Not that simple. Most of the content on the disc is in xbox specific formats. Some of it may work, but not much, especially with the endian switch complicating things. You'd still need full GPU side emulation to get that stuff to work (display lists, textures, etc).
Basically you'd save having to create a CPU emulator, but getting even a few studios to port games would be hard. On top of the actual rebuilding, they'd have to go through the certification process again, and all for maybe a handful of sales.
In visual studio you can attach to a running process on another machine and debug it exactly as if it were running locally. You can step from sql to C++ to.Net (lots of languages) to DirectX shaders... You can modify the code of said running process, recompile it, and patch it back into memory without restarting it. You can even do these things if the other machine happens to be an xbox. It's a massive set of tools, and the debugger is just one insanely awesome part of it.
No doubt, for example it would be great for the center console of an aircraft (or any vehicle), so that two pilots could have different displays. However, the article was clearly talking about two people two people watching different TV channels, and I just don't think it would be more than a gimmick for entertainment.
No bullshit in regards to it actually existing, I'm not doubting that. I'm sure it's cool technology and all, I just don't see it actually being useful. I don't want to watch TV from an angle, and I can't play games properly unless I am sitting pretty squarely in front of the display, it just doesn't feel right. I'm sure I'm not alone in that. Some TVs let you use headphones to listen to the PIP channel, so you can do a similar thing, albeit sharing screen space. That's pretty much a useless gimmick, and this is just the evolution of it, better in some ways, worse in others.
As soon as you put on headphones, you are no longer doing something with other people; you may as well just have a handheld device. The only possible use I can think of for this is a replacement for split-screen in games.
-Os might actually be fastest. Smaller code = better cache utilisation. Depends on the machine, but I know with the stuff I am working on, -03 is slower than -O2 because it makes the code so much bigger.
Hardware vendors ship with Windows because it's what their customers want, a familiar, working system out of the box. Microsoft isn't paying them to do it, just giving discounts, so Linux would still be cheaper, but it's just not a viable alternative in a desktop computer (you can buy servers bundled with Linux). Until Linux steps up and offers something that consumers actually want, or Apple decides to get out of the hardware business, things aren't going to change.
Of course it's not impossible to do both, but given the limited resources available, compromises have to be made.
This branchy code that everyone is talking about, it only represents a tiny fraction of what a game does, time wise, to the point where you have things like the Unreal Engine. Most of the controller logic in Unreal is executed by an interpreted script language. UnrealScript is 10-100x slower than native code (it even says so in the docs), but it's still used, because this stuff DOESN'T TAKE VERY LONG.
Based on this, both Sony and Microsoft made the radical decision to create hardware that actually speeds up the things THAT TAKE A LONG TIME.
That could be true, and if you've got to the point where the 10% stuff approaches 50%, it's time to start building hardware for it. I just don't see that happening really, because a lot of that branchy code will end up being replaced by more abstract systems which are better suited to running in parallel.
I haven't yet worked on any next-gen titles, but I have experience on current-gen Xbox, PS2, and PC games. Our game logic has a lot of branchy code, really branchy, spaghetti code, with lots of indirect branches, etc. This code represents the majority of the actual volume of our source, but in practice only takes something like 3-10% of CPU time. Everything else in the game is suited to parallelisation.
So we have this 10/90 split (at worst) of time spent excecuting the wacky code vs. the pure number crunching. Why on earth would I want a processor that is better at doing the 10, when I could have one that is better at the rest?
Who cares about first party games? How about if Sony pictures made films that only played on Sony DVD players? Nintendo should be respected as a game developer, but what is the point in clinging to a proprietary platform? I say let game developers make games, and let console developers make consoles.
The Cell gets it's speed by assuming it is passed in order instructions, unlike almost any other type of CPU. It will choke on anything out of order. Sony better make a kick ass compiler, as I don't see game developers to keen on making assembly langauge level tweaks on something they are unfamilliar with.
First of all, your description of in-order execution is basically nonsense. All processors are passed instructions in-order. The difference with out-of-order processors is that they can reorder the instructions on the fly. I think Sony/IBM/Whoever have the right idea with this, let the compilers do the optimisation, not the hardware.
Slashdot Mirror
What does this have to do with Live? Isn't everything in the article stuff that has been in use for ages?
Yes, it is possible.
Not that simple. Most of the content on the disc is in xbox specific formats. Some of it may work, but not much, especially with the endian switch complicating things. You'd still need full GPU side emulation to get that stuff to work (display lists, textures, etc).
Basically you'd save having to create a CPU emulator, but getting even a few studios to port games would be hard. On top of the actual rebuilding, they'd have to go through the certification process again, and all for maybe a handful of sales.
In visual studio you can attach to a running process on another machine and debug it exactly as if it were running locally. You can step from sql to C++ to .Net (lots of languages) to DirectX shaders... You can modify the code of said running process, recompile it, and patch it back into memory without restarting it. You can even do these things if the other machine happens to be an xbox. It's a massive set of tools, and the debugger is just one insanely awesome part of it.
We could put it on our keyboards too, we just need to get rid of ^.
Maybe he doesn't care about the politics, and he really does just want the best browser. That's what I want at least.
No doubt, for example it would be great for the center console of an aircraft (or any vehicle), so that two pilots could have different displays. However, the article was clearly talking about two people two people watching different TV channels, and I just don't think it would be more than a gimmick for entertainment.
Angry! And sad. Very sad.
I would have never got the joke without you.
No bullshit in regards to it actually existing, I'm not doubting that. I'm sure it's cool technology and all, I just don't see it actually being useful. I don't want to watch TV from an angle, and I can't play games properly unless I am sitting pretty squarely in front of the display, it just doesn't feel right. I'm sure I'm not alone in that. Some TVs let you use headphones to listen to the PIP channel, so you can do a similar thing, albeit sharing screen space. That's pretty much a useless gimmick, and this is just the evolution of it, better in some ways, worse in others.
As soon as you put on headphones, you are no longer doing something with other people; you may as well just have a handheld device. The only possible use I can think of for this is a replacement for split-screen in games.
When I inhale, my sides go out, not in (well, my front at least).
Unless they want it to look like you are inhaling from the Xbox, but then they might as well have just called it Xbox 420.
-Os might actually be fastest. Smaller code = better cache utilisation. Depends on the machine, but I know with the stuff I am working on, -03 is slower than -O2 because it makes the code so much bigger.
Hardware vendors ship with Windows because it's what their customers want, a familiar, working system out of the box. Microsoft isn't paying them to do it, just giving discounts, so Linux would still be cheaper, but it's just not a viable alternative in a desktop computer (you can buy servers bundled with Linux). Until Linux steps up and offers something that consumers actually want, or Apple decides to get out of the hardware business, things aren't going to change.
Of course it's not impossible to do both, but given the limited resources available, compromises have to be made.
This branchy code that everyone is talking about, it only represents a tiny fraction of what a game does, time wise, to the point where you have things like the Unreal Engine. Most of the controller logic in Unreal is executed by an interpreted script language. UnrealScript is 10-100x slower than native code (it even says so in the docs), but it's still used, because this stuff DOESN'T TAKE VERY LONG.
Based on this, both Sony and Microsoft made the radical decision to create hardware that actually speeds up the things THAT TAKE A LONG TIME.
That could be true, and if you've got to the point where the 10% stuff approaches 50%, it's time to start building hardware for it. I just don't see that happening really, because a lot of that branchy code will end up being replaced by more abstract systems which are better suited to running in parallel.
Sort of off topic...
I haven't yet worked on any next-gen titles, but I have experience on current-gen Xbox, PS2, and PC games. Our game logic has a lot of branchy code, really branchy, spaghetti code, with lots of indirect branches, etc. This code represents the majority of the actual volume of our source, but in practice only takes something like 3-10% of CPU time. Everything else in the game is suited to parallelisation.
So we have this 10/90 split (at worst) of time spent excecuting the wacky code vs. the pure number crunching. Why on earth would I want a processor that is better at doing the 10, when I could have one that is better at the rest?
Seriously, it was nonsense.
It won't have even a tiny effect on sales. Look how few people use linux on the desktop, and you can quite easily install that without any hacking.
It can't be a very big proportion of people, you can't even rent VHS tapes at most places around here any more.
Real AI code is not branchy, the kind of hacked controller logic used in most games today is.
Who cares about first party games? How about if Sony pictures made films that only played on Sony DVD players? Nintendo should be respected as a game developer, but what is the point in clinging to a proprietary platform? I say let game developers make games, and let console developers make consoles.
If you can listen to it without realising that it's Christian, then:
a) Is it?
b: Does it matter?
'Christian music' is 100% marketing, lots of music is inspired by faith, and successful without the moniker.
The Cell gets it's speed by assuming it is passed in order instructions, unlike almost any other type of CPU. It will choke on anything out of order. Sony better make a kick ass compiler, as I don't see game developers to keen on making assembly langauge level tweaks on something they are unfamilliar with.
First of all, your description of in-order execution is basically nonsense. All processors are passed instructions in-order. The difference with out-of-order processors is that they can reorder the instructions on the fly. I think Sony/IBM/Whoever have the right idea with this, let the compilers do the optimisation, not the hardware.
It should have at least been edited.