ARM leaps forward!

[takemehomegrandma]

This post is probably most interesting to AROS currently (and to everyone else interested in general computer technology evolution of course ):

Just days after the Exynos 5 Dual CPU made its debut in Samsung's new Chromebook (we will most certainly see it in a future Galaxy IV phone as well), the worlds first Cortex-A15 ARM CPU showing truly impressive performance, ARM officially launches the "Next Generation" 64-bit ARMv8 architecture (backards compatible to ARMv7):

"ARM Launches Cortex-A50 Series, the World’s Most Energy-Efficient 64-bit Processors"
http://www.arm.com/about/newsroom/arm-launches-cortex-a50-series-the-worlds-most-energy-efficient-64-bit-processors.php

These are the first two ARMv8 cores available for licensing from ARM themselves. Here is a summary:

Cortex-A53: ~1x "today’s superphone" performance using 1/4 of the power
"The most efficient ARM application processor ever"

Cortex-A57: ~3x "today’s superphone" performance using 1x of the power
"Provides computer performance comparable to a legacy PC,  while operating in a mobile power budget"

And like before with big.LITTLE, these cores can be combined in various ways in a single CPU chip: "The Cortex-A53 processor combined with the Cortex-A57 and big.LITTLE processing technology will enable platforms with extreme performance range while radically reducing the energy consumption"

Among those CPU manufacturers that is going to make their own products based on these cores you see Broadcom, Calxeda, HiSilicon, Samsung and STMicroelectronics, and a bit surprisingly AMD:

"AMD has signed a license for a 64-bit processor design from ARM, ending its exclusive commitment to x86"
http://www.techworld.com.au/article/440450/amd_sell_arm-based_server_chips_2014/

Other big players making CPU's based on this new 64-bit ARMv8 architecture you'll find Applied Micro, Cavium and of course nVidia.

nVidia kind of shook the ground in the industry when they announced their future ARM strategy together with Microsoft almost two years ago.

This one is IMHO the most interesting one, the one I personally am most curious about. These CPU's ("Denver" core) will be "designed to support future products ranging from personal computers and servers to workstations and supercomputers", "we are designing a high-performing ARM CPU core in combination with our massively parallel GPU cores to create a new class of processor"

According to this blog they are looking to go head-to-head with x86: "Denver frees PCs, workstations and servers from the hegemony and inefficiency of the x86 architecture.  For several years, makers of high-end computing platforms have had no choice about instruction-set architecture.  The only option was the x86 instruction set with variable-length instructions, a small register set, and other features that interfered with modern compiler optimizations, required a larger area for instruction decoding, and substantially reduced energy efficiency.

Denver provides a choice.   System builders can now choose a high-performance processor based on a RISC instruction set with modern features such as fixed-width instructions, predication, and a large general register file.   These features enable advanced compiler techniques and simplify implementation, ultimately leading to higher performance and a more energy-efficient processor."

And I think this is a lot bigger deal for nVidia than most people think, they are making something completely different out of it than "just" putting out another "CPU core with GPU" to the market. In this very interesting blog/interview (read it, really, do it!), nVidia chief Jen-Hsun Huang describes it as an upcoming paradigm shift, and a "re-invention" of the whole company:

Nvidia 1.0 was PC graphics (made possible by "fab-less production").
Nvidia 2.0 was the creation of the "GPU"
Nvidia 3.0 (about to happen) is about parallel processing (in a "newish" way, as I read it)

Nvidia is for the first time designing both the CPU core(s) and GPU on their own, in-house, towards goals and a purpose they have defined themselves. That interview is 1.5 years old, and in it, Mr. Huang mentions they (a few hundred engineers) have been working internally with this for 3.5 years already. I have seen a post somewhere (can't remember) suggesting that these new chips will integrate the "GPU" and "CPU" on the silicon in a previously never seen manner, the on-chip internal data bandwidth between those parts will be enormous, it will be so considerable that those previously "separated parts" will kind of  merge in practice. Again, nVidia themselves are labeling what they are now doing with ARM/Denver as being equally significant as the fab-less chip production and the rise of the GPU concept, it will be the next step, and will constitute "Nvidia 3.0".

That's why I'm so curious to see it. It's definitely going to be more than "just another CPU-core/GPU SoC bundle, only faster". Making a faster "Tegra" won't exactly warrant a "Nvidia 3.0" label, there has got to be more than that to it.

"Every single one of this project are fully funded and the expectation is within the next three or four years [1.5 years ago] we’re going to bring to the mobile market performance that is nearly a hundred times higher than today’s’ PC. And that’s the roll out if you will of our Nvidia 3.0 strategy."

"ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers."

1.5 years ago he said that it takes about 5 years to design a custom CPU, and at that time they had already been working 3.5 years on it. There are many speculations about release dates, some say 2013, and that seems kind of probable.

ARM Naysayers - REPENT!

[takemehomegrandma]

OH, bulls...

According to the nVidia president in that interview, these are the things that matters to them, and the reason to their decisions (made a couple of years ago):

• Energy efficient computing (of growing importance on all markets, not just mobile)
• Mobile computing (one of the sides of "energy efficient computing")
  
• A common platform, ranging from tiny handhelds all the way up to supercomputer
• Unique control over the IP of that platform (the whole "eco-system" actually)
• Not making some x86 commodity that won't ever be able to compete with Intel

ARM brings this, x86 doesn't.

"Now inside the company we say the way we distilled Nvidia 3.0 down into actions is three arms. We say go parallel, go mobile and go ARM. Now a lot of you have asked me over the years what is our CPU (central processing unit, or computer’s brain) strategy and I’ve said over the years it was ARM. And I said it so matter of factly and it almost sounded like a joke. But it was the truth. It was the same thing I was telling our company inside that our CPU strategy is ARM. I believe that ARM will do for CPUs what Taiwan Semiconductor Manufacturing Co. did for foundries (the contract chip manufacturers)."

"Instead of hundreds of millions of devices, it’s several billion. And so that would make the ARM processor the most valuable instruction set architecture (or chip processing architecture) in the world."

"our strategy with Project Denver was to extend the reach of ARM beyond the mobile, the handheld computing space. To take the ARM processor, partner with them to develop a next-generation 64 bit processor to extend it so that all of computing can have the benefits of that instruction set architecture. It is backward-compatible with today’s ARM processors."

"And so everyone now sees the picture that our CPU strategy really is ARM, that we intend to take the ARM for mobile devices all the way to supercomputers. ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers."

"People still thought a cloud over our heads was our big battle with Intel. People said that Nvidia’s Intel chip set (MCP) business is going away and of course we announced that our dispute with Intel has been resolved. We’ve extended our cross license with Intel and the licensing revenues that would come to our company would be approximately $1.5 billion over six years. That by and large replaces and some the business that we lost with MCP."

"There are two reasons why we decided not to do x86. Aside from, well the second reason is what I said earlier that in fact it’s the wrong instruction set architecture. The first reason is simply very large of course. The world’s not waiting for us to build yet another x86 and we’re not going to go hire a bunch of the world’s best engineers so that we can wake up in the morning to go do something that somebody else has already done 25 years ago. It’s not logical. And so it’s another way of saying it’s a commodity. Intel has got every single price point covered from $10 all the way up to $1000. There is not one nook and cranny we can cover by ourselves. AMD has covered everything else."

"At AMD, they actually make perfectly good CPUs. I’ve never met a CPU at AMD that I didn’t like. They are all fine. They just can’t win."

"even if they give me rights to make an x86 chip, I will be building a commodity that at every price point they have an alternative to. And if they have an alternative to everything that I make, and it’s easier to buy from Intel, it’s just really not possible to distinguish yourself in an x86 world. And so that’s sort of the reason why, that’s one of the negative reasons why you don’t do it. But the positive reason is we all want to go make a contribution to something and make a difference in the world. I mean you’re going to go spend $1 billion in r&d, you go spend $1 billion building something that matters."

Everything makes perfect sense in my eyes.

[takemehomegrandma]

To meet x86 head on you need to be able to beat them in all of it's strong areas, not just defend your own.
 
They can complain about x86 being the wrong instruction set, but to prove it they need to launch an ARM core that can beat an i7 and they can't.
 
As soon as ARM is faster as well as cheaper and less power hungry than x86, then everyone will switch everything to it.

I think we will see something novel regarding parallel processing, like a combination of a "traditional" CPU (probably a few of them on chip) with more than enough horse power for most traditional applications, connected to a massively paralleled, high speed GPU that will join in on "general computing". I think the key here would be the removal of any bottle necks in between them, to get *massive*, super fast, direct, on-chip bandwidth in between them. That would bring a whole new meaning to CUDA, and this together would bring *a whole new class* of processors.

Could be wrong of course! But there are interesting things ahead...

[takemehomegrandma]

That would explain the flood of new PPC designs including two new 64bit cores (a first for Freescale). Yep, look like they're abandoning the market.:laugh1:

If they will have to cut somewhere (which they probably will), then it won't be their ARM section. They are even replacing some PPC based applications with ARM themselves, and this is something happening in many embedded industries which has been traditional PPC strongholds, like automotive. AFAIK.

Not that PPC matters anymore to anyone interested in desktops/laptops, but anyway...

[takemehomegrandma]

It's only a matter of time before Microsoft abandons PPC in it's XBox.

As I've said before, I wouldn't be totally surprised if the new X-box will be ARM based, powered by Denver/nVidia. Watch it happen!

We'll see in time I guess...

[takemehomegrandma]

Once again there are rumors of a potential, upcoming architectural switch for Apple Mac's:

http://www.bloomberg.com/news/2012-11-05/apple-said-to-be-exploring-switch-from-intel-chips-for-the-mac.html

This is far from being the first rumor of this kind, many similar has circulated over the net during the last few years. One almost begin to wonder if there is a certain threshold when one can actually start using the old "where there is smoke, there is fire" saying.

The usual responses to these rumors are: "Bullsh!t, ARM doesn't have enough performance", and the people saying so are usually sitting in the car of present, driving down the road of time, and looking at the various ARM CPU's in their rear mirror. A natural thing to do (looking in the rear mirror that is), since the road goes over a hill a bit further down the road, a hill blocking the view of whats at the other side of the crest. So since you don't have a picture of the future, but you do of the past, your comments couldn't possibly be anything other than the one above.

But just because you and me and all the other common people can't see at all what's on the other side of the hill, it doesn't matter that there aren't people there doing stuff already, things we are about to see in a year or two when we actually drives over that hill and gets our first view over the previously hidden horizon.

If I may speculate, I think Mr. Jen-Hsun Huang, founder and chief of the nVidia corporation, will be standing there with a few new CPU's based on their "Denver" concept. "within the next three or four years we’re going to bring to the mobile market performance that is nearly a hundred times higher than today’s’ PC." (That was 1.5 years ago) "ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers."

They have been working with this for years, and should almost be ready. He is the guy who brought us a true computer graphic evolution (through competition and fabless production), he is the guy who brought us the GPU, the concept that took graphics and gaming to a whole new dimension by its enormous performance in running massively paralleled code. He has delivered. He has reformed and evolved a whole industry. Twice. Now he is very confident it will happen again with a whole new class of processor. And while you and me can't really see this yet (we will come over that hill in a year or two), I have no doubt whatsoever that both Apple and Microsoft is sitting down at nVidias briefings and presentations (or by now, maybe even demonstrations) of the thing. And within context like this, the "migration" rumors like the one mentioned above doesn't sound like "bullsh!t" anymore.

What did the nVidia chief have to say about a possible Mac OS migration to ARM?

Q:  Is ARM on the Mac OS possible?
A: "I don’t know their plans but if you look at it from 10,000 feet, it seems to make sense, right? Because if they go Mac on ARM, they could address some of their concerns with their own SOC. So instead of paying $150, they can pay $15."

Nothing in his answer about whether it would be technically doable at all, if it would make technologically sense, that seems to be *a given* in his reply (and he has unique inside knowledge of the products ahead, nobody can deny that). No instead he focus on IP and the economic side of doing their own SoC. Which Apple seems to be quite aware about themselves, releasing their own SoC CPU's every half a year or so.

So it will be indeed very interesting to see what's behind that crest of the hill that the road of time will pass over in a year or so.

[takemehomegrandma]

If it leads to other companies offering all this swanky new ARM tech in useful form factors (laptops and desktops) I'm all for it. If it just means they're looking at merging their tablet and desktop/laptop lines, meh. One Windows 8 was way the hell more than enough.

While I think there will be desktops and (particularly) laptops for some time still in the future, there is a shift going on where both the importance and impact of "traditional" computers like these declines. It has been going on for quite some time now (all statistics proves this, but you don't have to read boring statistics, it's enough to observe your surroundings for one day, and you'll see what I mean), and it will accelerate in a close future, that's a safe bet. And this no matter the underlying architecture. The user pattern changes, as simple as that.