Feeds

back to article FIGHT! Intel disputes ARM's claims of Android superiority

Last month, The Reg published test results performed by ARM that the UK chip designers said show that mobile processors based on its technology have significant advantages over Intel's chips when running Android apps. Intel would beg to differ. "Someone's gotta be the truth squad around here, right?" Intel corporate …

COMMENTS

This topic is closed for new posts.
Anonymous Coward

not quite as creative as ARM, but...

having collected a bunch of downvotes on the ARM presentation article by pointing out the "creative accounting" in the ARM guy's comparison, it's a good time to point out the notable non-comparison in the Intel presentation - they are just comparing translated vs. native on x86, and in particular do not compare x86 vs ARM performance at a given price or power-consumption level, or compare x86 vs ARM power consumption for a given application.

Vendor benchmarks are very much like used car ads, it's what they don't say that hints as the facts they'd rather not have known...

16
2
Anonymous Coward

Re: not quite as creative as ARM, but...

Exactly. This seemed to me nothing more then some slimy salesman effort at fiddling some numbers. Remarkably "one careful elderly lady owner. Barely left the garage"

"Someone's gotta be the truth squad around here, right?" Intel corporate communications manager Bill Calder told The Reg.

Erm, rly Intel? You must think we're all morons. The industry and community can't determine any of this for ourselves? We need some self-imposed corporate overlord to decree from on high? Well, if you insist. Won't be you then. Your Netburst and (T)itanic bollox still makes me chuckle...

12
4
Anonymous Coward

Re: not quite as creative as ARM, but...

Oh come on. ARM's own presentation was about the impact of binary translation and ALSO only compared Intel (native) to Intel (translated).

YOU may want to see a comparison of Intel vs. ARM but that's not what's being discussed here by either Intel *or* ARM. Do try to follow what's going on.

3
0
Thumb Up

Re: ...in particular do not compare x86 vs ARM ...

THANK you, AC. I was going to post the exact same comment. Hard to benchmark like-for-like across two different machines but this is surely the meat-and-potatoes of what ARM are banging on about - native ARM vs translated ARM.

0
0
Silver badge

Re: that fps test is worthless

..or nearly worthless. Measuring frame locked rates tells you the game runs fast enough (or not) and nothing more about actual performance. No-one serious about benchmarking ever does it because it hides how much time the system is idling waiting on the next frame time.

The first few entries tell you the translated version was struggling to hit 16ms/frame but don't tell you how much faster the native version was. The 60 v 60 runs could have been an unlocked 60fps v 120fps but we have no way of telling.

I wonder why they chose such a poor comparison? (No, I know why they did it;)

And totally missing is any comparison of native ARM v native x86.

0
1
Silver badge

Re: that fps test is worthless

Yes, but if the game runs fast enough, WHO THE HELL CARES?

Too many benchmarks miss the entire question that's interesting: "Does it do what I want?" For a moblie device that means "does it hit the locked framerate, and which one has the longest battery life while doing that."

0
1
Gold badge

Re: that fps test is worthless

I care. If the device is capable of emitting 300fps and is capped at 60fps then it's idle for goodly chunks of time. That means those transistors can be powered down between frames, or spend their time doing something else. It probably means that system gets way better battery life than some system that is struggling with all it's might to hit the 60fps barrier.

3
0
Silver badge
Paris Hilton

All together now,

"Well they would say that, wouldn't they?"

14
1
h3
Bronze badge

I hope intel make a mobile SoC using their best technology soon.

1
1
Silver badge
Facepalm

Re: Mobile Soc

They did already.

"I hope intel make a mobile ARM SoC using their best technology soon."

Fixed it for you.

Intel have made ARM.

They didn't sell all ARM to Marvell

They had a great RISC of their own design (i960)

They have an ARM Licence.

Their biggest assets are not x86 and Itanium etc but Process Technology, Marketing and Industry relationships. So it's pure stupid ego that they don't make an ARM Soc. It would add some genuine competition and we would then get better stuff from AMD, Qualcomm, Samsung etc.

x86 due to marketing clout and "PC Compatibility" killed off "better" CPUs such as 680xx series, PowerPC, MIPS, Alpha and Alpha64, Sparc etc.

Now we don't so much need legacy x86 Compatibily as Win 8.x isn't that compatible compared with XP, which unlike previous MS OS which went past "Sell by" and "Use By" has been publicly "hung drawn and quartered".

So a good time for Intel to put their might behind ARM instead of polishing the T**d

2
1

This post has been deleted by a moderator

Silver badge

Intel are good

They have managed to achieve amazing things even dragging that dead-weight x86 architecture behind them. They have only kept competitive by being streets ahead in process.

Imagine what they could do if they let go of x86 and put their talents into making the best ARM out there.

Their XScaale ARM parts were amazing. With Intels current process capabilities they could do something that with blow the competition away.

3
1
Silver badge

Re: Intel are good

But what happens if Intel start producing great ARM kit?

ARM gets better - massively successful. More software is written for ARM. 4/8 Core devices increase. Tablets sprout keyboards Productivity apps for ARM become usable. More 48/96 core servers appear. More server-side apps appear. HP starts shipping them, Lenovo starts shipping them, Samsung starts shipping them.

All the architecture improvements are available to all vendors, fab superiority is all Intel has left and that won't command the same premiums and profit that x86 architecture and fab superiority (for some value of "superiority") provide in the desktop and server markets.

Like MS with RT, Intel would like a slice of the mobile market but the definitely don't want mobile to encroach on desktop or mobile tech to find its way into servers. Better (desktop) performance at ARM mobile prices is not what Intel want, which is reason enough to shun their attempts at mobile.

Its Apple and Samsung which drive ARM because neither of these have a stake in desktop or server markets but these two have ARM profit-centres. Yes, both have x86 laptops, but both are processor agnostic because they buy their (expensive) x86 chips.

5
1
Silver badge

Re: Intel are good

Imagine what they could do if they let go of x86 and put their talents into making the best ARM out there.

The chips would be fantastic but their margins would suffer. They need the x86 lock-in to preserve those margins and it's what the salesforce knows how to sell.

Actually, since Intel has already started contract manufacturing this might happen sooner than anyone expects.

3
1
Anonymous Coward

Re: Intel are good

"They need the x86 lock-in to preserve those margins "

Correct!

And how do they continue to get those margins across the range after there is an affordable cost-competitive x86 SoC for phones that will ALSO do the necessary for 80+% of Windows desktops.

0
1
Bronze badge
Meh

I am so sick of Intel's whining about ARM versus Atom or CISC whatever

- ARM is RISC

- Intel is CISC

Enough said.

Meanwhile:

"…If apps haven't been recompiled to run natively on Intel-based Android devices and therefore have to be converted at runtime from native ARM code into native Intel x86 code using "binary translation" (aka "bridge technology"), the conversion caused "a huge impact to the user and to the performance of the system."

Then don't do that! Sheesh.

Intel: You've done RISC before. Do it again and deal with it. Then maybe you can compete in the market without the whining.

1
3
Anonymous Coward

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

Obviously Intel doesn't want to pay any licensing fees to ARM if they can avoid it, and they do have a popular and widely-supported instruction set of their own, so while their strategy doesn't seem to be super successful right now it's pretty logical.

Intel might be able to make the best ARM chip ever but if they start doing that, ultimately they will end up just increasing the popularity of ARM, increasing the viability of their competition, and possibly cannibalizing their own efforts at mobile (and even desktop/server) chips.

2
1
Silver badge

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

ARM is RISC - Intel is CISC

That hasn't been true for years. Intel has been more RISC than CISC since the Pentium (IIRC) and certainly since the P4 debacle.

2
0
Silver badge

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

Intel still has an ARM licence.

It's not at all about Licence fees but Ego and Margin.

1
1
Anonymous Coward

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

"It's not at all about Licence fees but Ego and Margin"

What margin did Intel get out of IA64, overall? Whose egos at Intel/HP were so big that IA64 lasted ten years longer than it should have done?

I think we should be told.

0
1

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

RISC Vs CISC is almost irrelevant in modern devices. A modern CISC device has an instruction set decoder producing a stream of RISC micro-ops; that's really not that different to running a java/c# byte code interpreter/JIT compiler producing a stream of native instructions.

0
0

Re: I am so sick of Intel's whining about ARM versus Atom or CISC whatever

Intel only did RISC in the past because they didn't want to share with AMD. Because of a court ruling, anything Intel implements with the x86 CPU's AMD can implement, and vice-versa. Intel made the Itanium RISC CPU to break away from that. The problem was people really cared about their old programs and legacy compatibility. I know I do. It was at this time that AMD capitalized on Intel's RISC CPU project and created x86-64. It was far superior than the Pentium IV but because of Intel's illegal and underhanded tactics AMD had a hard time getting their CPU to take off. Some motherboard OEM's even had to sell their motherboards under a false name in order not to anger Intel. Slowly companies started to sell AMD CPU's. When HP ditched the Itanium, even though they helped Intel make and market it, for the Opteron then Intel saw the light. They deep-sixed the Itanium and with full resources back in x86 they soon were able to make a better CPU than AMD.

0
0
Silver badge

We all have the ARM device

We have ARM devices and know they work well. The Intel devices are an unknown quantity, and that makes Intel's road an uphill climb here. Obviously many were disappointed by previous generations of their product - but it was only recently that the premium ARM devices became passable. They are not so far behind. If they can get some units in people's hands this problem will go away. Benchmark arguments aren't going to do it.

Fortunately it looks like they have some killer reference designs in the pipeline. If they don't skimp on display quality, and do what they must on pricing they they should be OK.

1
0
Silver badge

Better windows than windows?

This was the argument that IBM made with OS/2. And it was true. So true in fact that, because companies could use OS/2 to safely run their multiple DOS/Windows programs, they didn't buy OS/2 software.

If binary translation from ARM to x86 is so good that no one notices, no one will bother doing native x86 versions. Developers won't really care if it becomes a push-button option in their IDE.

But Intel's real problem will be convincing manufacturers that it won't jack up prices once it has killed the competition. Currently, Intel is able to sweet talk some manufacturers into using its kit because it is a useful bargaining chip with Qualcomm. But ARM is developing faster and the number of manufacturers is increasing: Qualcomm, Broadcom, Samsung, nVidia, TI, Mediatek, …

1
0

Given that it's a Samsung tablet

Want to bet that a the 9% of misbehaving apps are just because of TouchWiz?

2
4
Bronze badge

There are three kinds of lies: my benchmarks, your benchmarks and his benchmarks.

2
0
Bronze badge

Superior?

x86 is superior to MIPS and Alpha for running Windows.

x86 is superior to ARM and anything else for running Linux for the same reason - binaries are more commonly available for x86.

So as ARM is the de facto standard for Android, that settles it equally. Most Android apps don't use native code, it's true, so there is some degree of flexibility. However, if one is building an Android tablet, using a ColdFire or SPARC or MIPS or PowerPC chip is just willful and self-indulgent perversity even if one could sort of get away with it.

1
3
Silver badge

Re: Superior?

Not particularly if you compare Like with Like running Native code. The Problem with MIPS and Alpha (in NT3.5 / NT 4.0) was lack of native versions and also the x86 only Win3.x and Win9x were the main MS OS then. NT3.5 and NT3.51 rarely used other than servers. NT4.0 started to see much higher Workstation penetration but hampered by MS and Channel promoting Win9x to Business (It originally was really meant to be Home Gaming OS). Win9x even killed the Pentium Pro which ran NT4.0 great but was a pig on Win9x because so much of it was really 16 bit Win3.1 with Explorer Shell.

The Pentium Pro even outperformed the later PII running Win98 or NT4.0 if running NT4.0 and 32bit applications. The cost of RAM also help kill the Pentium Pro too.

SPARC or MIPS or PowerPC etc have been Niche products for years now. Even routers and Set-boxes rarely use MIPS now, that was last major market for MIPS. Apple abandoned Power PC 10 years ago? Oracle doesn't care about SPARC.

Of course no-one will make a tablet with SPARC or MIPS or PowerPC unless a new insanely cheap SoC version of those is released for a Tablet.

1
0
Anonymous Coward

Re: Superior?

"x86 is superior to ARM and anything else for running Linux for the same reason - binaries are more commonly available for x86"

?????????????????????????

Presumably that's why so many Linux-based chunks of consumer and professional electronics (TVs, routers, NAS boxes, phones, etc) are x86-based.

Perhaps you could name a few widespread examples, because I certainly couldn't.

Whereas if readers look around their homes or offices, every Designed for Linux device to be seen or bought will almost inevitably have ARM Instead.

That may change. But till it does, x86 is only commercially relevant where Windows is relevant. And that's not a growing market.

2
0
Bronze badge
Coat

Re: Superior?

Oooo, this brings back memories! Suddenly I'm back in the days of the StrongARM and the sale of DEC Semi to Intel. All those Acorn RISC PC users that suddenly realised that they had, much to their chagrin, "Intel Inside"! And yes, I was one of those Acorn users!

Heh! Never mind. We got over it.

In other words... move along! Nothing to see here!

0
0
Anonymous Coward

Lies and FUD on both sides

I had an Orange San Diego for a long time and had absolutely zero issues with it - all the apps I tried worked perfectly (not into gaming so did not try any). Also, the battery life was better than my other phone, an HTC Evo 3D.

I not a fan of Intel, but the Orange San Diego was very good value for money, much like the Moto G is now.

0
0
Bronze badge
Paris Hilton

Ditto most of above

The group intel convened in November 2013 had little budget to spend and were given a 6 month deadline to come up with something substantial to slate ARM?

If so, is this the best it can do?

0
0

The real story here.

It is great that intel is managing to run these applications on their legacy devices. It would be a shame to have to clutter more landfills with unloved intel mobile devices, so any technology that gives them a sliver of usefulness is good news for everyone.

1
0
This topic is closed for new posts.