Microsoft to design its own CPUs - Next Xbox In Development

[AirRaid]

http://tinyurl.com/ydtq7k (nytimes.com)
http://games.kikizo.com/news/200610/063.asp
http://tinyurl.com/yadece (teamxbox.com)
http://www.eurogamer.net/article.php?article_id=68767


it seems Microsoft will try to design its own multi-core CPUs. the
same thing was said shortly after Xbox1 launched, that MS would design
its own CPU+GPU for Xbox2 but that did not exactly happen, MS only
worked closely on the design of CPU+GPU for Xbox2 (Xbox 360) with IBM
and ATI. maybe this time however, Microsoft is actually going to
attempt to become a CPU designer. whatever they do, they'll need
something that's competitive with the next-generation of CELL
processor(s) that will power PS4.

next generation of CPUs, in 5-6 years are going to have upto many
dozens of cores in on one chip
(Intel has an 80 core CPU planned for 2010, and they were talking about
100s of cores by 2015)


as for the GPU of the next-generation Xbox, I doubt Microsoft will go
it alone. they have two real options, ATI (AMD now) and Nvidia. my
bets are on AMD (ATI) to remain backward compatible with Xbox 360.

 

[Ed]

(Intel has an 80 core CPU planned for 2010, and they were talking about
100s of cores by 2015)

Intel talks a lot of BS, it's all part of trying to keep their stock
price from hitting rock bottom. 80 cores glued together, that should be
something! lol!

 

[nobody]

http://tinyurl.com/ydtq7k (nytimes.com)
http://games.kikizo.com/news/200610/063.asp
http://tinyurl.com/yadece (teamxbox.com)
http://www.eurogamer.net/article.php?article_id=68767


it seems Microsoft will try to design its own multi-core CPUs. the
same thing was said shortly after Xbox1 launched, that MS would design
its own CPU+GPU for Xbox2 but that did not exactly happen, MS only
worked closely on the design of CPU+GPU for Xbox2 (Xbox 360) with IBM
and ATI. maybe this time however, Microsoft is actually going to
attempt to become a CPU designer. whatever they do, they'll need
something that's competitive with the next-generation of CELL
processor(s) that will power PS4.

next generation of CPUs, in 5-6 years are going to have upto many
dozens of cores in on one chip
(Intel has an 80 core CPU planned for 2010, and they were talking about
100s of cores by 2015)


as for the GPU of the next-generation Xbox, I doubt Microsoft will go
it alone. they have two real options, ATI (AMD now) and Nvidia. my
bets are on AMD (ATI) to remain backward compatible with Xbox 360.

MSFT the chipmaker - who could've thought! Time to rename the
company. "MegaHard - your MicroSoft on Viagra"<lol/>.
But if we look at this beyond the jokes, MSFT has tons of ca$h and no
expertise in the field. Growing the expertise in-house is a lengthy
and very expensive process. The easier way is to buy an existing
source of expertise. The candidates would be TMTA <shrug
Expertise="insufficient"/>, VIA (same thing), NVDA, and AMD(including
ATi). INTC is out of play with its huge market cap (123B at the time
of writing), despite their mediocre performance during last years (but
if Otellini continues the trend, they may get into the play as
well<lol/>). IBM also is out of questions (137.7B). Sun actually
fits the bill in terms of cash, but anti-trust issues (namely Solaris)
may get in the way. The most logical candidate is AMD - both top
notch x86 and high end graphics in one not-so-big gulp, no software or
other strings attached. This might actually be beneficial for AMD - a
cash infusion wouldn't hurt, and may even close the gap with INTC on
transition to ever smaller process - about a year as it stands now.
Also would give the badly needed cushion in case their future K-x
architecture ever "does not live up to expectations" (case in point -
Intel's Netbust, AMD would not survive a disaster like this). If
anyone stands to lose - it's INTC stock holders. MSFT muscle behind
AMD will certainly beat INTC (not necessarily the company, but surely
the stock price) into the dust.

NNN

 

[Del Cecchi]

MSFT the chipmaker - who could've thought! Time to rename the
company. "MegaHard - your MicroSoft on Viagra"<lol/>.
But if we look at this beyond the jokes, MSFT has tons of ca$h and no
expertise in the field. Growing the expertise in-house is a lengthy
and very expensive process. The easier way is to buy an existing
source of expertise. The candidates would be TMTA <shrug
Expertise="insufficient"/>, VIA (same thing), NVDA, and AMD(including
ATi). INTC is out of play with its huge market cap (123B at the time
of writing), despite their mediocre performance during last years (but
if Otellini continues the trend, they may get into the play as
well<lol/>). IBM also is out of questions (137.7B). Sun actually
fits the bill in terms of cash, but anti-trust issues (namely Solaris)
may get in the way. The most logical candidate is AMD - both top
notch x86 and high end graphics in one not-so-big gulp, no software or
other strings attached. This might actually be beneficial for AMD - a
cash infusion wouldn't hurt, and may even close the gap with INTC on
transition to ever smaller process - about a year as it stands now.
Also would give the badly needed cushion in case their future K-x
architecture ever "does not live up to expectations" (case in point -
Intel's Netbust, AMD would not survive a disaster like this). If
anyone stands to lose - it's INTC stock holders. MSFT muscle behind
AMD will certainly beat INTC (not necessarily the company, but surely
the stock price) into the dust.

NNN

Explain to me why Microsoft would want to take on the grunt work of chip
design? If they have an opinion as to the high level design, surely the
folks they choose to partner with would listen. After all there is the
Golden Rule.

del

 

[Anne & Lynn Wheeler]

Explain to me why Microsoft would want to take on the grunt work of chip
design? If they have an opinion as to the high level design, surely the
folks they choose to partner with would listen. After all there is the
Golden Rule.

there are heavily into it with the game machines ... and they've
already hired some number of people in the area.

current state of the art trying to achieve real-time and realism is
requiring enormous software tricks to leverage the existing hardware
implementations ... something akin to the speculation about new POWER6
chip possibly having highly skilled people doing large amount of very
customed circuit optimization.

there is some speculation that by sufficiently orienting the hardware
design to the task ... that advanced, realistic game implementations
can be accomplished by a much larger percentage of the software
programming population.

 

[krw]

there are heavily into it with the game machines ... and they've
already hired some number of people in the area.

The point is; why buy the cow if the milk is free? Chip design in
expensive work. Chip fab even more so. Let others take the risks.

current state of the art trying to achieve real-time and realism is
requiring enormous software tricks to leverage the existing hardware
implementations ... something akin to the speculation about new POWER6
chip possibly having highly skilled people doing large amount of very
customed circuit optimization.

Nothing new about this. The bleeding edge always has a large
custom content.

there is some speculation that by sufficiently orienting the hardware
design to the task ... that advanced, realistic game implementations
can be accomplished by a much larger percentage of the software
programming population.

That's quite possible, but the question remains, why would MSFT
want to do the grunt work? Architecture, perhaps. I'm with Del on
this one; makes no sense, at least as stated.

 

[Some Guy]

The point is; why buy the cow if the milk is free?

Because Mac OS has moved to Intel chips.

MS is positioning themselves to be both the supplier of the CPU and
the OS.

Motorola died because there weren't enough MAC customers buying 68k
CPU's.

This is a warning shot across Intel's head by MS.

It could also be because of the retirement rate of Intel P3 and P4
CPU's. The stability of supply maybe is less than what MS wants for
the X-box.

 

[David Kanter]

The point is; why buy the cow if the milk is free? Chip design in
expensive work. Chip fab even more so. Let others take the risks.

How about because the milk isn't free. What do you think IBM was paid
for the xbox360? I agree that MS won't get into the fab business, but
honestly hiring some folks to design chips could be a good idea.

Nothing new about this. The bleeding edge always has a large
custom content.

IBM historically has avoided lots of custom logic AFAIK, so POWER6 may
be a real departure here.

That's quite possible, but the question remains, why would MSFT
want to do the grunt work? Architecture, perhaps. I'm with Del on
this one; makes no sense, at least as stated.

Maybe MS doesn't feel like giving profits to folks who compete with
them? Just a thought here...

DK

 

[David Kanter]

The point is; why buy the cow if the milk is free? Chip design in
expensive work. Chip fab even more so. Let others take the risks.

How about because the milk isn't free. What do you think IBM was paid
for the xbox360? I agree that MS won't get into the fab business, but
honestly hiring some folks to design chips could be a good idea.

Nothing new about this. The bleeding edge always has a large
custom content.

IBM historically has avoided lots of custom logic AFAIK, so POWER6 may
be a real departure here.

That's quite possible, but the question remains, why would MSFT
want to do the grunt work? Architecture, perhaps. I'm with Del on
this one; makes no sense, at least as stated.

Maybe MS doesn't feel like giving profits to folks who compete with
them? Just a thought here...

And honestly, buying a relatively small start up that has a good chip
for gaming applications probably wouldn't be too hard. That way, you'd
get a proven team, and you get to keep the expertise in house. This
would also have some advantages for MS, as they might actually learn
what EDA requires and come up with a compelling product for that
market.

DK

 

[krw]

Because Mac OS has moved to Intel chips.

Who cares about Mac?

MS is positioning themselves to be both the supplier of the CPU and
the OS.

Don't believe it. Ther is no room/money left in CPUs.

Motorola died because there weren't enough MAC customers buying 68k
CPU's.

No, Motorola "died" because of bad management.

This is a warning shot across Intel's head by MS.

Perhaps, but it's an idle threat.

It could also be because of the retirement rate of Intel P3 and P4
CPU's. The stability of supply maybe is less than what MS wants for
the X-box.

Huh? WTF does the P3 or P4 have to do with X-box?

 

[krw]

How about because the milk isn't free. What do you think IBM was paid
for the xbox360?

Can't comment, except to say that it was a *lot* less than what
they'd have to spend to do it themselves. The architecture and HLD
was already done.

I agree that MS won't get into the fab business, but
honestly hiring some folks to design chips could be a good idea.

Chips, perhaps. CPUs, not a chance. Some high-level architects to
provide direction perhaps but a completely new CPU architecture?
Not a chance.

IBM historically has avoided lots of custom logic AFAIK, so POWER6 may
be a real departure here.

Depends on how you define "custom", I guess. Customized
transistors, no. Custom circuits; done all the time. Mostly
standard cell, but custom is used when needed.

Maybe MS doesn't feel like giving profits to folks who compete with
them? Just a thought here...

???Who's competing with M$???

 

[krw]

And honestly, buying a relatively small start up that has a good chip
for gaming applications probably wouldn't be too hard. That way, you'd
get a proven team, and you get to keep the expertise in house. This
would also have some advantages for MS, as they might actually learn
what EDA requires and come up with a compelling product for that
market.

Just what "relatively small start up" has a bleeding edge CPU these
days? How many "relatively small start up" CPU companies are there
these days? Now, I could see them trying to tackle a graphics chip
(nvidia?).

 

[Anne & Lynn Wheeler]

The point is; why buy the cow if the milk is free? Chip design in
expensive work. Chip fab even more so. Let others take the risks.

there are a fair number of fabless chip operations (spun-off and/or
outsourced). the capital investment risk in the actual fabs is why
some number of chip operations have spun them off.

there were some number of operations in the 70s/80s that went thru
this transition period from custom/proprietary to cots (commercial off
the shelf) ... where the cost trade-offs was to use (less) expensive
off-the-shelf chips and devote relatively scare talent to more
critical parts of their business.

some of this could be considered in the light of the commerce
dept. meetings on hdtv and competitiveness in the late 80s and early
90s ... supposedly whoever owned the custom hdtv chip business
.... that there would be so many of these chips, that they would become
commodity standard.

so given a relatively high-volume activity ... are there marginal,
incremental activities that you can invest in that would still show a
positive ROI. part of this may be that you have already invested in
all the obvious stuff ... so it isn't directly a trade-off decision
between where to apply scarce funding and skills; you've aleardy done
all the more obvious things (both $$$ and skills). so now can you show
incremental return-on-investment by moving into other parts of the
activity; possibly considered as a form vertical integration.

so question is there enuf revenue ROI related to some specific target
markets to moving into more vertical integration operation (and is
there sufficient custom chip volume that it becomes somewhat akin to
what numerous people feared was going to happen with hdtv).

 

[Kai Harrekilde-Petersen]

How about because the milk isn't free. What do you think IBM was paid
for the xbox360? I agree that MS won't get into the fab business, but
honestly hiring some folks to design chips could be a good idea.

Only if they expect to do it on a regular basis. Staffing and tooling
up for a single project is simply a waste of money - have somebody
else do it for you.

Maybe MS doesn't feel like giving profits to folks who compete with
them? Just a thought here...

What's the opportunity cost? I'm all with Del here - it doesn't make
sense for MSFT to do the grunt work.

Hiring a bunch of knowledgable people in order to manage the partner, yes.
Hiring skilled architects to design and write the spec, yes.
But the grunt work - I simply cannot see that it would be worth MSFT's
money.


Kai

 

[nobody]

Explain to me why Microsoft would want to take on the grunt work of chip
design? If they have an opinion as to the high level design, surely the
folks they choose to partner with would listen. After all there is the
Golden Rule.

Ask Gates - or the ones who started the rumor. I am just speculating
what may (but not necessarily will) happen IF the rumor is not
completely baseless.

NNN

 

[ranjit_mathews]

Intel talks a lot of BS, it's all part of trying to keep their stock
price from hitting rock bottom. 80 cores glued together,

.... all running at 10GHz, undoubtedly.

 

[Tony Hill]

Because Mac OS has moved to Intel chips.

?? And that affects the X-Box how exactly? Honestly Macs and OS-X
are barely even a blip on Microsoft's radar for desktop computers
(Apple is still languishing at around 2-3% of worldwide sales), hardly
something for MS to get worried about.

MS is positioning themselves to be both the supplier of the CPU and
the OS.

Motorola died because there weren't enough MAC customers buying 68k
CPU's.

Macs made up a *VERY* small proportion of the processors that Motorola
sold. Besides, they didn't die at all, just split the chip business
off so that it could be more focused (ie the exact opposite of what is
being proposed here). And Freescale (the spin-off semiconductor
company) is doing fairly well for themselves, $250M profit on $1.6B
revenue for the third quarter of this year. I don't think that
they're missing Mac sales too much.

This is a warning shot across Intel's head by MS.

It could also be because of the retirement rate of Intel P3 and P4
CPU's. The stability of supply maybe is less than what MS wants for
the X-box.

MS doesn't buy chips off the spot market for the XBox, they have a
guaranteed supply as per their contract! Intel will provide the
original XBox chips for as long as the MS contract requires them too.
Same goes for IBM providing the XBox360 chips and the next generation
chips beyond that.

I'm in full agreement with the others, absolutely no point at all in
MS getting involved in the processor market, there are more than
enough competent suppliers out there who have already coughed up the
several billion dollars required for fabs and the R&D teams. MS will
probably get involved more closely on the specs and architecture of
the chips for their next-gen console (sorta like Sony's involvement
with the Cell chip for the PS3), but the actually processor design and
production will be handled by someone with expertise in that field.

 

[David Kanter]

Can't comment, except to say that it was a *lot* less than what
they'd have to spend to do it themselves. The architecture and HLD
was already done.

I absolutely agree, but that is not necessarily true for the Xbox3. If
the marginal effort is substantially higher (say, a ground up design),
then it may not be worth it.

Chips, perhaps. CPUs, not a chance. Some high-level architects to
provide direction perhaps but a completely new CPU architecture?
Not a chance.

So first of all, I'd point out that it isn't clear exactly what kind of
folks MS is hiring. Are they simulation guys, or circuit guys?

Secondly, it really depends on how often they plan to be doing MPU
design. If they plan to support a game console for the next 10-15
years, then it could easily make sense. Perhaps instead of putting
together a team that can do a from scratch design, they pull together a
team which can do all the shrinks and adaptations for different
processes. That could provide some real cost savings, as the teams
could be much smaller and not as experience as the folks who do the
first effort.

[snip]

???Who's competing with M$???

I don't know...maybe it's the largest software company in the world?

Let's try and list the IBM products that compete or substitute for MS
products:

S/390
OS/400
AIX
Linux
DB2
Rational's SCM tools
and the list goes on...

So why does MS want to go out of their way to help IBM again? ETS and
their fabs sure aren't cheap...

DK

 

[prep]

MSFT the chipmaker - who could've thought! Time to rename the ....

anyone stands to lose - it's INTC stock holders. MSFT muscle behind
AMD will certainly beat INTC (not necessarily the company, but surely
the stock price) into the dust.

The #1 candiate on the organ donor list would be SGI/MIP would
it not? They are also well trained in BOCICA to the gatesters
of bill.

--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be.

 

[nobody]

Let's try and list the IBM products that compete or substitute for MS
products:

S/390 Legacy

Legacy

AIX

And the market share (I mean of total servers, not just 'nix)
is what? Single digits? Or even less than that?

Linux

Gee, I thought it's freeware, not a product (by which I mean
something that sells at a profit). And since when Torvalds is on IBM
payroll?

DB2

Again, what's the market share outside mainframe?

Rational's SCM tools

Not an expert on this one, so leave this blank.

and the list goes on...

OK, let's continue...
Any non-exotic general purpose (desktop/workstation/server)
OS? The answer is: None, after OS/2 was taken off life support, and
even before that it was long comatose.

Any office suit? Lotus 1-2-3 is ancient history.

Email? As much as Outlook/Exchange sucks, it's a great
product comparing to Notes/Domino (arrrrgh!).

Where IBM eats MS and others lunch, and dinner too, is their
consultancy and outsourcing business, and a lot of it includes
development of MS-based products and support of MS-based systems
(Windows, SQL Server, etc.). Been interviewed by one of IBM guys for
a .NET consulting position, but we just didn't like each other (I
generally dislike Indians, and it shows despite my best effort to
suppress it).

So why does MS want to go out of their way to help IBM again? ETS and
their fabs sure aren't cheap...

Aren't cheap for whom? As of today AMD market cap stands at 9.8B - a
nice chunk of change for anyone, but, let's see...
MSFT market cap is 280B. AMD is exactly 3.5% of that. Buying it
outright will result in less of MSFT stock move than some
not-really-remarkable trading days produced. And there is no need to
buy the whole thing - about 25% would be enough to stuff AMD board
with MS people and dictate the directions. The only question: is MS
really up to making their own chips, or it's just a baseless rumor?

NNN