There's nothing about the ARM ISA that relegates it to low power and low performance. Consider a Sandy Bridge CPU with an ARM decoder instead of x86 = instant high performance ARM processor.
I really don't understand why the next XBOX and PS3 aren't jumping all over AMD Fusion.
Win 8 isnt even out yet, does this mean win 9 will be out in 2 years??
The same way they did last time?I guess I dont understand how the new architecture would work, and how they would work with backwards compatability.
I cant believe that they would not have at least one generation of backwards compatability.
Arm->x86 would not be any harder than powerpc -> pc. Basically, all the higher-level infrastructure would look the same on xbox and windows, and if you work in asm, you have to do separate codepaths.And also, how would this work with porting to the PC if they go to this new architecture.
MS has heavily pushed for higher-level programming tools for gaming for a long time now. This makes porting between different (MS) platforms easier, not harder. If you work on XNA, you don't need to care if the cpu below you is x86, arm, or powerpc.I have this terrible fear that in order to save a few bucks, microsoft would put the final nail in the coffin of PC gaming by making it almost impossible to make PC compatable games. Maybe though this could be a good opportunity for AMD to make some sort of SoC or graphics integration module.
Is it really that easy?
I thought there were other differences beyond the decoder?
Nothing.Wow, I wonder what this is going to do to the PC gaming industry?
Well, MS has already paid for an ARM architectural license: http://www.eetimes.com/electronics-news/4204863/Microsoft-takes-ARM-license
Games have more than polygons, textures, and shaders, and they do need CPU power (if nothing else, rendering a more complicated scene is going to need more CPU). Consoles have never needed CPUs as good as a high-end gaming rig, but they do need decent CPUs, if they want more interesting and immersive games. Current ARMs (A10 and A15 are unknowns) would all be worse than taking Xenon, and giving it more cache and speed bump.A large issue that people are ignoring is that *consoles do not really need good cpus these days*. They need more memory, more GPU power, but frankly, cpu is not a component that determines how good the games appear.
MS has this thing called windows phone...
Interesting though that both PS4 and Xbox loop are now rumored to be using an ARM CPU
The reason they won't go x86 is they can't have what they want - which is their own custom chip that they own design rights for to their specification. The only people that can make x86 are Intel and AMD, and they will only provide you with a finished chip designed to fit into one of their x86 motherboards of which only they are allowed to make too.
ARM has no such constraints - you license the cpu for peanuts and then you can do whatever you like, including fairly easily making your own SOC, and fabbing it wherever you want. Hence the push from many of the players who make their own hardware (apple, sony, ms) away from x86 towards ARM.
> with multiple dedicated assistive cores for graphics, AI, physics, sound, networking, encryption and sensors, according to MS Nerd blog.
A CPU core for sound, which takes almost no CPU load now? A CPU core for "AI" which is part of the general logic of the program?
Even a CPU core for physics doesn't make sense compared to having that be done by the GPU when needed and letting the GPU use that power for graphics when it is not needed.
100% BS. Clicking the link is giving them hits for a garbage post.
As long as it has a real GPU designed by Nvidia or AMD this'll be ok. I don't think ARM or PowerVR can make a GPU anywhere close in performance to a modern day PC graphics card now or in the future.
Because the fusion GPU is slow...
Games have more than polygons, textures, and shaders, and they do need CPU power (if nothing else, rendering a more complicated scene is going to need more CPU).
Current ARMs (A10 and A15 are unknowns) would all be worse than taking Xenon, and giving it more cache and speed bump.
...unless the A15 makes us go,"wow,"
, I wouldn't expect it soon.
There is almost no chance the NEXT XBOX or PS use an ARM CPU. They going with a system as powerful as the new PSP? MS would be much better off with a 28NM XENOS with an extra core. They already own the license and have development tools around it. The next XBOX will have some kind of IBM CPU in it. Any ARM coming out soon is not going to even be remotely powerful enough. Even the big A15 is still only going to be in the league of the freaking ATOM at much lower power level.
XBOX has never had backcompat. Then again, I still think that PowerPC 470S would be the better pick for the console.i just don't know why you'd break compatibility between current 360 games and the next gen. easy enough to keep powerpc and amd. amd has been pushing the power envelope downward for the last several revisions. and there's nothing about powerpc that makes it consume 100 watts if you're not on the absolute bleeding edge.
So, physics just magically happens in some silicon ether, not in the CPU. Script-powered world interactions don't use the CPU? More objects, especially NPCs, don't need more CPU? You seemed to ignore that part. The CPU isn't just a front-end for the GPU. It runs all of the game that isn't in the GPU, and then also acts as a front-end for the GPU. Current consoles are stretched about as far as they can go on their old wimpy CPUs, and need newer and better wimpy CPUs.No, today, it really doesn't. The job of a cpu in modern (dx10+) rendering is mostly to send datasets to the GPU. So what you need is really good IO. Which you would have with beefy unified memory bus.
That would be easy for IBM to work around; and I'm going to take a wait and see approach to the ARM. At this point, we don't even know what the A9 would be like, if given freedom from cell phone and tablet constraints. Few good test platforms are out there, and nobody who has had one has taken the time to (or, if they are doing commercial development with it, been allowed to) try to get some gauge of real-world performance. RISC CPUs can always show high int throughput, so I'll keep my skeptic hat on.But we do know something about the A15. Specifically, it's 3-issue OoO instead of 2-issue, and arm is marketing it with 40% clock-for-clock speedup over A9. Also, it has advertised speeds up to 2.5GHz. It would shred Xenon, especially in latency-sensitive integer loads.