Soldato
- Joined
- 2 Jan 2012
- Posts
- 12,407
- Location
- UK.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD Might Be Developing Next Generation APU With x86 Cheetah and ARM Cores – Features GCN 2.0 Cores
- Thread starter Boomstick777
- Start date
I can see this tech being awwsome in tablets and other smaller formfactors, perhaps even slim netbooks, as you can offer the user the already available Android apps paired up with a windows experience or perhaps a more powerful linux experience. Perhaps through dual boot or instant switch as we have already seen. But I have no idea how the ARM part will help the desktop marked. Maybe someone with a bit of knowledge regarding ARM could enlighten me ? 
But I have no idea how the ARM part will help the desktop marked. Maybe someone with a bit of knowledge regarding ARM could enlighten me ?
A good question, something i would like to know.
ARM CPU's can't run x86, probably for no reason other than; Intel wont let them.
But they can run x64
So what we have here is ARM Cores running x64 (which is AMD's own IP) and AMD Cores running x86 (Intel's IP licensed to AMD in exchange for x64 licensing)
AMD Cores run both x86 and x64. (x86_64)
Why and where do the ARM cores need to be a part of the architecture?
Last edited:
Soldato
- Joined
- 5 Dec 2010
- Posts
- 3,212
- Location
- Solihull
I'm no expert of CPU architecture, but this is my take:
but if you want something that purely routes the workload between cpu cores and the igpu. It'd make sense to use something lightweight and low powered like an arm core.
They're used in some ssds and will be part of high end Maxwell gpus, my guess would be for a similar reason.
but if you want something that purely routes the workload between cpu cores and the igpu. It'd make sense to use something lightweight and low powered like an arm core.
They're used in some ssds and will be part of high end Maxwell gpus, my guess would be for a similar reason.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
No, and no.
x64 is an extension of x86, you can't have x64, it's x86_64. Intel licence x86 to AMD and AMD license intel the x64 part. They will continue licensing them to each other because they currently have no other choice. In the future it's actually interesting as there is potential that if AMD is a big player in ARM, particularly in servers but in desktop. If Windows has moved to support ARM better(it's in the process of happening already) and Android likely moves up the range of laptops to higher and higher performance ones, then AMD could dump x86 and not renew the license with Intel, which would screw Intel hard.
Can ARM use x86, no, Arm the company don't have a license to work on x86 chips and they wouldn't want to. AMD can mix an ARM core with x86 isa as they own a license for both and can do whatever they want, will they... answer is there is really no reason to mix standard ISA's, ever.
ARM isa(well the latest one) is significantly newer and has a lot more registers than x86-64. It's very possible for it to be faster, and many, if not most of the main people in the industry believing so.
AS for using an ARM core for a low power core logic, no. What people refer to as cores and core logic, cores are a very small part of an overall CPU chip, core logic routes date to cpu cores to be processed, you wouldn't use a ARM cpu to route data.
Nvidia's Denver is by all accounts a chip that rather than the usual ARM ISA based instructions into the decoder and through the integer/fpu cores for processing, you have ARM instructions pass through a translation layer then into an effectively completely alien integer/fpu pipeline. It's possible to get huge performance like that, or create massive problems, more complex and more ripe for problems.
AMD could make their own core instruction set and make an ARM and a X86 translation layer then make incredibly similar chips except for the translation layer, but I don't think this is what they are doing at all.
The general plan appears to be for next year having a low-ish power pin compatible ARM and x86 versions of chips for the same platform. This is almost certainly a say 2-15W arm/4-25W x86-64 situation. A further improved Beema/Mullins at 20nm with significantly better performance again.
Then the year after we have K12, which sounds like it will eventually be AMD or ARM and likely aimed at say 15-80W range. The general gist of the situation is that next year the lower power chips are completely different except for the pin compatibility.
The K12 architecture sounds like everything on the chip, memory controller, fixed function, anything except the actual cores themselves will be the same and AMD will have this one chip with either arm or x86 cores in them. Effectively everything but the core being identical is something very different and very new.
Unfortunately this false.
ARM CPUs use their own instruction set which is completely and utterly different to x86 and x86_64.
ARM CPUs can run both 32 bit software and 64 bit software but that is entirely different to the 32 bit or 64 bit instruction set that Intel uses.
Do not get confused. ARM CPUs can not run software compiled for Intel compatible CPUs.
They're fundamentally different, ARM is RISC (reduced instruction set) X86 is CISC (complex), reflecting the typical application of each.
http://en.wikipedia.org/wiki/ARM_architecture
http://en.wikipedia.org/wiki/ARM_architecture
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,505
- Location
- Dormanstown.
Indeed it did, and it needs to die unless Microsoft plan to invest mega bucks into it.
