PowerPC confirmed for Xbox-2 ?

[Sprite Scaler]

Up until today (at least for me) it seemed Xbox-2's CPU core was up in
the air. While it was known that IBM would be fabbing / manufacturing
the CPU, I think it was somewhat uncertain who would be *designing*
it. it could have been an AMD-designed, IBM-manufactured chip. but
now it seems it *will* be PowerPC-based.



-------------------------------------------------------------------------------
IBM claims Xbox2 deal a defeat for Intel

Intel insouciance may conceal deep concern


By INQUIRER staff: Friday 07 November 2003, 07:57

CHIP FIRM Intel may think that Microsoft not using its chips is a
matter of absolute indifference, but an internal email from Big Blue
suggests the boot is on the other foot.
Sources have shown us how IBM announced the deal inside the firm, in
this memorandum below.

According to IBM, not only has it "thrown them [Intel] out of Xbox",
but it also kept them [Intel] out of the Apple G5.

The memo appears to show that if there's one company that bears a
grudge, that's Big Blue. As our own Good Doctor Spinola has it: "Life
is too short not to wreak a bitter and bloody revenge".

You may also want to read Eva Glass' story from March 2002 about
Microsoft and the Xbox2, which discusses cutting out Nvidia, and any
X86 manufacturer.


( here's the interesting part )

'A few minutes ago we announced that we have won a deal to design,
develop and make the processor for the next generation of Microsoft's
Xbox game system.

Microsoft selected us for one very simple reason -- we are the leaders
in the advanced technologies they need for Xbox.

This is a major win for us on several fronts:

It illustrates and capitalizes on our deep technology and service
capabilities. The new Xbox technologies will be based on the latest in
IBM's family of state-of-the-art processors;

It is the single largest win for Engineering & Technology Services and
a substantial foundry win for our 300 mm facility;

We've handed Intel another defeat. Earlier this year, we kept them out
of the Apple G5 and now we've thrown them out of Xbox. (Not bad,
considering one Intel executive recently called us "trivial");

We are now the undisputed leader in providing advanced chip technology
for the gaming industry. We make the CPUs for Nintendo's GameCube
systems. We are shipping in volume the GeForce FX advanced graphics
processor for NVIDIA, the premier supplier of graphics chips for the
gaming industry. We are working with Sony Computer Entertainment to
develop the processor for its PlayStation3 system. And now we've won
the next-generation Xbox.'
-------------------------------------------------------------------------------

http://www.theinquirer.net/?article=12552


Anyone disagree?

[No.6]

All this wondering what CPU will xbox 2 have is meaningless. Just enjoy your
xbox that you now have. Are you that bored of the xbox all ready. If you are
well I suggest find some other hobby to make you happy.

[Gorf]

"Sprite Scaler" <spritescaler@directvinternet.com> wrote in message
news:8dca4a2e.0311070759.27562c8b@posting.google.com... Up until today (at least for me) it seemed Xbox-2's CPU core was up in the air. While it was known that IBM would be fabbing / manufacturing the CPU, I think it was somewhat uncertain who would be *designing* it. it could have been an AMD-designed, IBM-manufactured chip. but now it seems it *will* be PowerPC-based.

Heh... wonder if people will STILL call it a PC, or whether they'll call it
a Mac ;)


--
Gorf

Hoc nomen meum verum non est

[Michael Cargill]

"No.6" <nospam@forme.net> wrote in message
news:NoPqb.3175$y95.1666@nwrdny01.gnilink.net... All this wondering what CPU will xbox 2 have is meaningless. Just enjoy
your xbox that you now have. Are you that bored of the xbox all ready. If you
are well I suggest find some other hobby to make you happy.

What are you on about? Talking about Xbox 2 is no different to talking
about games that are not out yet.




--
Craftily Guffed by Michael Cargill
----------------------------------
Eat My Cheese!
Go, Dog. Go!

[Maniac]

how do you get power pc from that?

On 7 Nov 2003 07:59:24 -0800, spritescaler@directvinternet.com (Sprite
Scaler) wrote:
Up until today (at least for me) it seemed Xbox-2's CPU core was up inthe air. While it was known that IBM would be fabbing / manufacturingthe CPU, I think it was somewhat uncertain who would be *designing*it. it could have been an AMD-designed, IBM-manufactured chip. butnow it seems it *will* be PowerPC-based.

[Bagpuss]

On Fri, 07 Nov 2003 10:27:32 -0800, Maniac <maniac@nospam.com> wrote:
how do you get power pc from that?

Well its either power pc (RISC chip, e.g. a G5) based or cyrix based
and feck me I don't think they will be using a cyrix processor :-)
--
GSX600F - Matilda the cracked teapot, gaffer tape included
I'm not a nerd, I'm an empowered freak
Wipe off cat poo to reply

[Maniac]

it could also be x86, as in amd. there's nothing in there that says
they will go power pc or risc; so it's all still conjecture at this
point. this article says nothing new.

On Fri, 07 Nov 2003 18:53:55 +0000, Bagpuss
<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:
On Fri, 07 Nov 2003 10:27:32 -0800, Maniac <maniac@nospam.com> wrote:how do you get power pc from that?Well its either power pc (RISC chip, e.g. a G5) based or cyrix basedand feck me I don't think they will be using a cyrix processor :-)

[Bagpuss]

On Fri, 07 Nov 2003 12:30:51 -0800, Maniac <maniac@nospam.com> wrote:
it could also be x86, as in amd. there's nothing in there that saysthey will go power pc or risc; so it's all still conjecture at thispoint. this article says nothing new.

But then they still have the issue of NVidia backwards compatiblity in
the ATI graphics chip set. Its not just as simple as "its DX API" you
have to take into account different pixel shaders and coders taking
into account features of the NVidia chipset.
--
GSX600F - Matilda the cracked teapot, gaffer tape included
I'm not a nerd, I'm an empowered freak
Wipe off cat poo to reply

[Maniac]

On Fri, 07 Nov 2003 21:52:02 +0000, Bagpuss
<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:
On Fri, 07 Nov 2003 12:30:51 -0800, Maniac <maniac@nospam.com> wrote:it could also be x86, as in amd. there's nothing in there that saysthey will go power pc or risc; so it's all still conjecture at thispoint. this article says nothing new.But then they still have the issue of NVidia backwards compatiblity inthe ATI graphics chip set. Its not just as simple as "its DX API" youhave to take into account different pixel shaders and coders takinginto account features of the NVidia chipset.

well, 'compatibility' really wasn't discussed, but yea, i agree.
depends on how 'custom' the chip is, and honestly, i don't know
whether software written for xbox goes thru an api or it writes
directly to hardware. i always assumed it was via api...

[George]

"Maniac" <maniac@nospam.com> wrote in message
news:uvonqvopcppj9tc2uadv6kjt58ijqcdth4@4ax.com... how do you get power pc from that?

"The new Xbox technologies will be based on the latest in IBM's family of
state-of-the-art processors;"

Unless they've something new up their sleeve the PowerPC chips would seem to
be their current state of the art processors.

There's a bit more info on the story here:
http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2493
including speculation that XBox 2 may not be backwardly compatible.

I've wondered for a while, (since both announced ATI as partners for the
next gen) whether Xbox 2 and Gamecube 2 are one in the same. I thought it
would be a conflict of interests supplying both. Who would get priority?
Then Microsoft announce IBM as the cpu supplier, the same supplier of cpus
for the Gamecube. Perhaps as a partner Nintendo have convinced Microsoft of
the PowerPC's benefits. Isn't speculation great? :-)
--
Remove N0_SPAM_PLEAZE to email
It's a spring thing
http://www.hookeslaw.com

[Maniac]

On Sat, 8 Nov 2003 02:02:02 -0000, "George"
<newzgroupzN0_SPAM_PLEAZE@ntlworld.com> wrote:
"Maniac" <maniac@nospam.com> wrote in messagenews:uvonqvopcppj9tc2uadv6kjt58ijqcdth4@4ax.com... how do you get power pc from that?"The new Xbox technologies will be based on the latest in IBM's family ofstate-of-the-art processors;"Unless they've something new up their sleeve the PowerPC chips would seem tobe their current state of the art processors.

amd, the possibility of a new or custom design, and there's always the
far fetched idea of cell ;p
There's a bit more info on the story here:http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2493including speculation that XBox 2 may not be backwardly compatible.

heh.. well, what was you first clue? ati? ibm? or ???

there's still always the chance it will be.. or maybe it won't. we
just don't know enough, and i haven't seen any information of
substance that would allow decent speculation.
I've wondered for a while, (since both announced ATI as partners for thenext gen) whether Xbox 2 and Gamecube 2 are one in the same. I thought itwould be a conflict of interests supplying both. Who would get priority?Then Microsoft announce IBM as the cpu supplier, the same supplier of cpusfor the Gamecube. Perhaps as a partner Nintendo have convinced Microsoft ofthe PowerPC's benefits. Isn't speculation great? :-)

and the supplier of 'cell' for sony (toshiba is involved also)... i
doubt that anything would be a 'std' chip shared by any console
manufacturer. with this much lead time i would expect custom chips
even if based on similar architecture.

sure, it's fun to speculate, but in the end it just doesn't matter if
the end product delivers. :)

[Company Man]

"George" <newzgroupzN0_SPAM_PLEAZE@ntlworld.com> wrote in message
news:bohimh$1dpujn$1@ID-176063.news.uni-berlin.de... "Maniac" <maniac@nospam.com> wrote in message news:uvonqvopcppj9tc2uadv6kjt58ijqcdth4@4ax.com... how do you get power pc from that? "The new Xbox technologies will be based on the latest in IBM's family of state-of-the-art processors;" Unless they've something new up their sleeve the PowerPC chips would seem
to be their current state of the art processors. There's a bit more info on the story here: http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2493 including speculation that XBox 2 may not be backwardly compatible. I've wondered for a while, (since both announced ATI as partners for the next gen) whether Xbox 2 and Gamecube 2 are one in the same. I thought it would be a conflict of interests supplying both. Who would get priority? Then Microsoft announce IBM as the cpu supplier, the same supplier of cpus for the Gamecube. Perhaps as a partner Nintendo have convinced Microsoft
of the PowerPC's benefits. Isn't speculation great? :-) -- Remove N0_SPAM_PLEAZE to email It's a spring thing http://www.hookeslaw.com

OT to this thread, but interesting nonetheless, is the interview on this
same site with Ed Fries (head of Xbox marketing and development) at some
kinda conclave called 'X03'

http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2267

wherein it is revealed that the 'box now has such strong developer support
that Softee is now narrowing the focus of their inhouse development efforts
to only those titles which will be revolutionary and groundbreaking.
<#insert pithy raconteur-like observation here#>

[Bagpuss]

On Fri, 07 Nov 2003 16:37:42 -0800, Maniac <maniac@nospam.com> wrote:
On Fri, 07 Nov 2003 21:52:02 +0000, Bagpuss<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:On Fri, 07 Nov 2003 12:30:51 -0800, Maniac <maniac@nospam.com> wrote:it could also be x86, as in amd. there's nothing in there that saysthey will go power pc or risc; so it's all still conjecture at thispoint. this article says nothing new.But then they still have the issue of NVidia backwards compatiblity inthe ATI graphics chip set. Its not just as simple as "its DX API" youhave to take into account different pixel shaders and coders takinginto account features of the NVidia chipset.well, 'compatibility' really wasn't discussed, but yea, i agree.depends on how 'custom' the chip is, and honestly, i don't knowwhether software written for xbox goes thru an api or it writesdirectly to hardware. i always assumed it was via api...

Personally I doubt its as much DX API as everyone thinks. The _big_
thing about the NVIDIA chipset is that you can upload your own
microcode for the DX Shaders. At the time of the XBox release the DX
API had no cross platform code standard for the microcode. Now there
is a standard, but I can't remeber if its released yet. There are two
that I know of one for NVIDIA (cgi) and the ATI one (rendermonkey? or
something). The microcode for the two chipsets is different and I can
imagine investing in compatibility being somehing that ATI would not
want to spend a great deal on.
--
GSX600F - Matilda the cracked teapot, gaffer tape included
I'm not a nerd, I'm an empowered freak
Wipe off cat poo to reply

[Tony Hill]

On Sat, 08 Nov 2003 10:53:16 +0000, Bagpuss
<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:On Fri, 07 Nov 2003 16:37:42 -0800, Maniac <maniac@nospam.com> wrote:well, 'compatibility' really wasn't discussed, but yea, i agree.depends on how 'custom' the chip is, and honestly, i don't knowwhether software written for xbox goes thru an api or it writesdirectly to hardware. i always assumed it was via api...Personally I doubt its as much DX API as everyone thinks. The _big_thing about the NVIDIA chipset is that you can upload your ownmicrocode for the DX Shaders. At the time of the XBox release the DXAPI had no cross platform code standard for the microcode. Now thereis a standard, but I can't remeber if its released yet. There are twothat I know of one for NVIDIA (cgi) and the ATI one (rendermonkey? orsomething). The microcode for the two chipsets is different and I canimagine investing in compatibility being somehing that ATI would notwant to spend a great deal on.

This sort of thing is something that Microsoft could handle, you just
need to throw a little bit of processing power at the job, but nothing
too serious. MS would just need to trap any calls to these shadders
and translate them into something that the ATI chip could understand.
Given that the XBox2 is to use a fair bit more powerful hardware than
the original XBox, it shouldn't be too much of a problem. The odd
game wouldn't work, and some games might work but have slight visual
defects every now and then, but for the most part this could be made
to work without too much difficulty.

However, if you're changing the processor as well, that adds another
level of complexity. You can still do the above, and you can take
care of the difference in processors through a combination of
recompiling the API and emulation for non-API calls, but this also
takes some extra processing umph. Just how much processing power you
need to do the emulation depends on just how well optimized your APIs
and emulation are (read: how much money you spend on it).

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

[Bagpuss]

On Sat, 08 Nov 2003 23:56:46 GMT, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:
On Sat, 08 Nov 2003 10:53:16 +0000, Bagpuss<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:On Fri, 07 Nov 2003 16:37:42 -0800, Maniac <maniac@nospam.com> wrote:well, 'compatibility' really wasn't discussed, but yea, i agree.depends on how 'custom' the chip is, and honestly, i don't knowwhether software written for xbox goes thru an api or it writesdirectly to hardware. i always assumed it was via api...Personally I doubt its as much DX API as everyone thinks. The _big_thing about the NVIDIA chipset is that you can upload your ownmicrocode for the DX Shaders. At the time of the XBox release the DXAPI had no cross platform code standard for the microcode. Now thereis a standard, but I can't remeber if its released yet. There are twothat I know of one for NVIDIA (cgi) and the ATI one (rendermonkey? orsomething). The microcode for the two chipsets is different and I canimagine investing in compatibility being somehing that ATI would notwant to spend a great deal on.This sort of thing is something that Microsoft could handle, you justneed to throw a little bit of processing power at the job, but nothingtoo serious. MS would just need to trap any calls to these shaddersand translate them into something that the ATI chip could understand.

Not quite so simple unless they use DX Apis to upload the code even
then you have to make sure they are not doing something funny in the
code that could trip up the translator. You upload the code to the
shader processors. It then becomes code outside the OS or kernel
control. The ATI chipset could translate but you then have the chance
of clock cycles being different potentially causing problems.
Given that the XBox2 is to use a fair bit more powerful hardware thanthe original XBox, it shouldn't be too much of a problem. The oddgame wouldn't work, and some games might work but have slight visualdefects every now and then, but for the most part this could be madeto work without too much difficulty.

I think its a lot more work than many people believe. Its getting
accurate emulation thats the problem not just getting it to work.
However, if you're changing the processor as well, that adds anotherlevel of complexity. You can still do the above, and you can takecare of the difference in processors through a combination ofrecompiling the API and emulation for non-API calls, but this alsotakes some extra processing umph. Just how much processing power youneed to do the emulation depends on just how well optimized your APIsand emulation are (read: how much money you spend on it).

You if you want the console to either a) cost a fortune or b) be
limited by the need for backwards compatibility then it can be done.

The PS2 was marketed with its backwards compatibility as a big
feature. Before the media told us it was important no one gave a shit.
No doubt MS will advertise it and talk about how great and powerful
the new machine is rendering old games obsolete and backwards
compatbility will no longer be necessary as it removes the freedom to
innovate and all that crap.

No matter what you can still gauranatee one thing. No matter how good
the console is it will still be crippled by games being stright ported
from the PS3 :-)
--
GSX600F - Matilda the cracked teapot, gaffer tape included

Wipe off cat poo to reply

[Emilio Lizardo]

In article <gtlsqvkmhk68170lqdr5edkmo2o7qi2afe@4ax.com>,
Bagpuss <bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:
No matter what you can still gauranatee one thing. No matter how good the console is it will still be crippled by games being stright ported from the PS3 :-) --

What a fucking moron.

No wonder you were brain dead enough to buy a FlopBox.

[Tony Hill]

On Sun, 09 Nov 2003 15:11:36 +0000, Bagpuss
<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:On Sat, 08 Nov 2003 23:56:46 GMT, Tony Hill <hilla_nospam_20@yahoo.ca>wrote:This sort of thing is something that Microsoft could handle, you justneed to throw a little bit of processing power at the job, but nothingtoo serious. MS would just need to trap any calls to these shaddersand translate them into something that the ATI chip could understand.Not quite so simple unless they use DX Apis to upload the code even

They don't need the code to be uploaded by DirectX, they just need to
trap it. Microsoft knows exactly how their hardware works, and they
can trap any instructions they so chose. It's not like the software
uses some dark magic to get code from the disk to the GPU. Even if
they aren't using DirectX, it's still going through the CPU and the
operating system, even if only somewhat peripherally.
then you have to make sure they are not doing something funny in thecode that could trip up the translator.

That's the hard part, and that's why it won't work on 100% of games
and why you will have some visual artifacts from time to time.
However, it should be easy to handle for MOST games, even if there are
one or two odd-balls that are broken. If the game is popular enough,
Microsoft just needs to study it for a bit and fix/work around any
problems, ie they just need to throw money at it.
You upload the code to theshader processors. It then becomes code outside the OS or kernelcontrol. The ATI chipset could translate but you then have the chanceof clock cycles being different potentially causing problems.

Little problem here, little problem there, nothing that can't be fixed
by studying what the code is and what problem it is causing. Again,
all it takes is time and money, it's just a question of how much money
MS wants to invest in this.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

[gamechat66]

x-no-archive: yes

"Tony Hill" <hilla_nospam_20@yahoo.ca> wrote in message
news:fc887cab444d116d1a157c59af16adff@news.1usenet.com... However, if you're changing the processor as well, that adds another level of complexity. You can still do the above, and you can take care of the difference in processors through a combination of recompiling the API and emulation for non-API calls,

The first thing, getting the ATI to emulate the NVidia, seems doable. The
second thing, recompiling the API (and OS) seems doable too, but still that
would only allow it to run XBOX2 games natively (since they'll be compiled
for that same platform). In fact, they MUST do the latter if they switch
architectures, whether they want backward-compatibility or not; although it
will probably be a newer OS and newer API.

The hard part, that perhaps you've overlooked, is the first gen XBOX games
are IA32 (x86) binaries and the latter wouldn't help them at all: They
would need to create an IA32 VM running inside the new OS. They'd probably
not only need to have that, but a complete XBOX1 VM actually. And that's
where the whole idea goes from seemingly doable to near impossible, imho.
I'm certainly no expert on these things, but I don't see how that next-gen
processor is going to have the horsepower to run an XBOX1 VM inside the new
OS and acheive anything like original speed, with the ATI having to emulate
the NVidia further exacerbating the problem. But what do I know...

[Sprite Scaler]

"George" <newzgroupzN0_SPAM_PLEAZE@ntlworld.com> wrote in message news:<bohimh$1dpujn$1@ID-176063.news.uni-berlin.de>... "Maniac" <maniac@nospam.com> wrote in message news:uvonqvopcppj9tc2uadv6kjt58ijqcdth4@4ax.com... how do you get power pc from that? "The new Xbox technologies will be based on the latest in IBM's family of state-of-the-art processors;" Unless they've something new up their sleeve the PowerPC chips would seem to be their current state of the art processors. There's a bit more info on the story here: http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2493 including speculation that XBox 2 may not be backwardly compatible. I've wondered for a while, (since both announced ATI as partners for the next gen) whether Xbox 2 and Gamecube 2 are one in the same. I thought it would be a conflict of interests supplying both. Who would get priority? Then Microsoft announce IBM as the cpu supplier, the same supplier of cpus for the Gamecube. Perhaps as a partner Nintendo have convinced Microsoft of the PowerPC's benefits. Isn't speculation great? :-)


While it's easy to think that Xbox 2 and Gamecube 2 are one and the
same, they aren't. Xbox 2 GPU is being designed at ATI's Marlborough
office, while Gamecube 2 GPU is being done at ATI's Santa Clara
offices. ATI stated some months ago that they're both being designed
with different goals in mind.

Xbox 2 CPU is now known to be an IBM-designed & IBM-manufactured core.
Gamecube2's CPU is unknown. it's rumored/reported to be an NEC effort
possibility in colaboration with other companies such as Cray,
Hitachi, Fujuitsu, Matsushita. or it might be an IBM PowerPC/ core
again.

[George]

"Sprite Scaler" <spritescaler@directvinternet.com> wrote in message
news:8dca4a2e.0311100547.3b38bce3@posting.google.com... "George" <newzgroupzN0_SPAM_PLEAZE@ntlworld.com> wrote in message
news:<bohimh$1dpujn$1@ID-176063.news.uni-berlin.de>... "Maniac" <maniac@nospam.com> wrote in message news:uvonqvopcppj9tc2uadv6kjt58ijqcdth4@4ax.com... how do you get power pc from that? "The new Xbox technologies will be based on the latest in IBM's family
of state-of-the-art processors;" Unless they've something new up their sleeve the PowerPC chips would
seem to be their current state of the art processors. There's a bit more info on the story here: http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2493 including speculation that XBox 2 may not be backwardly compatible. I've wondered for a while, (since both announced ATI as partners for the next gen) whether Xbox 2 and Gamecube 2 are one in the same. I thought
it would be a conflict of interests supplying both. Who would get
priority? Then Microsoft announce IBM as the cpu supplier, the same supplier of
cpus for the Gamecube. Perhaps as a partner Nintendo have convinced
Microsoft of the PowerPC's benefits. Isn't speculation great? :-) While it's easy to think that Xbox 2 and Gamecube 2 are one and the same, they aren't. Xbox 2 GPU is being designed at ATI's Marlborough office, while Gamecube 2 GPU is being done at ATI's Santa Clara offices. ATI stated some months ago that they're both being designed with different goals in mind. Xbox 2 CPU is now known to be an IBM-designed & IBM-manufactured core. Gamecube2's CPU is unknown. it's rumored/reported to be an NEC effort possibility in colaboration with other companies such as Cray, Hitachi, Fujuitsu, Matsushita. or it might be an IBM PowerPC/ core again.

Interesting but I can see problems ahead. What happens if both consoles
take off big time and both companies need more processors than ATI can
produce? Who gets priority? Seems a conflict of interests to me. I
thought Nintendo mentioned they would be working with IBM too but then it
all went quiet. Probably adding two and two together to get five but as I
say it's nice to speculate. :-)
--
Remove N0_SPAM_PLEAZE to email
It's a spring thing
http://www.hookeslaw.com

[Bagpuss]

On Mon, 10 Nov 2003 05:18:24 GMT, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:
On Sun, 09 Nov 2003 15:11:36 +0000, Bagpuss<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:On Sat, 08 Nov 2003 23:56:46 GMT, Tony Hill <hilla_nospam_20@yahoo.ca>wrote:This sort of thing is something that Microsoft could handle, you justneed to throw a little bit of processing power at the job, but nothingtoo serious. MS would just need to trap any calls to these shaddersand translate them into something that the ATI chip could understand.Not quite so simple unless they use DX Apis to upload the code evenThey don't need the code to be uploaded by DirectX, they just need totrap it. Microsoft knows exactly how their hardware works, and theycan trap any instructions they so chose. It's not like the softwareuses some dark magic to get code from the disk to the GPU. Even ifthey aren't using DirectX, it's still going through the CPU and theoperating system, even if only somewhat peripherally.

With a single process model you'll need hardware interrupts. On top of
that you'll have to detect activity that is occurring over the DMA
channels or specific memory addresses. Detecting CPU instructions
wouldn't be simple. What about programming the DMA to do automatic
upload? That can become CPU independant, hence you need to detect DMA
activity too which can be out of the realm of the CPU.

On top of that they may not being using the Operating System. Remeber
on the XBox the game itself becomes the OS as per other consoles.
then you have to make sure they are not doing something funny in thecode that could trip up the translator.That's the hard part, and that's why it won't work on 100% of gamesand why you will have some visual artifacts from time to time.However, it should be easy to handle for MOST games, even if there areone or two odd-balls that are broken. If the game is popular enough,Microsoft just needs to study it for a bit and fix/work around anyproblems, ie they just need to throw money at it.

Thow *lots* of money at it. Would you like to have your new console to
be priced with the old console price (and some) shoved on top to pay
for the development of emulation code.

Looking between the lines (in terms of CPU choice e.t.c ) game code
may well be .net based with a CPU that can process .net bytecodes in a
similar way to JVM rather than raw x86 or G5 instructions. MS want
this and press releases have said similar.

You upload the code to theshader processors. It then becomes code outside the OS or kernelcontrol. The ATI chipset could translate but you then have the chanceof clock cycles being different potentially causing problems.Little problem here, little problem there, nothing that can't be fixedby studying what the code is and what problem it is causing. Again,all it takes is time and money, it's just a question of how much moneyMS wants to invest in this.

I personally hope they don't want to invest any. When the XB2 comes
out you'll be able to get an XBox for £50 or so quid. I'd rather not
have the price hoiked up by 100 or 200 just for backwards
compatibility with *some* games.
--
GSX600F - Matilda the cracked teapot, gaffer tape included

Wipe off cat poo to reply

[Tony Hill]

On Mon, 10 Nov 2003 06:24:47 GMT, "gamechat66" <gamechat66@yahoo.com>
wrote:"Tony Hill" <hilla_nospam_20@yahoo.ca> wrote in messagenews:fc887cab444d116d1a157c59af16adff@news.1usenet.com... However, if you're changing the processor as well, that adds another level of complexity. You can still do the above, and you can take care of the difference in processors through a combination of recompiling the API and emulation for non-API calls,The first thing, getting the ATI to emulate the NVidia, seems doable. Thesecond thing, recompiling the API (and OS) seems doable too, but still thatwould only allow it to run XBOX2 games natively (since they'll be compiledfor that same platform). In fact, they MUST do the latter if they switcharchitectures, whether they want backward-compatibility or not; although itwill probably be a newer OS and newer API.

yup, new games will be different, no forwards compatibility. Nothing
out of the ordinary here though.
The hard part, that perhaps you've overlooked, is the first gen XBOX gamesare IA32 (x86) binaries and the latter wouldn't help them at all: Theywould need to create an IA32 VM running inside the new OS. They'd probablynot only need to have that, but a complete XBOX1 VM actually.

Yup, a complete XBox environment will be required, but a lot of the
API calls can be handled by the new native XBox2 API.

FWIW Microsoft refers to this process as "thunking", and the
architecture as a whole as "Windows on Windows", or simply 'WoW'.
They do something of this sort already with WindowsXP and Win2003 for
the Itanium. Handling the XBox will be a bit trickier because they'll
need to trap some direct hardware calls which are not possible in
regular Windows, but it is possible to do.
And that'swhere the whole idea goes from seemingly doable to near impossible, imho.

Not impossible, just not cheap. It will take a lot of developer time
and money that MS may decide is better spent elsewhere.
I'm certainly no expert on these things, but I don't see how that next-genprocessor is going to have the horsepower to run an XBOX1 VM inside the newOS and acheive anything like original speed, with the ATI having to emulatethe NVidia further exacerbating the problem. But what do I know...

Well, this is the problem, just how much horsepower will be required?
Also, it won't be constant from one game to the next, so the frame
rates could suffer on certain games. The odd game of course will
simply not work at all because it does something funny that MS didn't
take into account.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

[Tony Hill]

On Mon, 10 Nov 2003 18:42:35 +0000, Bagpuss
<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:They don't need the code to be uploaded by DirectX, they just need totrap it. Microsoft knows exactly how their hardware works, and theycan trap any instructions they so chose. It's not like the softwareuses some dark magic to get code from the disk to the GPU. Even ifthey aren't using DirectX, it's still going through the CPU and theoperating system, even if only somewhat peripherally.With a single process model you'll need hardware interrupts. On top ofthat you'll have to detect activity that is occurring over the DMAchannels or specific memory addresses. Detecting CPU instructionswouldn't be simple. What about programming the DMA to do automaticupload? That can become CPU independant, hence you need to detect DMAactivity too which can be out of the realm of the CPU.

You seem to be under the impression that the GPU is off in it's own
little world with no connection to the operating system or anything
else. It's not. Everything it does goes through the OS to a certain
degree. OS still sees that a DMA transfer will be made, even if it
doesn't touch the data itself.

Remember, the processor reads the instructions from memory, not the
GPU.
On top of that they may not being using the Operating System. Remeberon the XBox the game itself becomes the OS as per other consoles.

The game can make direct access to hardware, and yes, that is tricky.
However it doesn't become the operating system, it just skirts around
the edges of the existing OS from time to time.
That's the hard part, and that's why it won't work on 100% of gamesand why you will have some visual artifacts from time to time.However, it should be easy to handle for MOST games, even if there areone or two odd-balls that are broken. If the game is popular enough,Microsoft just needs to study it for a bit and fix/work around anyproblems, ie they just need to throw money at it.Thow *lots* of money at it. Would you like to have your new console tobe priced with the old console price (and some) shoved on top to payfor the development of emulation code.

As always, the cost depends on a lot of factors, particularly with
things like consoles where the cost of the system is not necessarily
reflected in the retail price.

I don't know how much time/money it would cost to run XBox games on
this XBox2 system. Microsoft may determine that it will cost a LOT of
money for relatively minimal benefit and not do it. Or they may
determine that the cost is sufficiently low and the benefit
sufficiently high that it's worth while.

All I'm saying is that it is possible to create emulation on the
console that will work for most (not all) games. They were going to
have to do emulation regardless of what hardware changes they made,
however if MS had stuck with an x86 processor, it would have been
relatively easy to do. Switching to a PowerPC chip makes it rather
difficult.
Looking between the lines (in terms of CPU choice e.t.c ) game codemay well be .net based with a CPU that can process .net bytecodes in asimilar way to JVM rather than raw x86 or G5 instructions. MS wantthis and press releases have said similar.

I have seen this, though I'm not sure how successful that idea will
be. Consoles typically run at pretty low-levels, either APIs that
directly access hardware or even going lower to talk to hardware
directly. Now they're talking about adding two more layers of
abstraction into the mix. It doesn't really sound like a winning idea
to me.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

[George]

"Tony Hill" <hilla_nospam_20@yahoo.ca> wrote in message
news:65793801bb0cb91b12912a63a3fc6cd0@news.1usenet.com...
And that'swhere the whole idea goes from seemingly doable to near impossible, imho. Not impossible, just not cheap. It will take a lot of developer time and money that MS may decide is better spent elsewhere.

Would it be cheaper to do what Sony did and just stick an Xbox in the
NextBox as a separate chip or set of chips?
--
Remove N0_SPAM_PLEAZE to email
It's a spring thing
http://www.hookeslaw.com

[gamechat66]

x-no-archive: yes

"Tony Hill" <hilla_nospam_20@yahoo.ca> wrote in message
news:65793801bb0cb91b12912a63a3fc6cd0@news.1usenet.com... On Mon, 10 Nov 2003 06:24:47 GMT, "gamechat66" <gamechat66@yahoo.com>The hard part, that perhaps you've overlooked, is the first gen XBOX
gamesare IA32 (x86) binaries and the latter wouldn't help them at all: Theywould need to create an IA32 VM running inside the new OS. They'd
probablynot only need to have that, but a complete XBOX1 VM actually. Yup, a complete XBox environment will be required, but a lot of the API calls can be handled by the new native XBox2 API.

Agreed. It seemed in your original post were implying that simply
recompliling everything for the new architecture would allow XBOX 1 games to
run, no problem (without regard to the other comments about ATI / nVidia).
I was just pointing out that it won't be as simple as that because they are
in IA32 binary format.
I'm certainly no expert on these things, but I don't see how that
next-genprocessor is going to have the horsepower to run an XBOX1 VM inside the
newOS and acheive anything like original speed, with the ATI having to
emulatethe NVidia further exacerbating the problem. But what do I know... Well, this is the problem, just how much horsepower will be required? Also, it won't be constant from one game to the next, so the frame rates could suffer on certain games. The odd game of course will simply not work at all because it does something funny that MS didn't take into account.

Yeah, how much indeed. Unless there is some breakthrough processor on the
horizon, I can't see it being done in the next generation. Just an opinion.
The more likely route, as others have pointed out, is simply to embed a
clone of the current processor on the board like Sony has apparently done.
Who knows the price, etc. of that and whether it would be cheaper in the
long run than trying to do it in software. It almost certainly would work
more dependably, and maybe AMD or even Intel could make them dirt cheap by
2005.

[Tony Hill]

On Tue, 11 Nov 2003 00:31:38 GMT, "gamechat66" <gamechat66@yahoo.com>
wrote: Well, this is the problem, just how much horsepower will be required? Also, it won't be constant from one game to the next, so the frame rates could suffer on certain games. The odd game of course will simply not work at all because it does something funny that MS didn't take into account.Yeah, how much indeed. Unless there is some breakthrough processor on thehorizon, I can't see it being done in the next generation. Just an opinion.The more likely route, as others have pointed out, is simply to embed aclone of the current processor on the board like Sony has apparently done.Who knows the price, etc. of that and whether it would be cheaper in thelong run than trying to do it in software. It almost certainly would workmore dependably, and maybe AMD or even Intel could make them dirt cheap by2005.

The processor in the current XBox is a "Coppermine" Celeron core
running at 733MHz. Intel and AMD can make a chip of that capability
dirt-cheap NOW. The die size of that chip would be only about 80mm^2
at the current 130nm production, and since the clock speed they could
get out of said chip is MUCH higher than 733MHz, they could probably
get by with a much smaller/simpler/cheaper die and higher clock. By
the time the XBox enters production, IBM will probably be using a 90nm
production line to make the chip, or possibly even a new 65nm line
(65nm is expected to be state of the art in late 2005/early 2006, but
consoles chips typically come from a slightly older manufacturing
process). Size of the x86 die could probably be well under 50mm^2,
and that's assuming it has separate L2 cache from the main processor.

The trick is that this doesn't exactly solve the problem, there are
still a lot of difficulties that would be involved to get the
co-processor to handle the legacy XBox software. This also really
complicates the design of the processor, which not only makes the chip
more expensive, but might also hurt it's performance in running native
code (probably not by any meaningful amount though). It's would
certainly make things a lot easier than not having an x86 chip sitting
there on-die, but adds it's own cost to the mix.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

[Bagpuss]

On Mon, 10 Nov 2003 23:17:31 GMT, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:
On Mon, 10 Nov 2003 18:42:35 +0000, Bagpuss<bagpusscatpoo.netcatpoo@ntlcatpooworld.com> wrote:They don't need the code to be uploaded by DirectX, they just need totrap it. Microsoft knows exactly how their hardware works, and theycan trap any instructions they so chose. It's not like the softwareuses some dark magic to get code from the disk to the GPU. Even ifthey aren't using DirectX, it's still going through the CPU and theoperating system, even if only somewhat peripherally.With a single process model you'll need hardware interrupts. On top ofthat you'll have to detect activity that is occurring over the DMAchannels or specific memory addresses. Detecting CPU instructionswouldn't be simple. What about programming the DMA to do automaticupload? That can become CPU independant, hence you need to detect DMAactivity too which can be out of the realm of the CPU.You seem to be under the impression that the GPU is off in it's ownlittle world with no connection to the operating system or anythingelse.

In certain circumstances yes. Same with DMA shifters.
It's not. Everything it does goes through the OS to a certaindegree. OS still sees that a DMA transfer will be made, even if itdoesn't touch the data itself.

The OS would see it if it was a multitasking OS. If the game _is_ the
OS (which is the case in the XBox) then only if the game was coded to
call the correct kernel calls could the intercept.
Now if the XB2 ran under an OS then it could do it if it sandboxed the
game.
Remember, the processor reads the instructions from memory, not theGPU.

Nope the GPU reads the instructions from its own shared segment of
memory. Have a look for pixel shaders. In particular Nvidia CGI
compiler, ASM, and ATI render monkey, you'll see where I'm comming
from in terms of GPU emulation then.
On top of that they may not being using the Operating System. Remeberon the XBox the game itself becomes the OS as per other consoles.The game can make direct access to hardware, and yes, that is tricky.

Its not that tricky. Espesh as you have with consoles a gauranteed
hardware platform.
However it doesn't become the operating system, it just skirts aroundthe edges of the existing OS from time to time.

I'm not sure what you mean by an OS in terms of consoles? Do you mean
calling the kernel functions or a proper OS that manages resources as
a govenor process like say Unix or Linux?
That's the hard part, and that's why it won't work on 100% of gamesand why you will have some visual artifacts from time to time.However, it should be easy to handle for MOST games, even if there areone or two odd-balls that are broken. If the game is popular enough,Microsoft just needs to study it for a bit and fix/work around anyproblems, ie they just need to throw money at it.Thow *lots* of money at it. Would you like to have your new console tobe priced with the old console price (and some) shoved on top to payfor the development of emulation code.As always, the cost depends on a lot of factors, particularly withthings like consoles where the cost of the system is not necessarilyreflected in the retail price.I don't know how much time/money it would cost to run XBox games onthis XBox2 system. Microsoft may determine that it will cost a LOT ofmoney for relatively minimal benefit and not do it. Or they maydetermine that the cost is sufficiently low and the benefitsufficiently high that it's worth while.All I'm saying is that it is possible to create emulation on theconsole that will work for most (not all) games. They were going tohave to do emulation regardless of what hardware changes they made,however if MS had stuck with an x86 processor, it would have beenrelatively easy to do. Switching to a PowerPC chip makes it ratherdifficult.Looking between the lines (in terms of CPU choice e.t.c ) game codemay well be .net based with a CPU that can process .net bytecodes in asimilar way to JVM rather than raw x86 or G5 instructions. MS wantthis and press releases have said similar.I have seen this, though I'm not sure how successful that idea willbe. Consoles typically run at pretty low-levels, either APIs thatdirectly access hardware or even going lower to talk to hardwaredirectly. Now they're talking about adding two more layers ofabstraction into the mix. It doesn't really sound like a winning ideato me.

Crank the processing power up to say 3Ghz, have a processor that can
run .net instructions native, add shaders and the like for rendering
and you can run the game like that. Sturmovik (?) runs the game AI in
a java virtual machine whilst the rest of the program is compiled (c++
IIRC). This is in essentially what JVM processors do in mobile phones,
the JVM is implenented in hardware.

--
GSX600F - Matilda the cracked teapot, gaffer tape included

Wipe off cat poo to reply