity like when Intel sold off their StrongARM
> line after which it blossomed!
>

Thanks for the encouraging words, Rick. Let us hope for the best - that
someone competent will buy the PPC line _and_ will keep it public.
In fact the time it took for the new QorIQ lines to get truly going
was longer than I had hoped, may be there just is some problem there
(e.g. some key employee lost on that lost plane over the Indian ocean,
someone who left them, someone who died etc.).
So your hopes may be the best we have at the moment to rely on.... I'll
pray you turn out to be right.

Dimiter

On 3/3/2015 11:54 AM, Dimiter_Popoff wrote:
> On 03.3.2015 &#1075;. 17:26, Dimiter_Popoff wrote:
>> On 03.3.2015 &#1075;. 17:00, David Brown wrote:
>>> ....
>>>
>>>  From the quick look at ST's site, it is mostly automotive PPC devices,
>>> and not multi-core devices at several hundred MHz.  But at least it is a
>>> second source for some of these sorts of chips, and if FreeXP (or
>>> whatever) cuts down on them, then perhaps ST will ramp up their
>>> selection.
>>>
>>
>> At that size I don't really care much about the core. I have never used
>> a power core on such a small device, come to think of it. I would be
>> fine with ARM as I am fine with Coldfire, TI DSP etc., there is no large
>> programming effort for these small devices to think long term about
>> (years and decades ahead).
>> Things begin to matter when the entire DPS and the environment I work
>> under need to be ported (yes I can live without a wintel or any other
>> machine here and keep things going); last time I did it from 68k to
>> power it took me about a year (mostly to write the first vpa
>> compiler, had to figure out a lot of things I had never done before
>> (assemble 68k sources for power object code had - has - not been done
>> and I wanted to do it such that it would last, which it did for nearly
>> 15 years now).
>> Doing it yet again would be a lot easier but it would still be a huge
>> effort.
>> Which is why I am so paranoid about what is going to happen, I really
>> want to have things run on a multicore 64 bit power machine... And there
>> are a few very good looking - though yet to mature - choices on the
>> QorIQ list.
>>
>> Dimiter
>>
>
> Just read a post on a Bulgarian language forum - a guy based in Chicago
> has spoken to a Freescale rep who said it would take a year or two until
> things are clear, but may be they would sell the power architecture
> family. It is obvious what this "may be" said that early means.
>
> So my worst fears have become reality after all. Time to think of a
> future as a taxi driver or as a milkman... Apparently the power
> family has become too powerful too be allowed for general public
> access - and I not a politburo member.
> The world had to go down the drain, hadn't it. I am not sure many
> people understand how the vast rift is between a power processor
> and any other, well, it is vast. True CPU technology is becoming
> classified - which will kill it altogether of course but
> try to tell that the people who are behind this, then they are
> not interested in what technology there _is_ but in what technology
> is out there not under their control.

Contrary to what seems to be a common belief I don't think it is very 
often that a company is bought just to shut down competing operations. 
It may be that the customers are more valuable than the technology, but 
clearly that does not apply in this case.

What is feasible is for operations to continue with minor changes to 
combine operations for efficiency or for competing product lines to be 
sold off.  It will just be too expensive to buy a product line only to 
axe it.

So don't worry that the power line will be shut down.  Instead consider 
that this may be an opportunity like when Intel sold off their StrongARM 
line after which it blossomed!

-- 

Rick

On 03.3.2015 &#1075;. 17:26, Dimiter_Popoff wrote:
> On 03.3.2015 &#1075;. 17:00, David Brown wrote:
>> ....
>>
>>  From the quick look at ST's site, it is mostly automotive PPC devices,
>> and not multi-core devices at several hundred MHz.  But at least it is a
>> second source for some of these sorts of chips, and if FreeXP (or
>> whatever) cuts down on them, then perhaps ST will ramp up their
>> selection.
>>
>
> At that size I don't really care much about the core. I have never used
> a power core on such a small device, come to think of it. I would be
> fine with ARM as I am fine with Coldfire, TI DSP etc., there is no large
> programming effort for these small devices to think long term about
> (years and decades ahead).
> Things begin to matter when the entire DPS and the environment I work
> under need to be ported (yes I can live without a wintel or any other
> machine here and keep things going); last time I did it from 68k to
> power it took me about a year (mostly to write the first vpa
> compiler, had to figure out a lot of things I had never done before
> (assemble 68k sources for power object code had - has - not been done
> and I wanted to do it such that it would last, which it did for nearly
> 15 years now).
> Doing it yet again would be a lot easier but it would still be a huge
> effort.
> Which is why I am so paranoid about what is going to happen, I really
> want to have things run on a multicore 64 bit power machine... And there
> are a few very good looking - though yet to mature - choices on the
> QorIQ list.
>
> Dimiter
>

Just read a post on a Bulgarian language forum - a guy based in Chicago
has spoken to a Freescale rep who said it would take a year or two until
things are clear, but may be they would sell the power architecture
family. It is obvious what this "may be" said that early means.

So my worst fears have become reality after all. Time to think of a
future as a taxi driver or as a milkman... Apparently the power
family has become too powerful too be allowed for general public
access - and I not a politburo member.
The world had to go down the drain, hadn't it. I am not sure many
people understand how the vast rift is between a power processor
and any other, well, it is vast. True CPU technology is becoming
classified - which will kill it altogether of course but
try to tell that the people who are behind this, then they are
not interested in what technology there _is_ but in what technology
is out there not under their control.

Dimiter

On 03.3.2015 &#1075;. 17:00, David Brown wrote:
> ....
>
>  From the quick look at ST's site, it is mostly automotive PPC devices,
> and not multi-core devices at several hundred MHz.  But at least it is a
> second source for some of these sorts of chips, and if FreeXP (or
> whatever) cuts down on them, then perhaps ST will ramp up their selection.
>

At that size I don't really care much about the core. I have never used
a power core on such a small device, come to think of it. I would be
fine with ARM as I am fine with Coldfire, TI DSP etc., there is no large
programming effort for these small devices to think long term about
(years and decades ahead).
Things begin to matter when the entire DPS and the environment I work
under need to be ported (yes I can live without a wintel or any other
machine here and keep things going); last time I did it from 68k to
power it took me about a year (mostly to write the first vpa
compiler, had to figure out a lot of things I had never done before
(assemble 68k sources for power object code had - has - not been done
and I wanted to do it such that it would last, which it did for nearly
15 years now).
Doing it yet again would be a lot easier but it would still be a huge
effort.
Which is why I am so paranoid about what is going to happen, I really
want to have things run on a multicore 64 bit power machine... And there
are a few very good looking - though yet to mature - choices on the
QorIQ list.

Dimiter

------------------------------------------------------
Dimiter Popoff, TGI             http://www.tgi-sci.com
------------------------------------------------------
http://www.flickr.com/photos/didi_tgi/