ARM gets weapon in server battle vs. Intel

10/2/2012 3:03 PM EDT

When I visited the Elbrus Laboratories in the early 1990s I was shown big iron Russian supercompters made out of 1970s and 80s vintage technologies. I was told that we built these out of necessity from vacuum tubes, and "they work". Russian/Soviet science and technology has come a long way, and has ways to go. It's gratifying that Elbrus is tackling ARM software and is presenting at the SV venue.

10/3/2012 11:23 AM EDT

I think the need for x86 compatibility is overblown and rapidly diminishing. Only legacy closed source applications have a need for such binary compatibility. Cloud infrastructure runs largely on open source technologies such as Linux and couldn't care less whether the instruction set is x86 or ARM. Moving an application from x86 to ARM is usually only a "make" away.

10/3/2012 11:27 AM EDT

"Make" is not the issue... it's all the testing, qualification, etc of the stack

10/3/2012 11:55 AM EDT

Why haven't they been bought by Intel yet?

10/3/2012 12:02 PM EDT

I just don't understand how an x86 emulation on ARM can help. ARM lags in performance, but makes it up in its market segment by superior power efficiency. Emulating x86 on such slower chips is bound to be slow-squared. The only reason to do it would be legacy applications, but the whole point of the post-PC era is that it's no longer the Wintel monopoly: backwards compatibility requirement is over.

The new software is portable---either because it's Open Source (like Android) or because it is written in Java or HTML5, or is really a light client running against a cloud-based back end.

Look at Transmeta---they couldn't make the emulation win, even though they emulated on a custom-fit microarchitecture rather than general ISA like ARM. Intel is just too good at improving their product for those that needed it.

The bottom line is, those sorry chaps who cannot move away from x86 will use Intel. Those who can port will make a decision based on cost, performance, power consideration, etc. I'll just say that I just got my $30 Raspberry running yesterday. It's slowish, but it runs off battery and did I mention that the entire computer costs $30?

10/3/2012 12:40 PM EDT

This does reduce the barrier to adoption for ARM in the server space though. I agree that the future is less sticky as far as ISA's go with most of the could applications running on higher abstracted platforms that hide the underlying h/w. But it is a compelling story to be able to say to the datacenter operators that they can run their existing software (albiet at lower performance) and all the new stuff with great power efficiency.

10/3/2012 3:34 PM EDT

x86 emulation has been attempted in the past including on big iron DEC Alpha; always fell short on overhead.

Although can yield an interesting engineering lesson, especially in Nx486 to 586 example on
reverse engineering x86 instruction set for implementation in RISC hard wired decoder.


Mike Bruzzone
Camp Marketing

10/3/2012 10:21 PM EDT

Several very good comments here. As some note, in the area of cloud platforms, the software assets are owned by the data center. So this type of emulaton technology is less relevant. However if you imagine an Enterprise severl where there is a mix of apps written in a high level (and therefore relatviely easy to recompile and run natively) language and an (annoying) application where only the x86 binary is available. Here, this type of translation is useful. And the performance hit will lessen. Trust me.

10/3/2012 10:35 PM EDT

It think this is interesting but not a game changer. Emulation has been around a long time.

For instance, I still have and use a Palm OS PDA. Original Palm devices used Motorola Dragonball processors. As devices required more power, they outgrew the Dragonball, and Palm and competitors shifted to ARM devices. To avoid making the vast amount of existing Palm software incompatible, Palm implemented a hardware compatibility layer. Motorola instructions were intercepted and converted on the fly to ARM instructions. In fact, it wasn't possible to write fully native apps. You wrote and compiled to Motorola code, but you could write "ARMlets" in native code to speed up critical operations.

Granted, the ARM CPUs were a *lot* faster than the Dragonball chips they supplanted (My ARM device runs at 200mhz. The last Dragonball device ran at 33mhz), so the overhead of emulation wasn't a factor.

It will be here, but that may not be critical. It will depend on exactly what X86 apps will run under it, and I would expect the emulation to improve over current levels, so while it might not be as fast as native code, it might be fast *enough*.

And it would be a stopgap in any case: it's a fair assumption that if ARM gains traction in the server space, proprietary X86 apps will be rewritten/recompiled for ARM to sell into that space.

10/3/2012 11:14 PM EDT

I think this is a nice project for some research work but not so much for a startup company. I wonder what they aim to monetize by making the emulator run x86 apps. The world of internet is becoming more and more ISA and OS agnostic where the only winning combination is performance and power.

10/3/2012 11:38 PM EDT

Fergie 65; on performance will lesson under emulation; exactly. If you mean by binary that the performane hit will lesson for applications
recomplied for 'specific x86 instruction calls" exactly. Hardware substitution learning always starts addressing core instruction translation equired for target application.

Mike Bruzzone
Camp Marketing

10/4/2012 3:05 AM EDT

FYI, Here's some analysis that was apparently cut from the current version of the story:

“It’s very possible this could be useful technology in a 2014 time frame,” said Kevin Krewell, senior analyst with market watcher Linley Group (Mountain View, Calif.).

Startup Transitive Technology helped Apple and SGI transition to the x86 from PowerPC and MIPS respectively before Transitive was acquired by IBM. Emulation “could be used for migration, but not as a long term strategy,” Krewell said.

Emulation is typically used when the new architecture has higher performance than the old one, which is not the case—at least today--moving from the x86 to ARM. “By the time this software is out in 2014 you could see chips using ARM’s V8, 64-bit architecture,” Krewell noted.

“That said, you will lose some of the power efficiency of ARM when doing emulation,” Krewell said. “Once you lose 20 or more percent of efficiency, you put ARM on par with an x86,” he added.

10/4/2012 3:06 AM EDT

BTW, ARM is quick to admit there is a wealth of 64-bit server software that still has hooks to the x86.

10/4/2012 10:59 AM EDT

The real question is: Can ARM compete with Intel in the long run? It boils down to process and economics, not instruction set. Whomever delivers the most work per Joule wins.

10/4/2012 11:22 AM EDT

Intel cannot invade Samsung's vertically integrated ARM space, which will only grow, I think.

10/4/2012 6:23 PM EDT

Good luck to ARM, we need a real competitor to Intel.

10/8/2012 5:24 AM EDT

I wonder how much that 40% of native performance is due to the x86 code. The ARM has more registers so does less stack shuffling to get its work done. Translated code has to pay the Intel Tax.

11/7/2012 9:36 PM EST

Does anyone happen to know where I might be able to find this paper that was presented last week at ARM TechCon? I am currently taking a graduate level computer architecture course and I would love to present this article to the class. Any information anyone might be able to offer would be much appreciated.