ARM Researching Novel Chip Memory

An anonymous reader writes "ARM may be best known as processor designer but the company is now working on a non-volatile memory that could scale down to 5nm, according to an Electronics 360 report. The memory is something different called Correlated-electron RAM that was originally developed by a professor at University of Colorado. ARM is joining a research collaboration to try and make the memory an option at ARM-friendly foundries."

I love ARM

[EmperorOfCanada]

I love that ARM didn't initially go head to head with Intel and thus ended up not getting crushed by them (think transmeta/AMD). I thus have hopes that this not only works because it is cool but because ARM is cool and deserves another win for what they have done.

Re:

[sumdumass]

Interesting take on it. ARM was originally loved because it effectively ran circles around x86 style processors at a fraction of the power. It's been a while and I'm not sure if this is still the case but I often root for old time favorites of long ago just out of habit.

Judging from the prevalence of ARM technology in today's hardware, I would think it is still better then the Intel and AMD alternatives.

Re: I love ARM

[Anonymous Coward]

The Intel equivalents are roughly equal in terms of performance but are relatively expensive to buy for manufacturers. The base of Android would have to be recompiled for x86 (relatively easy) but there are some Android games/apps that are made Arm specific for performance gains and wouldn't run without development on other platforms.

Its just a more costly less compatible alternative, if Intel want to make any gains they would need to demonstrate real performance benefits than just performance parity. I did

Re:

[The123king]

The problem with Intel is they've sat on their laurels with the x86 architecture for too long. I give them many points for trying to capture the high-end market with Itanium, but they didn't realise it was going to be the low end market that would come around and steal away the next big market. They were too busy protecting their face to realise they were gonna get kicked in the balls.

Re:

[AmiMoJo]

Android for x86 has a binary translation layer for native ARM code that seems to do a pretty good job. There is no need to even re-compile such apps and performance seems to be near native.

Re:

[Bengie]

Intel's new 250mw CPUs have more mips and flops than ARM with the same power rating, and the Intel CPU has better idle power draw. Intel is starting to beat ARM at their own game and with full x86 CPUs. ARM is trying to break into Intel territory and looks to have "decent" power draw and performance, but many of their current CPUs do not scale well past 1ghz. Like their ARM A9 consumes 4x the power just to gain an extra 10% performance on the high end. These "low power" CPUs are great at low frequencies, bu

Re:

[nojayuk]

ARM is a design operation, it produces no silicon of its own. Intel is an end-to-end company both designing and fabricating its own product with world-leading fabs, die shrinks, finFETs etc. Intel hasn't bothered much with low-power performance up till now as there wasn't a demand for it until relatively recently -- performance was for the data centre, supercomputers and workstations, low power consumption was for embedded devices and braindead phones and handhelds. Laptops were in the middle but a three-ho

Re:

[fast turtle]

Most ARM fans keep repeating the same fucking mantra - It uses less power then x86. Bullshit. It's not just the CPU that results in power demand for mobile systems (Tablets, phones and what not) it's the GPU, Screen, disk and all the other parts that go into the thing and Intel has finally paid enough attention to overall power budget that they're actually beating most of the ARM based SoC's out there in both power savings and performance.

Look at the power demand of the Nvidia Tegra SoC's. Tegra3 was decent

hard disk and gamer GPU on my phone?

[raymorris]

The disk? Did you really just say that? You just said the DISK in a phone draws so much power that the CPU doesn't really matter? Umm, phones don't have hard drives. They don't need powerful GPUs either, and the screen is off 96% of the time.

Intel x86 (actually AMD 64) sometimes makes sense in a desktop, where you do have a couple of hard drives, a powerful GPU, etc. A phone is not a desktop. A phone is a low power device. Finally Intel doesn't use a THOUSAND times as much power as ARM anymore, so it's no

Re:

[viperidaenz]

Intel held on to the ARM licenses they had when building the StrongARM and XScale CPU's a decade ago.

Re:

[Bengie]

I like your summary better :)

Re:

[The123king]

It's still too little too late. Look at Apple when they moved to PowerPC. Apple's computers could run circles around the competing x86 chips, but by then the x86 architecture had become so prevalent there was no hope in hell that PowerPC would dominate. It's the same with between x86 and ARM in the mobile world. Sure, there will be x86 chips that are faster and more power efficient, but everyone's so ingrained into ARM that they'll never really make inroads into the market.

Re:

[fdrebin]

... but everyone's so ingrained into ARM that they'll never really make inroads into the market.

"Never Say Never Again"
Or... look up "You can't be fired for buying IBM"
Or... "DEC Forever" or (one that a was told to me personally in 1989) "PCs will never have the power of Workstations!"

Re:I love ARM

[0123456]

I've heard this before, but Intel likes to play games with power numbers. AMD's TDP is the limit of the processor if EVERYTHING were on. Intel's is under normal usage.

Yeah, right.

That's why, when you actually measure the power consumption, you usually find Intel CPUs are far more efficient than AMD's.

Re:

[viperidaenz]

Maybe that's because AMD don't have to technology to turn off all the unused parts of the chip or use all the execution units all at once so everything is all on all the time.

Re:I love ARM

[AmiMoJo]

The problem is that no matter how good Intel chips are only Intel makes them. ARM is widely licensed and a manufacturer has a vast selection of processors and most important system-on-chip silicone to choose from, at all performance and price levels. There is also the language issue, with far eastern manufacturers preferring support and documentation in their native languages and from local companies.

More over you generalize too much. Saying "ARM A9 consumes 4x the power" is meaningless because there is no "ARM A9" chip, only various implementations of the A9 spec and they are all different.

Re:

[EmperorOfCanada]

I think have hit on a hugely important factor. By having many different manufacturers there is no worry about ARM playing some games like cutting you off, or strongarming you into some new marketing ploy. If one manufacturer tries to screw you there are many others happy to do business.

On a side note the server I have long been waiting for is a 16+ CPU arm server. A typical web site is continuously getting a zillion microscopic requests that shouldn't trouble a low end ARM chip. So why not spread them ou

Re:

[TheRaven64]

By having many different manufacturers there is no worry about ARM playing some games like cutting you off, or strongarming you into some new marketing ploy. If one manufacturer tries to screw you there are many others happy to do business.

This is part of ARM's strategy with ARMv8. They intentionally delayed their own designs so that they wouldn't compete with their partners. Now, if you want to license a 64-bit core from ARM, they have a low-power in-order design and a better-performing out-of-order superscalar design, but several of their partners also have their own ARMv8 implementations that were built with advice from ARM engineers but are independent implementations. They will each have different power/price/performance trades, helpi

Re:

[Ginger Unicorn]

Hey! I love ARM because Acorns were full of ARM. Apple indeed...

Re:

[myurr]

It was the first processor I learned to write assembly for, back on the Acorn Archimedes. Brilliant computers, so far ahead of their time, and I wouldn't be half the programmer I am now were it not for learning on those machines.

Re:

[wonkey_monkey]

MOV PC,R14

Ahhh... that's the stuff.

Re:

[dpilot]

Oh wow! Shades of: //STEP01 EXEC PGM=IEFBR14

Also proof that a program that does absolutely nothing can indeed have bugs. Look up the history some time. (It's on Wikipedia.)

Re:

[TheRaven64]

Great for assembly programmers, not so great for pipeline designers. One of the weaknesses of the ARM design is that making the PC a general-purpose register meant that every instruction needs a little bit of extra logic in the decoder to tell you whether it's a branch, which then complicates the branch predictor. It's fun being able to do a ldr with the pc as the destination register, right up until you have to implement a long pipeline with that instruction. This is why ARMv8 makes the pc a special reg

Re:

[The123king]

I love ARM because it powers everything from my iPad to my HTC One to my WD HDD to my TV to my Raspberry Pi to my CD ROM drive to my router to my... *mumble mumble drone drone*

Re: I love ARM

[astar]

I love ARM. Validating the first Cpu designs in your head when IBM teams were in Fail mode inventing pipelining. Noticing that their chip was unexpectedly working even with a fail power supply. Low royalities and not big revenue even though they have designed most of the computers on the planet.

Re:

[flyingfsck]

ARM started as Acorn in 1978, the same time that Intel created the 8086 processor. The current popular ARM processors are actually MIPS processors which likewise goes back to the early 1980s. So this stuff is oooooold.

Re: I love ARM

[Anonymous Coward]

Same anon: I'm an idiot, please disregard that post.

Re:

[fast turtle]

At least you realize it. Maybe now you can do something about it.

Old Proverb "Wise Man keeps mouth closed instead of opening it and proving they're an idiot".

In my case, the only excersise I get now if from "flying off the handle, then jumping to conclusions, followed by butting in" so I wonder why everyone dislikes me?

Re:

[edmudama]

It's Proverbs 17:28

Re:

[Megol]

ARM started as Acorn in 1978, the same time that Intel created the 8086 processor. The current popular ARM processors are actually MIPS processors which likewise goes back to the early 1980s. So this stuff is oooooold.

MIPS?!? Did you just make that up? Do you think x86 are MIPS too?!?

ARM and MIPS are processors and there the similarities end.

Re:

[ButchDeLoria]

Also both RISC.

Re:

[Megol]

The original ARM ISA is pretty complex for a RISC design and isn't as strongly pipeline optimized as many RISC (even ignoring the lack of branch delay slots). Few registers, multi-clock instructions (load/store multiple), addressing modes with register updates, not being heavily optimized for pipelining and no explicit zero register are just a few more or less unique features of the original ARM compared to other RISC processors. Making almost every instruction support conditional execution is pretty much u

Re:

[Megol]

So the idea is that having a zero register makes the processor a MIPS one? There have been zero valued registers/accumulators/storage locations in many processors long before RISC was "invented" which isn't that strange as a zero value is a very common immediate data value.

Re:

[TheRaven64]

That's an idea that predates MIPS. You'd do better if you claimed MIPS heritage because AArch64 removes the PC as a GPR, but (unlike MIPS) it does provide PC-relative addressing (MIPS does PC-relative memory accesses by forcing the calling convention to make every function call in position-independent code a JALR $25, so that $25 is guaranteed to contain the PC on function entry). MIPS also lacks the complex addressing modes of AArch64 (and AArch32). Unlike AArch64, it has branch delay slots, doesn't hav

Re:

[dreamchaser]

ARM processors are not MIPS architecture. The latter still exists and has little in common with ARM other than both being RISC.

Re:I love ARM

[itsdapead]

I love that ARM didn't initially go head to head with Intel and thus ended up not getting crushed by them (think transmeta/AMD).

Actually, they did start out (as Acorn) by going head-to-head with Intel. Others have mentioned Acorn but not really pointed out that the original 1987 ARM was a credible competitor to the 80286 and 68000 [wikipedia.org]. (By "credible competitor" I mean "left the 68k and 286 choking on its dust"). It was only ever really used in that way in the Acorn Archemedes [wikipedia.org] and RiscPC [wikipedia.org] which never made it big outside of the UK - although it outlived most of the other non-Wintel personal computers.

OK - when ARM was spun off they did, as you say, rather sensibly, end up going after the embedded market, but ARM might never have happened if Acorn had gone with the 80286 [wikipedia.org] for their BBC Micro successor.

Re:

[Darinbob]

They did go head-to-head, just not on the PC.

SRAM

[Anonymous Coward]

efficient SRAM would be a bigger deal. DRAM is holding us back right now.

Novel chip memory

[rossdee]

Soon to be seen in Kindles and Nooks

Re:

[Bengie]

The lone inventor is mostly something for a new area of research. Once an area of research is mature, it's near impossible to compete with larger commercial entities.

Re:

[viperidaenz]

Farnsworth invented the smell-o-scope!

Re:

[RaceProUK]

Don't forget the Finglonger :)

Re:

[viperidaenz]

He didn't actually invent the Finglonger.
He used his What-If machine to find out what would happen if he did invent it.

Re:

[RaceProUK]

And in a later episode, he actually had one.

Re:

[TheRaven64]

Individual researchers are less likely to make significant contributions, but small teams still do. Individuals have ideas. Small teams produce proof-of-concept implementations. Larger teams produce working prototypes. Big teams produce finished products. It's hard to compete with a large company when it comes to bringing products to market, but it's quite easy for university research labs to get things to the proof-of-concept stage. They then have the choice of spinning out a company to try to commer

Re:

[account_deleted]

Comment removed based on user account deletion

bahahahh

[Anonymous Coward]

gotta love you ARM fanboys, as misguided as you are.

ARM's days are numbered. It can try to come up with whatever hacks it wants, but in the end, they can't beat physics. And if anyone understands that, it's Intel. MIPS won't change that. So prepare to watch ARM flail around while it loses significant market share to Intel over the next year.

As for Nvidia... they love to overhype and underdeliver in hopes people will just settle for what they're offered. Nvidia will partake in the same woes as ARM over the n

Re:

[crutchy]

yeah and linux is just a fad!

microsoft ftw! :-)

The only winner in this game is Intel.

might want to google "china"

Re:

[Anonymous Coward]

Wow, I had a good laugh at your misguided fanboism.

Even if ARM stopped making any chip technology to be used in computers, tablets and smartphones they would still be the leader in the chips in all those other devices you rely on in your daily life.

Pull your head out and you'll see that ARM is everywhere. Are they the best? Not always, but they are often the best for what they are used for.

Re:

[dreamchaser]

That's like saying Windows is the best for what it is used for because it has a large market share. ARM is prevalent in mobile devices because when smartphones and tablets started to rise in popularity Intel at the time had no low power option to use as their answer to ARM. I'm not so sure ARMs 'days are numbered,' I'm just pointing out that numbers does not necessarily mean ARM is the best at what they are used for anymore. Intel has closed a great deal of ground in that regard.

Re:

[Anonymous Coward]

ARM is the most sold processor architecture on the planet.
Every modern car has like 20 to 100 ARM cores build in, and - if at all - (for the radio) a single Intel procesor.
iPhones, iPads and plenty of other mobile devices run on: ARM.
Claiming that Intel makes the long run is just nonsense. And if Intel is not sooner or later abandoning the stupid x86 architecture altogether they will go where Microsoft is going or where MIPS already went.

Re:

[viperidaenz]

MIPS were pretty popular in ECU's, so were 6502 and 68k. With the speed at which the auto industry moves, they probably still are.

Re:

[crutchy]

i disagree about arm's days being numbered and all that, but in all fairness a lot of embedded applications are likely also powered by traditional microcontrollers supplied by freescale (motorola), atmel, st micro, ti, pic, etc.

Re:

[Pinky's Brain]

Intel has good technology ... but everyone is going to be a loser in the coming years and Intel is becoming a tempting target for a leveraged buyout, which might well destroy them as an IDM.

Re:

[viperidaenz]

MIPS [wikipedia.org] and ARM [wikipedia.org] are different architectures.

5nm? Die Shrink?

[v4vijayakumar]

a non-volatile memory that could scale down to 5nm?

Why 5nm is significant? Is it something to do with die shrink? http://en.wikipedia.org/wiki/D... [wikipedia.org]

Re:

[Bengie]

Probably near the distance where quantum tunneling becomes an issue. When your design requires a layer of insulation and electrons can pass through it, then your design breaks.

Might be a waste for bulk CMOS

[Peter Gasperini]

The fact that ARM is doing such theoretical research and trying to turn it into practical applications is excellent. However, they might want to consider investigating the use of this memory in non-CMOS applications, particularly for sensors. Nobody is going to make a 5nm bulk CMOS node - we've hit the Last Node at 14nm.

Re:

[viperidaenz]

10nm is coming next year... 7nm in 2017. 5nm may be a few years after that.

Re:

[Anonymous Coward]

Except that 16/14nm isn't much more logic dense than 22/20nm. Now we keep making the minimum feature size smaller, but the gate length is about the same size (e.g, FinFet). Of course types of circuits scale better than others (e.g. rams), but one of the reasons to not scale down is that power wall (it's currently better to have larger devices to minimize static current leakage than have minimum sized devices and melt the silicon as soon as you turn it on).

At 10nm, quantum tunnelling is a significant imped