XBox, PowerMac G5, and NT?

[AidenShaw]

another 64-bit myth

Another advantage of a 64bit cpu over a 32bit one, is the possibility to read 64 bits of data at once, which is faster than reading 64 bits in two times like you'd have to do on a 32bit cpu (depending on what you're doing).

But a 32-bit CPU can read 128-bit data! Look at the following list:

• 8-bit integers - 32-bit & 64-bit
• 16-bit integers - 32-bit & 64-bit
• 32-bit integers - 32-bit & 64-bit
• 64-bit integers - 64-bit
• 32-bit floating - 32-bit & 64-bit
• 64-bit floating - 32-bit & 64-bit
• 128-bit SIMD - 32-bit & 64-bit

The 32-bit CPU can process data in 64-bit chunks, even in 128-bit chunks. The "twice as much data" statement is a 64-bit myth.

The only difference is that the 64-bit CPU has native, single-instruction 64-bit integer operations. This is really the only advantage (other than 64-bit memory, which OS X does not support) of the 64-bit CPU, and for most programs it's not important.

So the "possibility" of reading 64-bits is not unique to a 64-bit CPU....

 

[the_mole1314]

Well, I think we might see some more OSX games, mabey more XBox2 ports to Mac, don't you think? I wonder how hard it would be to make the games for OSX. I don't think that hard. But even if it's hard, atleast developers will have Dual G5s in their hands, and mabey they'll use them beyond the XBox, and oh, mabey on the Mac perhaps?

 

[wrldwzrd89]

Windows NT for PowerPC existed...

I don't see Microsoft bringing it back...

I also don't think there's much significance to the fact that they have an NT kernel on PowerPC. Presumably it's all just for development for the Xbox.

arn

I read in the Inquirer article that it is quite possible that Microsoft is considering giving Intel the boot. Although this seems highly unlikely at first, imagine the implications for Apple and IBM's PowerPCs if MS 'switches' to PPC as their preferred processor for deploying Windows. I'd imagine that if MS were to do this, they would make it part of the transition to Windows Longhorn and include it with all the other sweeping changes to Windows (such as hopefully eliminating drive letters).

 

[ennerseed]

Man if MS sells the XBox at a loss and makes it up with license fees, it seems like a good idea to let Apple's machines play those games... They will make a sh*t load more money on fees.... well it sounds nice and easy.

 

[Frobozz]

because the xBox 2 has 3 processors in it, so one would assume that the games made for it would need more processing power that a dual G5...

I believe the processors have to be powers of 2 (1,2,4,8,16,32,64, etc...) I think the information you are quoting about 3 processors is misinterpreted because the article never mentioned 3 CPU's. I would suspect a northbridge chip or some other ISA and 2 CPU's-- resulting in 3 "processors". All would be developed by IBM of course.

 

[Bear]

I read in the Inquirer article that it is quite possible that Microsoft is considering giving Intel the boot. Although this seems highly unlikely at first, imagine the implications for Apple and IBM's PowerPCs if MS 'switches' to PPC as their preferred processor for deploying Windows. I'd imagine that if MS were to do this, they would make it part of the transition to Windows Longhorn and include it with all the other sweeping changes to Windows (such as hopefully eliminating drive letters).

I doubt that is what Microsoft plans. Unless that is why they bought VirtualPC? VirtualPC would ease the processor migration issues. (Except for games.)

I still doubt it is going to happen.

 

[Bear]

I believe the processors have to be powers of 2 (1,2,4,8,16,32,64, etc...) I think the information you are quoting about 3 processors is misinterpreted because the article never mentioned 3 CPU's. I would suspect a northbridge chip or some other ISA and 2 CPU's-- resulting in 3 "processors". All would be developed by IBM of course.

Of course it could be two GPUs and one CPU. Or some other combination of processor types.

However, multiprocessor systems come in all quantity of processors from 2 up. Just because people talk about 2 and 4 processor systems, i doesn't mean it's a power of 2. Some newer high end systems require you to add 2 or 4 processors at a time. Sun's F15K mainframe supports up to 106 processors and they can be added in 2 or 4 processor increments.

 

[DharvaBinky]

PPC Native Direct X is the key!

The real gold in this rumor, should it prove to be true, is the fact that XBox uses a DirectX 8 API for the graphics, and XBox2 should use DirectX9 or 10 (probably 9, if it's being developed now). Take a quick look at this page of an ArsTechnica analysis of OS X's graphics architecture:

http://arstechnica.com/reviews/2q00/macos-x-dp4/macos-x-dp4-4.html

Quoting:

Apple describes Core Graphics Rendering as a "black box" that converts input to PDF and then converts the PDF to various output formats. The diagram below illustrates:

Core Graphics Rendering: The PDF Black Box


The inputs for Core Graphics Rendering include PDF files, Core Graphics Rendering's own native C API, and QuickDraw. QuickTime and OpenGL also leverage Core Graphics Rendering, but to a lesser extent since those two technologies implement their own versions of certain two-dimensional graphics capabilities. The final input arrow is simply labeled "inputs." Apple explains it by stating that future APIs in the front end may be supported.

<ars commentary>I don't know about you, but one word immediately springs to mind: DirectX. Well, obviously only the portions of DirectX that deal with screen drawing, but it's clear that Apple has designed Mac OS X's graphics subsystem to be as flexible as possible. And with QuickDraw, QuickTime, and OpenGL already implemented, I'm hard-pressed to name another API (other than X, which is already in the busy hands of Darwin hackers) that would be implemented in the future. This is pure speculation at this point, of course, but if it does come to pass, remember that you heard it here first. </ars>

The output arrows include screen rendering (i.e passing pixel values to Core Graphics Services for compositing onto the screen), PDF, PostScript, and raster data.



I'm ready for SWG on Mac. Please.

Dharvabinky

 

[wrldwzrd89]

I doubt that is what Microsoft plans. Unless that is why they bought VirtualPC? VirtualPC would ease the processor migration issues. (Except for games.)

I still doubt it is going to happen.

I have my doubts too that this will happen, but I thought I should bring it to the attention of the members of MacRumors. You are correct about Virtual PC; you are also correct regarding games (I've tried running basic games on Virtual PC - the sound was choppy and the game was almost unplayable, even though it's only 2D. This was on a 667 MHz PowerBook G4.). I have to wonder, though, about Microsoft and how they will 'improve' Virtual PC (product activation, a la Windows XP, anyone?).

 

[Bear]

I have my doubts too that this will happen, but I thought I should bring it to the attention of the members of MacRumors. You are correct about Virtual PC; you are also correct regarding games (I've tried running basic games on Virtual PC - the sound was choppy and the game was almost unplayable, even though it's only 2D. This was on a 667 MHz PowerBook G4.). I have to wonder, though, about Microsoft and how they will 'improve' Virtual PC (product activation, a la Windows XP, anyone?).

Actually I can think of one thing that would speed up VirtualPC a lot. And that is to let all the system calls jump thru to PowerPC code. Less code to emulate means a faster running program. I'm not sure how much of this could be done unless a PowerPC version of Windows was the operating system running on the system itself.

And as for activation, well, I'm gonna keep a G4 system around until I don't need VirtualPC anymore. I don't even intend to apply the patches that Microsoft has provided for VirtualPC. 6.02 works well for me and since I'm not running XP under it, I don't need the latest patches to support the XP patches. (Did Microsoft cause that to happen on purpose? Force people to register with them and such?)

 

[Henriok]

I suppose they could use a later version, allthough that NT 4 is rocksolid in terms of stability and allready tested with PowerPC hardware.

Yeah well.. the NT4 kernel didn't have any support for something as mundane as USB. The NT-kernel in WinXP or later is probably more advanced in terms of memory, networking, device handling, I/O, 64-bitness, lot's of processors and so forth. And DirectX 10 will probably run much better on a fresh kernel than it would on an 8 year old one.

 

[Frobozz]

Of course it could be two GPUs and one CPU. Or some other combination of processor types.

However, multiprocessor systems come in all quantity of processors from 2 up. Just because people talk about 2 and 4 processor systems, i doesn't mean it's a power of 2. Some newer high end systems require you to add 2 or 4 processors at a time. Sun's F15K mainframe supports up to 106 processors and they can be added in 2 or 4 processor increments.

Ahhh. I see, said the blind man.

 

[wrldwzrd89]

Actually I can think of one thing that would speed up VirtualPC a lot. And that is to let all the system calls jump thru to PowerPC code. Less code to emulate means a faster running program. I'm not sure how much of this could be done unless a PowerPC version of Windows was the operating system running on the system itself.

And as for activation, well, I'm gonna keep a G4 system around until I don't need VirtualPC anymore. I don't even intend to apply the patches that Microsoft has provided for VirtualPC. 6.02 works well for me and since I'm not running XP under it, I don't need the latest patches to support the XP patches. (Did Microsoft cause that to happen on purpose? Force people to register with them and such?)

The problem with that idea is that Windows programs expect to be using x86 instructions, which must be translated to PowerPC instructions. Virtual PC would speed up quite a bit if the conversion routine could be optimized more (surely VPC spends a lot of time in this one routine). Lookup tables, binary search trees, and dynamic recompilation are all methods that could possibly speed up emulation.

Unfortunately, I have to put up with any activation Microsoft includes in Virtual PC since I don't have the option of running VPC with anything other than Windows XP. Actually, I don't use VPC currently (the copy I once had now belongs to another family member), so if I need VPC for something, I'd have to buy a whole new copy. Even if that were the case, I'd consider buying a real Windows box and RDC-ing (Remote Desktop Connection) to it.

 

[jeffmc425]

The Inquirer reported over the weekend that the Software Development Kit for the upcoming Xbox 2 is being seeded to developers "on dual Apple Power Mac G5 systems running a custom Windows NT Kernel."

It was previously announced that Microsoft had chosen the PowerPC as the processor for their upcoming gaming system.

Meanwhile, Windows NT for PowerPC is something that existed in the past, but was dropped.

Remember, MS had NT running on a Big Endian, as well as a Little Endian Power PC back in 1994. That whole project was shelved, but the sources stayed int eh source pool. Resurecting and upgrading the system specific HAL and drivers is not a big effort. It is the compiler

 

[pgwalsh]

Windows NT for PowerPC existed...

I don't see Microsoft bringing it back...

I also don't think there's much significance to the fact that they have an NT kernel on PowerPC. Presumably it's all just for development for the Xbox.

arn

I have a copy of Windows NT for PPC. Never tried to install it on PPC though. No point.

 

[jeffmc425]

But a 32-bit CPU can read 128-bit data! Look at the following list:

• 8-bit integers - 32-bit & 64-bit
• 16-bit integers - 32-bit & 64-bit
• 32-bit integers - 32-bit & 64-bit
• 64-bit integers - 64-bit
• 32-bit floating - 32-bit & 64-bit
• 64-bit floating - 32-bit & 64-bit
• 128-bit SIMD - 32-bit & 64-bit

The 32-bit CPU can process data in 64-bit chunks, even in 128-bit chunks. The "twice as much data" statement is a 64-bit myth.

The only difference is that the 64-bit CPU has native, single-instruction 64-bit integer operations. This is really the only advantage (other than 64-bit memory, which OS X does not support) of the 64-bit CPU, and for most programs it's not important.

So the "possibility" of reading 64-bits is not unique to a 64-bit CPU....

Most likely, MS would use canonical logword support in the PPC to support the 64 Bit CPU on the 32 bit OS. Remember the DEC Alpha? In 1991-1995, it was supported by MS under NT. it was a 64 BIT cpu, but also had support for canonical longword addresses. So 32 bit pointers were simply "sign-extended" so 32 BIT OS's could run on it. VMS, WinNT and Linux were 32 BIT OS's that ran on it. Ture64 Unix was the only 64 bit OS at the time that Alpha supported.

Now, given all that, if everyone codes nicely and lives byt the rules of variable arguments and such, all things work with a 32 BIT OS on a 64 Bit machine. But if you try to get cleve, you will find the brick wall quickly.

Jeff

 

[wrldwzrd89]

Most likely, MS would use canonical logword support in the PPC to support the 64 Bit CPU on the 32 bit OS. Remember the DEC Alpha? In 1991-1995, it was supported by MS under NT. it was a 64 BIT cpu, but also had support for canonical longword addresses. So 32 bit pointers were simply "sign-extended" so 32 BIT OS's could run on it. VMS, WinNT and Linux were 32 BIT OS's that ran on it. Ture64 Unix was the only 64 bit OS at the time that Alpha supported.

Now, given all that, if everyone codes nicely and lives byt the rules of variable arguments and such, all things work with a 32 BIT OS on a 64 Bit machine. But if you try to get cleve, you will find the brick wall quickly.

Jeff

I don't believe the PowerPC G5 needs to use canonical longword pointers because of its ability to run 32-bit code natively. That being the case, a port of the most recent Windows NT kernel for PowerPC should run just fine on the PowerPC G5, as well as previous PowerPCs. Remember, the PowerPC G5, like AMD's Athlon64, is unlike 64-bit CPUs before it in that it can run 32-bit code with no emulation or performance overhead compared to 32-bit CPUs.

 

[AidenShaw]

That whole project was shelved, but the sources stayed int eh source pool. Resurecting and upgrading the system specific HAL and drivers is not a big effort. It is the compiler

It was "shelved" after being released and supported - it wasn't a development effort that was stopped before shipment. (You probably know that, but it needed clarification.)

And, BTW, Visual Studio PPC backends exist for Windows CE ("Pocket PC") targets today....

 

[concentricity]

I agree to most of your assumptions. I guess that the important question will be if there is an easy way of compiling the games under OS X as well - what I believe and if it will be done. And the second part is questionable. If there are many games available for Macs it would definitly increase the overall value of Macs and does M$ wants this to happen ?

I think the equotation is how much many can they make with games sold for Macs in comparison to how much money the lose because more people switch to Macs and are not running a copy of Windows.

Do you think MS would rather have people switch from "Wintel" to Linux, or Wintel to Mac? At least if they switch to Mac, MS is going to be selling them stuff. MS won't be selling anything to the linux switchers.

 

[MorganX]

G4 Macs were used to develop the first GameCube games on and a shop in London has a combined Mac and NES that people used to use to program NES games on.

Although the if this Xbox2 thing is true, then it will hopefully mean it will be easy to port Xbox2 games to the Mac in future and maybe an emulator that will run well on Macs.

I don't think it means a port will be easy, but, I don't see why Xbox2 Game DVDs couldn't boot into the Xbox kernel and run on any G5, if that was desired by the developers.

One thing is for certain, Apple will sell some additional hardware. Maybe not significant amounts, but having a G5 machine in every major game developers studios doing major next generation game development can't be bad forApple.

 

[agreenster]

Okay, but this is NOT going to fit in my entertainment center.

That is some hilarious photoshop work. Nice job! Now why didnt I think of that?

Nice to see IBM doing so well. That only means good things for future processors and Apple.

 

[ryaxnb]

Maybe this will make it easier for game developers (i.e. not hackers) to port their games.

 

[AidenShaw]

Remember the DEC Alpha? In 1991-1995, it was supported by MS under NT. it was a 64 BIT cpu, but also had support for canonical longword addresses. So 32 bit pointers were simply "sign-extended" so 32 BIT OS's could run on it. VMS, WinNT and Linux were 32 BIT OS's that ran on it.

Actually, NT 3.1 was released in 1993, and the Alpha didn't release until Fall '92.

Were you running NT on an EV-3?


McLeman?

 

[TrenchMouth]

I really dont feel that this will do anything for the Mac gaming sene, and that doesnt really bother me, my Mac is great with just one or two games that i like to play.
I think the real winners here are the hardware and those that like the Xbox. it has really proven itself as a console so i am looking forward to a new installation. if its going to incorperate PPC CPUs then great, they worked wonders for the GCN and i cant wait to see what they do for the xbox.
do not confuse this as a win for Apple or MS even. Its a step in the right direction for IBM and ATi who, so far as speculation is concerned, are going to be supplying the hardware for the entire next generation of respectable consoles. With this in mind I think we can all get over which is the best console and get on with the real deal, which is making a playing great games. Its nice that some Apple hardware is in the mix, but i would take it at face value and nothing more. Save up some money, all 3 should be out within 2 years...

 

[welborn]

Trip G5s, Direct X for PPC and video cards (be vs. le)

1. I've never heard of anyone refer to a machine's processors as anything other than the main CPU. If someone says "three processors", I know that they're not including the I/O, clock, GPU, MMU or other separate processors.
(Or my old IIfx would've been like a 6-processor system with its old 6502s managing ADB and such)

One question is ... wouldn't it be useful for Apple to put out a high-end triple-G5 system? This is for two reasons: to be more useful as a development platform for Xbox2, and to beat them to the speed punch. Hell-- maybe a quad, just for the hell of it.

2. Direct X for powerPC is great in theory, but it'll probably only run as interfacing with the modified NT kernel. It probably won't help much with Mac gaming (at least, not without lots of extra development to make it work in the MacOS X world).

3. Anyone know whether the big-endian vs. little-endian is a software or Hardware difference? If it's hardware, perhaps XBox 2 will help lower video card prices for Macs. (That is the big difference between PC and Mac video cards, right?)