Archived from groups: comp.sys.ibm.pc.hardware.chips (
More info?)
George Macdonald wrote:
> On Wed, 22 Sep 2004 13:02:54 GMT, Robert Myers <rmyers1400@comcast.net>
> wrote:
>
>
>>George Macdonald wrote:
>
<snip>
>
> IBM sold Power 440 IP? Are you talking about the AMCC deal? Didn't look
> like an outright sale to me - just a license of some IP with a takeover of
> some responsibility for logistics & marketing.
<quote>
http://www.siliconstrategies.com/article/showArticle.jhtml?articleId=18901180&_requestid=208096
IBM to sell PowerPC line to AMCC for $227 million (Updated)
Silicon Strategies
04/13/2004, 8:35 AM ET
SAN DIEGO--Applied Micro Circuits Corp. (AMCC) on Tuesday (April 13)
announced a plan to acquire intellectual property and the assets
associated with IBM Corp.'s 400 series of embedded PowerPC standard
products for about $227 million in cash.
AMCC has also taken a license for IBM's Power Architecture. IBM will
continue to manufacture the PowerPC products for AMCC. The agreement
also provides AMCC with access to IBM's advanced CMOS process and
systems-on-a-chip (SoC) design methodology.
</quote>
The fact that there are underlying required licencses associated with
the Power architecture is an important detail (just ask anybody who has
a Unix source code license), but short of IBM completely divesting
itself of the Power architecture, I don't see how much more of an
outright sale the deal could have been.
> I don't think their picture
> is anywhere as bleak as you paint it and I don't see, with the IBM
> infrastructure, why a processor group would have to show a profit on its
> own.
>
"Bleak" is a pretty loaded word.
Power _doesn't_ have to show a profit on its own, and it could continue
indefinitely as the copestone of IBM's high end strategy. IBM has
really gotten smart about open source, and maybe it will be able to
sustain critical mass for Power with it's low-end Linux-only boxes. Or
maybe it won't.
<snip>
> As for Alpha, surely the cost of making the chip could have been fixed -
> the fab was ancient by the time Intel agreed to take it over.
Who knows? Probably compared to the cost of what has actually happened
with Itanium, doing whatever needed to be done to bring Alpha completely
into the Intel juggernaut would seem to be a bargain in retrospect. It
probably didn't look that way at the time the decision was made.
Chipmaking has to involve alchemy. What costs are immutable and what
costs are not and how are those costs tied to design? Like I would have
a clue.
<snip>
>>>So will we end up with just x86-64 and ARM as *the* computer
>>>architectures to choose from?
>>
>>Aside from the embedded market, maybe.
>
>
> I thought ARM *is* the embedded market or are you thinking of the bottom
> end of it more?
>
John Mashey has recently mentioned both the embedded market and
Tensilica more than once as examples of interesting action in computer
architecture, and there are some really hot network processors that have
appeared recently. I don't know what end that is, but I don't think any
of those processors have an ARM heritage.
>
>>The problem (as always, from my limited perspective) is that none of the
>>revolutions in microprocessor design have really been revolutions in the
>>sense that they answered questions there was a big payoff for anwering.
>> Intel thought IA64 was a revolution that answered an important
>>question (how to get significant parallelism without recoding
>>everything), but other architectures have been just about as successful
>>(or unsuccessful) in achieving the same goal.
>>
>>It's not as if there were no important questions worth asking--latency
>>tolerance, moving data around as the virtual real estate gets larger,
>>and, of course, power consumption--come to mind, but the demand drivers
>>just aren't big enough to drive a real revolution. Maybe if (say)
>>google succeeds in its plans for world domination and needs a real low
>>power revolution the way HPC needs a low power revolution.
>
>
> Revolutions are rare and I don't see why they'd be necessary as a sign of
> success. Steady progress with the odd discontinuity works fine for me.
>
You and Keith, except that I think, given the choice, Keith would
dispense with the odd discontinuity.
How many revolutions are at play here? Automatic computation, c. WWII.
The transistor, 1947. Integrated circuits, 1957. The microprocessor,
c. 1970. The personal computer, (as a real revolution, 1977-1981,
giving the Apple II and the IBM PC a tie). I think we're overdue.
>
>>>What will the Chinese do?... do they
>>>matter?<shrug>
>>
>>Of course they matter, but not soon enough for any but the most foolish
>>to speculate how.
>
>
> So far they're showing signs of going in the wrong direction - unique
> national standards for wireless could be just the start of something bigger
> and more destructive... their version of "playing by the rules"?? It's
> hardly a homogeneous culture so, with increased awareness of freedoms
> enjoyed elsewhere, I expect lots of Chechnya type unrest and attempted
> devolutions in the future. Many in the West who fear them economically
> make the mistake of regarding them as a monolithic society - IMO no where
> near as dangerous as they are painted.
Fortune magazine has a new article on Intel in China and on the
potential competitive threat from a Chinese semiconductor industry
(available on the net, but only with a subscription). The Chinese
apparently already have a home-grown chip that would compete with the
Pentium II.
The bigger picture for China over the longer haul? I have the same
skepticism you do: big, unwieldy society with adolescent ambition and
tremendous infrastructure problems.
RM