Slashdot Mirror
Microsoft Developing Console Chips
The Cheesecake writes "The New York Times is running an article that says that Microsoft is looking into designing and developing microchips. These will primarily be for the next generation of the Xbox. They also mention it could be used for things like voice recognition. They look to be doing this through a process designed by UC Berkley which makes it possible to reconfigure computer designs without the cost of making finished chips."
So now they shaft Intel and AMD.
Sun, SGI, DEC and IBM were disappointed in off-the-shelf CPU chips. Sun switched from Motorola to in-house SPARC. SGI bought MIPs to control CPU development. DEC had the most respected chip in the business. Apple used IBMs design. None of these enterprises were considered great commercial successes. Most of the survivors use Intel or AMD now. The big guys can come up with new versions each year or so and catch up to the "boutique" designs.
Call me crazy, but didn't Microsoft sign some sort of deal that they wouldn't enter the chip market a decade or so ago? Or am I remembering incorrectly?
---
Programming is like sex... Make one mistake and support it the rest of your life.
Here come the new hardware induced BSOD's...
Nuns. No sense of humor. -Kurgan
"They also mention it could be used for things like voice recognition."
Every time Microsoft introduces another new platform, whether OS, Office, HW, game console, or new executive, they promise voice recognition. Of course they never deliver.
Even the dedicated voice recognition researchers and developers don't have real voice recognition on any HW. MS doesn't do the kind of basic research necessary to move further down the road. And it doesn't even productize the R&D done by others - it copies or buys products from competitors. Or it keeps doing it wrong every time, until expectations are low enough that small improvements are declared victory.
The people who deliver useable voice recognition will work it out in the open telephony world, which has enough focus, money, constraints and momentum to actually get across the threshold to universal, untrained voice recognition that does something limited, but at least as perfectly as humans do.
Next we'll hear that these chips will be good for a "database filesystem"...
--
make install -not war
None of these enterprises were considered great commercial successes.
.com boom. How many do they have to sell to be a 'great commercial success'?
If you don't call DEC's Alpha chip a 'great commercial success', than what is? Does it still have to be on the market? What chip from the Alpha era is still on the market? They've all been redesigned since.
The only reason the Alpha 'failed' is because DEC's support business was so much more profitable than it's CPU business.
The big guys can come up with new versions each year or so and catch up to the "boutique" designs.
Intel only managed to catch up with the Alpha for two reasons: They stopped coming out with new versions of the Alpha, and Intel implemented patented Alpha designs without a license. It's not really a fair argument to you though, since Alpha wasn't 'boutique', and DEC *was* one of the 'big guys'.
Similarly, SPARC was *the* CPU of the
All the magic is out of CPU design. Lots of people know how to do it, and do it well. The hard part these days is in the manufacturing process, and you can buy that. There is no good reason not to design your own CPU if you can reasonably expect to sell enough of them,
It's nice to see Microsoft bringing quality to new markets ;). Seriously, Microsoft has to hedge it's bets - Windows and Office may not be cash cows forever. Twenty years from now Microsoft might be like IBM is today - important, influencial and profitable but not the young vigorous company it used to be. Microsoft should go for providing the best standards-based tools and environments it can. I believe that Microsoft place in the future is guaranteed and that at some point in the future they will be selling window managers for X alongside APIs that make everything easy to create and use (C#, XNA, Self configuring and healing networks, etc.). Microsoft's vast cash stores and pool of seventy odd thousand employees represents a major force in computing so don't be surprised when ten years from now you can download GPU updates if you were smart enough to buy a top-notch Microsoft console ;).
Shh.
It occurs to me that the DRM in the XBOX starts at the hardware level. So if MSFT wants to really lock down their systems making their own hardware would be a good place to start.
http://michaelsmith.id.au
My bullshit detector is going wild here. The magic is not all out of cpu designs. If you believe that, well you can go the way of the alpha.
The magic is not out of it. Most of the basics have been covered, that is true, and manufacturing process matters a lot; also true. But the manufacturing process is also not just a matter of "throwing a lot of money at it". However as I'm not as familiar with the manufacturing side of things, I'll stick to the areas I know, like processor design.
If the magic were gone in CPU design, and it was all about manufacturing, why do both ATI and Nvidia compete so heavily, and why can one produce chips faster than the others (although it switches too often for me to care). Sure they may not be general purpose processors, but they're highly important, and they are processors. Additionally while the changes in intels designs may be more on the level of incremental improvements (doubling cores etc) it doesn't mean there's not a lot of research going into it.
Now if you take the narrow mind that the magic in single core superscalare processor architecture is gone, sure. I'll agree with you there. That's a well studied problem. The research community moved on years ago. Simplescalar results no longer mean anything. However their is a large amount of research and development being done in the system design (the system level being restricted to a single chip). Cache-processor(s) interaction, efforts to improve programmability of chips etc etc.
Designing your own chip is a very very risky endeavor, even if you have multiple billions of dollars in the bank like microsoft does. If they manage to pull this off; more power to them. It's a very challenging process, and will not be done by grunts (i.e. it will require roomfuls of PhDs working on various parts of the project. Your comment about being reasonable to design your own... well I tend to disagree. Building the chip used in the xbox360 from scratch would have cost far more money than leveraging the design and knowledge expertise that IBM already had. Plus the chips needed for consoles have nowhere near the volume required to build their own fab. So in that case, that means they would contract out the fab work, and so the whole point is in their design. Sure looks like design is dead to me.
Phil
M$ has low customer apeal, as a brand label it sucks and beyond it's monopoly position on the os and office suite, which is currently being eroded, it has no customer loyalty. Take for example their new music player, how well will it sell, it depends upon how many of their potential customers they managed to piss off with WGA, families upgrading all their pcs with one oem package, their target market, now thats really stupid timing.
XBox 2 has failed to make an impact and it is becoming apparent that it is up to Sony to lose the next generation of consoles because M$ does not have the ability to take it.
M$ is trying to make the jump into a range of consumer products, mini broadband server/fire walls, portable and living room based media players as well as the game console. They do not have the brand desirability to achieve that goal, and will be forced to buy it. They have to move because they know the monopoly is expiring, their managment just lacks any real business skill beyond exploiting other companies mistakes (which doesn't happen anymore) and squeezing out a monopoly for all it is worth.
Chaos - everything, everywhere, everywhen