Slashdot Mirror
Microsoft Moving Into Chip Design With Xbox Next
adamsmith_uk writes "According to ZDNet, Microsoft will more actively participate in chip design for the next version of its Xbox gaming console, tentatively called Xbox Next. By switching from using relatively standard parts to more customized silicon, the company can better optimize its game console, due in 2005. At the same time, the move potentially gives the company a toehold in a completely new market."
By switching from using relatively standard parts to more customized silicon, the company can better optimize its game console, due in 2005.
Or cut back on piracy. Perhaps we will have to activate games online in the future!
adventure-today.com
Linux yes. Pirating games no. Sony, Nintendo, and Sega have all used "non-standard" hardware and there has always been a way to pirate games. Let's drop the anti-MS stuff for a second and realize that a games console doesn't need to be a general purpose PC. MS can get more bang for the buck by designing this animal to the spcific application of games only.
If they want to BE like Apple Computers, they should just throw caution to the wind and do it. I imagine this doesn't bode well for Intel or AMD...
To be honest, I'll bet they are really vying to make their own chips for home users and set top boxes and keep Intel and AMD on the backend.
Un-news
This really highlights the stupidity of MS's anti-hacking efforts. I don't ever remember a company spending so much effort and money on an attempt to remove functionality and desirability from their products.
-3Suns
~~~~
The Revolution will be Slashdotted
yes. but that plan had one small flaw -it was bollocks.
Sony used custom Si with the same die area as Itanium1, yet could afford to pull it off by selling in the millions.
MS thought that by reusing PC kit they could get in the business easily (true), and ride the continual fall in PC part cost. Unfortunately, PC parts had had their cost already sucked out of them, apart from the effective 5% a month cost reduction of the Si parts. HDD and the DVD dont have much cost reduction at all, so that HDD is $70 of rotating iron whose cost is fixed. The best bit: Sony also rode the fall in Si parts, didnt have an HDD to provide fixed cost and can cut the selling price of the PS/2 whenever their spreadsheet hints that MS may be about to break even on hardware.
I think the biggest mistake of MS was thinking they could sell the hardware at a loss and make money on the games. The trick is to sell the hardware at a profit and make even more money on the games. Sony do that. Adding the HDD was another error. All it does is replicate DLL hell and add the Bill Of Materials of the box.
And they are effectively removing the aspect of XBox that made it cost effective and appealing to developers: easy porting to the PC through common components and CPU architecture.
I very much doubt the ActiveX APIs on the next XBox will differ much from the ActiveX found on ordinary PC hardware. Most of the rest, the compiler takes care of. How difficult is it to port most apps from linux-ppc to linux-i386, or even from linux-ppc to freebsd-i386?
The XBox already runs on not-quite-standard hardware and not-quite-windows-2k/xp..
SCO employee? Check out the bounty
Once the process is decided that it it. You can't just switch to someone else.
This means that for once in their life MS is at the mercy of someone else.
Screw IBM and you just free up resources for Nintendo and Sony (Assume you know that they have chosen IBM as well), and delay your own product by 1-2 years, meaning the project is pretty much dead.
IBM is the Ring that Rules them All.
Help fight continental drift.
Some battles are over (debut of XBox and sales relative to PS2) but the war is far from seeing the dust settle. Sony is currently almost on it's knees. It's being forced to drive the axe through some heads and consolodate some of it's internal corporate structure. I've been watching this one very closely. MicroSoft I'm sure has smelled the blood in the water and will soon pummel the limping Sony into submission. MicroSoft can afford to be patient and go slow...It will. Sony's days are nearing sun set. -Helium03-
What luck for the rulers that men do not think. Adolf Hitler (1889-1945)
And they are effectively removing the aspect of XBox that made it cost effective and appealing to developers: easy porting to the PC through common components and CPU architecture.
Yes, it may become less similar to the PC, but I don't think this will matter that much. I bet they'll still stick with DirectX for the graphics architecture and a Win32-like API. If they preserve those there will be no problem. The only ones who really care about the change are going to be those making bleeding-edge single-platform releases that barely fit the machine - Halo 2 and Doom 3 come to mind.
As for the CPU - nobody *needs* assembler in videogames any more - it makes the games unportable and isn't worth the performance gain as compared to designing your algorithms correctly in the first place. In general, the biggest performance problems when developing for a console are non-sequential disk access when loading a level from a CD - and CPU cache misses (budget mass-market on-chip cache compared to a PC). Everything else should be peanuts to solve if you have a good profiler. And if you're a serious developer, you're already doing three-platform simultaneous releases.
Where do they think they'll find chip-design engineers who will work on Windows? I wouldn't do that type of work on that platform, and all the others I know will only use a Unix based system for their engineering work. Does this mean that MS will be installing a new Linux network to develop their new ASICs?
I'm sure it's possible, but designing ICs requires some serious software and hardware tools, and an OS that won't get in the engineers way.
It's good to use your head, but not as a battering ram.
Although Microsoft did not produce key components of the PC, yet it had a very active participation in designing the standards ruling the PC world. From 1184 paralell port to ISA Plug&Play, from ACPI to DirectX 9.0 it was Microsoft who decided how the hardware should interface to the OS and in cases like DirectX 9.0 it acutally dictated lots of the arcithecture of the hardware. So it's not a surprise that it get goes one step forward for a product that is going to carry it's own name on the box...
The IBM representative acknowledged that Microsoft is looking at the company's PowerPC technology, the underlying architecture behind the chips in Apple computers. PowerPC concepts will also be the basis of the Cell processor, which will contain multiple chip cores that handle a variety of tasks.
So, MS is going to have to port over a major portion of their kernel, including directx and a few other bits, to the G5? Is this like Apple internally porting OS-X to intel, but never letting the public have it?
Although they have enough problems getting developers to sign exclusivity contracts. They are eating a little of their own pie, by telling developers that you can develop on standard PC hardware and software, then do a straight port over, but you can't sell the PC version for awhile.
Next version better have full-on network multimedia capabilities. I want to run my ripped DVDs on the TV without more than a network cord to me server. My current multimedia computer is too loud, and quiet ones are either too expensive, or too low end (no surround sound, etc)
-Adam
Their compiler and tools group is extremely strong. I'd be surprised if it was at all any more difficult to port to Xbox Next than Xbox.
I'm sure 95% of it will still be a solid C compiler and directx api.
hell, it'd probably have a setting for Endian notation in the dev env too.
the main loss is that with general components they can send devkits to developers early and when the ps3 specs get announced, MS could simply bump up the included cpu and gpu on the release units - guaranteeing that it'd keep ahead.
more likely though, it's just a matter of cost. It was too expensive to pay for a general purpose machine in each xbox - when it wasn't needed. they just better have backwards compatibility - which would be the one true victim of a powerpc switch.
// "Can't clowns and pirates just -try- to get along?"
it might not be so much to prevent pirating or Linux computing on the X-box as a way to get into chip design and knowledge for the hardware required to implement Palladium - if MS is solely dependent upon Intel or AMD for implementation of "trusted computing" it may not be as able to control the implementation effectively, while if they have an internal resource to design and fabricate chips, they can compete more effectively with competing standards from chip manufacturers. Since the evil that is trusted computing is a centerpiece for Microsoft's future, knowledge of the technologies required for it makes sense for MS. (this is assuming that Microsoft doesn't already have internal resources for DRM - even in that case, this may be another way to try DRM out in the field and to see how it works/doesn't work before they release it as part of Palladium).
Ask any developer in the country: working with Microsoft means jumping through a ridiculous number of hoops, and complying with really awful regulations (like all that X-Box Live crap).
Why do developers do this?
Because development for the X-Box is otherwise relatively easy. The X-Box being a modified PC, means that porting PC titles to the X-Box is cake.
The modified PC architecture also allows Microsoft to raid E3 for hot-titles, and buy out (or sign advance release deals) on hot titles. ("Halo" for example was originally supposed to be a PC release).
But what happens when Microsoft begins to move away from standard components?
The first and most obvious advantage to Microsoft is cost. Owning the chip manufacturing reduces the overall cost of production, not only by cutting out the 3rd party, but through efficiencies of custom architecture. This will translate into a more competitive console price. Most people don't know it, but Microsoft is in a state of panic right now over console prices. GameCube and PS2 can undercut X-Box comfortably in the late-stage console cycle (2 years after a console's release).
But (buyer beware) even though the X-Box NEXT will carry a nice price-tag, the number of titles will be SIGNIFICANTLY LOWER.
Developers *hate* working with the X-Box team at M'soft, and if coding for the X-Box was as difficult as coding for the PS2 developers would choose 1 console and stick with it.
This is almost guaranteed to happen with the release of X-NEXT. Watch as Sony announces a larger than ever release calendar and Microsoft is forced to go on an acquisition streak in order to bulk up on releases.
Also watch as GameCube surprises everyone with their next console which will demolish Sony and Microsoft's benchmarks...
------ The best brain training is now totally free : )
every cent you can shave of the production costs of a unit makes a big difference
Your reasoning is spot on for any console manufacturer, but it's especially important for Microsoft because of the dreadful arithmetic of long-duration per-box losses resulting from slow growth of Xbox against the PS2.
The problem there is that Microsoft doesn't write a whole lot of games itself, so they're at the mercy of the usual game dev companies' choice of platform and rate of production. That rate has been slow, and every month that the ramp-up drags on with the PS2 light-years ahead in terms of game numbers represents another chunk of losses stemming from the high cost of the console versus number of games sold.
Exactly why Xbox hasn't exploded onto the scene and become a head-for-head PS2 rival after all this time is a good question which I haven't seen explained anywhere. It's nice hardware from a dev perspective, so why so few games? (Even the Xbox mags are disappearing from shops. Looks bad.)
With the present sluggish rate of new releases and with way under 200 Xbox games in most of the "Coming Soon" lists despite Xmas approaching, I don't see any light at the end of the Xbox tunnel for a long time to come. Under these inauspicious circumstances, I'd have to guess (and we can only guess) that bringing down the pre-console loss must be extraordinarily important to MS.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
The flip side to this is that it will throw their own developers off. They, bless their suffering hearts, must put up with all the ugliness on normal M$ work and then some. Time to buy another SDK, suckers! Considering the poor sales, I don't know where they will get then next batch. What M$ screw their develpers again? Say it ain't so.
Friends don't help friends install M$ junk.
I can just see the Intels, Siemens, TIs and NECs of the world lining up for patent suits on this one. If Microsoft plans to wade into this battle without any existing IP they are going to get smacked very hard with the infringement stick...
Next version better have full-on network multimedia capabilities. I want to run my ripped DVDs on the TV without more than a network cord to me server. My current multimedia computer is too loud, and quiet ones are either too expensive, or too low end (no surround sound, etc)
:]
Then just get a normal Xbox, throw a mod chip, then get a copy of XBox Media Player..
It works beautifully.
Also, if the stock Xbox makes too much noise, improve it. I ripped out all the shielding and replaced the fan with a quieter one. It runs a few degrees cooler without that insulation, but collects dust more readily. I just leave the bolts out so I can pop the lid off and vacuum it out when it comes time to vacuum out my PC's (monthly)..
Or you could quit trusting a HDD with all your rips and just slap them on a DVD(+/-)R and get a DVD player that'll read them reliably (all the progressive scan Sony's do, to my knowledge).
Yeah, that'll go over well with the developers.
...
"So, are we developing straight for the chip, or what?"
"Well, actually you'll be developing on top of DirectX, which lies on top of a cut down version of XP, which runs on VirtualPC, which runs on a microkernel, which then interfaces with the System Bios and its integrated DRM."
"I mean, yeah, it'll be running right on the bare metal, Real Soon Now(TM)..."
"That's what I thought you'd say."
-Adam
Back in the early days of RISC the same was being said of the idea of breaking down the CPU, eliminating complex instructions.
My point is that you have a very different set of tradeoffs going on to those in a general purpose PC. The main reason PCs have FPUs is to run benchmarks, if you look at the work most PCs do they don't really need them.
The real question is just how much stuff you can fit onto a single CPU chip. It is pretty certain that you want to integrate the GPU and the CPU. The on-chip/off-chip delay is going to be a major bottleneck. That does not leave a great deal to eliminate.
The way to settle the matter is not to flame on slashdot, take some actual games and compile the damn things for a range of simulated hardware options. That is actually what we used to do in the early days of RISC, the compilers were optimized to the code, (at first to the end user code, later on the benchmarks :-)
Sure you may think that floating point is essential for games, but it is a completely different question to ask whether the best way to spend your gates budget is on a slick full feature FPU.
Looking for an Information Security student project suggestion?
Try http://dotcrimeManifesto.com/
You're right, it would be interesting to run some benchmarks, but games nowadays *do* require a lot of calculations with non-integer numbers. In all but the simplest 2d games (i.e. puzzle games and the like), vast numbers of calculations are made. I'd be willing to bet that nothing out there makes as much use of the FPU as games software. Sure, some kind of fluid flow analysis software or rendering software like PRMan and Mental ray probably tax the FPU more, but in terms of the sheer number of FPU instructions that are excecuted everyday by PCs around the world, I'm sure games come out on top.
Please believe me when I say I'm not making this up - I have worked on real, published games for both the PC and console markets and we do use the FPU a great deal.
If you need further proof, just have a look at all the gaming hardware out there - the PS2 has two extra vector floating point units in its so-called 'Emotion Engine' CPU. Graphics cards (which are pretty much driven by games) are dedicated almost completely to floating-point operations nowadays - everything, from the vertex coordinates to the colours to the screen buffers are, or can be, floating point. This is the way that the gates are being put to better use - a lot of them are now dedicated to doing certain FP operations *very, very* quickly, on a scale unheard of on a conventional CPU. If the FPU is to be removed, it is likely to be replaced by additional specialised FP units, to do things like ray/sphere and ray/triangle intersections, physical simulation and the like. I don't think they'll entirely throw out the main FPU (although it could be scaled down a bit in that case - current CPUs generally have 4-part vector FP operations), simply because there'll always be a bit of calculation you need to do somewhere that doesn't fit into your specialised categories..
I suppose it is possible that the entire games industry has got it all wrong, and that this drive towards greater FP power is going in the wrong direction, but I doubt it.
P.S. Please don't feel I'm flaming - just explaining the way I see it.
Cheers,
Tim