Microsoft leaks details on Xbox Next

[Deadmeat]

...

No direct PC to XBOX2 conversion, knowledge on the PC won't help you on XBOX2?

Actually Xbox Next will look familiar to developers because it maintains API compatibility. Same Win32, same DX. Developers have moved away from assembly for sometime.

3 core cpu isn't gonna help the developers more than the VU0+VU1.

It is a totally different situation if you understand programming.

The weird thing is they may be on to something, how easy will it be to actually see the difference in the next gen.

That's what MS is counting on.

Xbox is 3X as powerful as PSX2 yet people don't see much difference. PSX3 could be 3X as powerful as Xbox Next yet they wouldn't see a difference; what they will see is the price difference and game selection.(MS revving up faster because it is based on same API structure)

 

[-tkf-]

Re: ...

No direct PC to XBOX2 conversion, knowledge on the PC won't help you on XBOX2?

Actually Xbox Next will look familiar to developers because it maintains API compatibility. Same Win32, same DX. Developers have moved away from assembly for sometime.

I doubt very much it will be that easy.

Anyway if MS doesn't fail again i will be surprised, usually MS needs more than 2 tries to get things right and for once the competition is fighting back.

 

[Deadmeat]

...

It'll only be one copy IN MEMORY, but all CPUs in SMP (or whatever-MP), still run the kernel of course, or else the OS won't function. How can you do inter-process communication or taskswitch without it? You can't!

Context switching and virtual memory is just a small part of OS services, the real resource hog is driver code and networking nowadays. Having SMP means one processor will run interrupt handler, TCP/IP stack and drivers, while other CPUs are free to spend 95% of its time in user code uninterrupted.

So you've said more than once. What actual evidence do you have to back that up?

IBM supercomputers.

despite that, SIMD instructions are still so viable Intel's busy extending the SSE instruction set with each generation of its microprocessors!

Usefulness of SSE is diminishing with each passing generation of DX.

You'll see that while altivec is a Moto trademark, it is in no way a Moto IP, it was co-developed by Moto, Apple and...you guessed it, IBM.

Co-development? More like one guy does all the work while the others sit an watch and write out requirements. In this case, Apple wrote the req, Moto worked to build it, and IBM never showed up to the party because of certain disputes. Only after Moto quit PPC R&D did IBM rejoin the party and start working on an Altivec compatible unit to meet the needs of legacy code.

Appears? How do you figure? XBoxes still have harddrives in them, don't they? So how could they "appear to be going that way anyway"?

Because we heard of this HD dissapearing rumor elsewhere before.

And greater flexibility?

I never said it. That's your word, not mine.

 

[Guden Oden]

Re: ...

3 core cpu isn't gonna help the developers more than the VU0+VU1.

It is a totally different situation if you understand programming.

*giggles madly*

Yes, totally different, I'm sure!

Tell me again, what's YOUR experience with programming...? Professional, and otherwise (and no, I do not mean the kind shown on TV.)

 

[function]

I have no doubt that the Xbox2 will have around the same performance as the PS2

<laughs> You'd better repair this statement quick, before someone comes after you with a knife.

Haha! Well spotted.

 

[Guden Oden]

Re: ...

Context switching and virtual memory is just a small part of OS services, the real resource hog is driver code and networking nowadays.

Driver code isn't part of the OS, and in a console you're not going to have much in the way of either virtual memory management, networking OR drivers. Your case is incredibly weak here.

What actual evidence do you have to back that up?

IBM supercomputers.

No, sorry, you can't use the argument itself as basis for the argument! That's circular reasoning! Don't you get it? Now try again, dammit.

Usefulness of SSE is diminishing with each passing generation of DX.

Oh, I'm SURE you, with all your understanding of programming, know all about the usefulness of SSE... See section about eating crow in previous post...

Co-development? More like one guy does all the work while the others sit an watch and write out requirements.

So you're busy making stuff up again? Were you there during development or something?

With ALL due respect, I think I trust a guy like Hannibal more than I do you...

Only after Moto quit PPC R&D did IBM rejoin the party and start working on an Altivec compatible unit to meet the needs of legacy code.

So, having been duly corrected on the subject of IBM having to license tech it in fact helped to develop, you have now changed your previously proven faulty stance to 'IBM not having been in on the party'; another completely fabricated scenario based on...what, really? IBM supercomputers????

Because we heard of this HD dissapearing rumor elsewhere before.

Forgive me for saying so, but you HAVE heard about rumors, right? Like, them not neccessarily being true and all that? So I ask you AGAIN...

JESUS. You CAN'T use rumors and innuendo as a basis for speculation about MS doing this or doing that! Also, how do you know this second rumor is not based off the first rumor, huh?

And greater flexibility?

I never said it. That's your word, not mine.

True, I misread, you said "reliability". Anyway, flash memory doesn't exactly live forever, and if it's soldered to the motherboard it'll be pretty much impossible to replace for a layperson. Especially after that particular device's not manufactured anymore.

 

[Fox5]

Re: ...

No direct PC to XBOX2 conversion, knowledge on the PC won't help you on XBOX2?

Actually Xbox Next will look familiar to developers because it maintains API compatibility. Same Win32, same DX. Developers have moved away from assembly for sometime.

3 core cpu isn't gonna help the developers more than the VU0+VU1.

It is a totally different situation if you understand programming.

The weird thing is they may be on to something, how easy will it be to actually see the difference in the next gen.

That's what MS is counting on.

Xbox is 3X as powerful as PSX2 yet people don't see much difference. PSX3 could be 3X as powerful as Xbox Next yet they wouldn't see a difference; what they will see is the price difference and game selection.(MS revving up faster because it is based on same API structure)

Xbox is not 3x as powerful as ps2 in every area, and is even weaker in some. If xbox was 3x as powerful and 3x better and had 3x as much as the ps2 in every area, then people would notice the difference. Oh, and it games were developed for xbox first and not ps2.

 

[bbot]

Oh you're gonna love their sound solution this time around.

Are they going to use the third processor as a dsp? Three processors seems like a strange number of processors for SMP.

 

[Guden Oden]

Using an entire G5 or whatever CPU for sound would be a terrible waste and a crime. Anyway, sound falls onto the shoulders of the chipset maker, SiS, I believe, since they won the contract to do the southbridge I/O chip, where it would be logical to place sound.

 

[Megadrive1988]

Xbox is not 3x as powerful as ps2 in every area, and is even weaker in some. If xbox was 3x as powerful and 3x better and had 3x as much as the ps2 in every area, then people would notice the difference.

here here.

absolutely right on.

no way in hell is Xbox 3x more powerful than PS2. not even 2x. not in every area that is. PS2 is more powerful than Xbox in a number of ways, while Xbox is more powerful than PS2 in other ways. bringing the GCN into the picture, its more powerful than the other two consoles in certain areas, and likewise, GCN is weaker in than its rivals in yet other areas.

the fact is, each of the three consoles has specific strengths over its two rivals.

 

[Dio]

The funny thing is I'd agree that Xbox isn't that much more powerful than PS2, but I'd disagree that nobody notices the difference. I find the texture quality on PS2 games is generally poor. Not licensing S3TC was a truly massive mistake.

(I was astonished to find that a texture palettisation utility reviewed in Develop was used "on 90% of last year's top fifty games". Palettisation should be dead as the dodo, given that DXTC is half the size and - in the vast majority of cases - hugely better quality.)

 

[Megadrive1988]

here's something interesting about the graphics processor:

What nobody is telling you and you'll know about this first, here on TeamXbox, is the revolutionary approach of the Xbox 2 to deal with today's biggest problem in graphics chips: memory bandwidth.

The graphic chip will contain not only a graphics rendering core but up embedded DRAM acting as a frame buffer that is big enough to handle an image that is 480i and can be 4 times over sampled and double buffered.

well, its not revolutionary anymore. it was done with PS2 and GCN. but it is new for Xbox 8)

also, something on the CPU

As the Mercury News article says, the Xbox 2 will use 3 64-bit processors. But we'll give you even more details. They'll be PowerPC processors, based on the PowerPC 976, the first dual-core 97x chip based on the POWER5 architecture, which will also be the first PowerPC built on a 65nm manufaturing process. Check our previous story for further details.

Each core is capable of processing two threads, thus meaning the whole system can process six threads simultaneously. It is worth noting that, because of its RISC nature, the new architecture is a big endian, a term that describes the order in which a sequence of bytes are stored in computer memory. Traditional x86 architecture use the little-endian approach.

The Xbox 2 is a multiprocessor-enabled, 64-bit platform. It is a proven server architecture that, in the case of Xbox, won’t take advantage of the greater than 4-GB physical memory space benefit of 64-bit computing but will make use of its other benefit: wider data paths and registers, something extremely useful in the execution and process of both integer and floating-point calculations.

http://www.teamxbox.com/news.php?id=5388

 

[zurich]

I still don't understand this '3' CPU number, doesn't that defy the laws of binary?

 

[Fafalada]

I find the texture quality on PS2 games is generally poor. Not licensing S3TC was a truly massive mistake.

People mainly complain about texture aliasing and lower resolution - S3TC wouldn't help either of these as most PS2 games already run predominantly 4bit art.
While the improved color integrity of texture assets would certainly help though, I believe fixing the mipmapping implementation on GS would have done a lot more good for overall look of most games.

given that DXTC is half the size and - in the vast majority of cases - hugely better quality.)

DXT is half the size(against 8bit) OR way better quality(against 4bit), but not both at the same time.
8bit with a GOOD quantizator easily matches the quality of DXTC on average.

 

[Megadrive1988]

I dont understand three cores either. I was thinking it would be in multipules of two.

POWER 4 and POWER 5 are both duel-core, are they not?

and you can have an MCM (multi-chip module) with 4 POWER4/5 chips. giving you 8 cores.

that's from my extremely limited understanding of IBM's POWER4/5 family anyway.

 

[zurich]

Another thought:

Could MS put the necessary guts in the XB2 for backwards compatibility, but leave it disabled? That way if consumers DID want to play XB1 games, they could buy the 'software' to enable it? That way NVIDIA would only be paid on a need-to-have basis.. think DVD playback kit, where MS left it disabled to avoid paying fees to the DVD consortium.

It could be as simple as a bootdisk or a controller adaptor or something.

 

[bbot]

PPC 976. My guess, based on the rumor posted on macosrumors.com, turned out to be correct. Which mean there will be a total of six microprocessor cores. Dave Baumann is correct too about what he heard about there being six of something.

That article on teamxbox.com should say 12 threads, not 6 threads.

 

[Megadrive1988]

why not just go for a full 4-way POWER5 MCM - each chip has 2 cores.
giving you 8 cores total.

oh wait, its not POWER5 ? im confused.

 

[nAo]

The funny thing is I'd agree that Xbox isn't that much more powerful than PS2, but I'd disagree that nobody notices the difference. I find the texture quality on PS2 games is generally poor. Not licensing S3TC was a truly massive mistake.

The real big problem with PS2 texture lies in mipmaps selection. LOD calculation is..I don't know how to define it...pure crap!

ciao,
Marco

 

[Guden Oden]

Three chips, six cores, a possible twelve logical CPUs assuming 2-way multithreading. I wanna see Deadmeat explaining to everyone EXACTLY in which way this thing will differ in programming compared to Cell, and how it will be SO much easier a beast to handle...