ARM CEO Dismisses Intel's New Mobile Attempt
ARM disses Intel's Medfield, calling it "good enough." Ouch.
In an interview at CES in Las Vegas, ARM CEO Warren East called Intel's latest attempt to enter the mobile market "good enough." The rival's new Medfield finally came out of hiding this week during the show, revealed to be heading to Android smartphones manufactured by Motorola Mobility and Lenovo for starters.
Yet East also acknowledged Intel as a serious competitor, and that the x86 giant will inevitably score a few smartphone wins. "Are they ever going to be the leaders in power efficiency? No, of course not. But they have a lot more to offer," he said. "They (Intel) have taken some designs that were never meant for mobile phones and they've literally wrenched those designs and put them into a power-performance space which is roughly good enough for mobile phones."
But even though Intel's current entry is "good enough," East said that ARM won't take Intel's x86 assault lightly. He pointed to the company's new Cortex A7 design that features the performance of chips used just a few years ago, but with 20-percent of the power consumption. This will obviously help consumers do more with their smartphones while battery sizes remain relatively consistent.
"It's like having a car with a fixed-size fuel tank and you want to drive 100 more miles," he said. "You've got to make the engine more efficient. That's what we do for a living."
East went on to talk about Microsoft and its decision to bring ARM's architecture into Windows 8. He also admitted that Android is "the flavor of the month" and "flavor of the year," and ARM wants to be a part of that. But there's also a special place in ARM's heart for Microsoft's flagship operating system, and seeing Windows 8 running on tablets using ARM chip designs is just simply "exciting."
"We've waited a long time for this to happen. Another six months, another 12 months doesn't matter," said East. "I'd much rather wait however long it takes to get a quality experience than compromise."
Sounds like true romance.
Watch out, or you will be left behind!
I do believe that only Intel and AMD hold x86 licenses, unless you mean ARM competing with x86 in PC's.
ARM has an edge in ultra low power. That is completely different than performance/power, which is very difficult to compare across performance levels. If you were to need processing on the order of what a 6 core Xeon server can put out, all cores working on the same problem, there is no ARM solution, no matter how many processors you use, that can produce the performance. This is because they have an extremely poor performance data path.
Alot of Apple fanbois/fangrls will be suprised when their Macbooks don't run any traditional apps because they didn't include any ARM to X86 emulators! That would be hilarious
Intel has a better process than any of these companies do, and it isn't by just a little. Also, almost all are just using ARM's IP for their cores, so they really aren't all that differentiated from one another.
As for ARM getting into the PC market, that is so much harder than Intel getting into phones that it isn't even funny. Who is going to buy a gimp notebook with no software? The PC software ecosystem is enormous compared to the several year old smartphone market. That is what they will be trying to sell. However, in the cell phone market, the browser is most of the value. Then 75% of apps run in an interpreted language, which means they work on multiple platforms. The remaining 25% are games, which will be something of a challenge, but not the headache that ARM will have trying to get into the x86 PC market.
x86 was able to beat RISC architectures in high performance computing because there was no legacy software and they will beat ARM in mobile because there is no legacy software. The issue has always been a question of when the power/performance would start to overlap enough that ARM would be vulnerable to x86.
Agreed, I certainly wouldn't dismiss Intel in any market segment, especially mobile (Graphics might be another story, at least for now). If I was ARM I would at least be concerned about Medfield, especially considering Intel's rapid release schedule in the coming years for their Atom based SOC lineup. The latest performance/power consumption previews of Medfield are nothing to laugh at, in fact they're quite impressive:
http://www.anandtech.com/show/5365/intels-medfield-atom-z2460-arrive-for-smartphones
Contrary to all the hype, ARM has no chance of displacing more than a negligible percentage of the x86's market. A lot of observers fail to recognize that as with anything else engineered, there's more than a single determining factor. And whenever someone engineers a piece of hardware to be good at one thing, they have to trade off another quality. Often, this also involves specialization: limiting an advantage to just one particular sector.
Low TDP is far from everything. It may be the most critical factor for CPUs for phones and tablets, (and significant in netbooks & ultrabooks) but in desktops, conventional laptops, workstations, and servers, a low TDP is definitely not the #1 concern. Outright performance-per-price (or at least performance-per-CPU) is a bigger concern. In this realm, x86 dwarfs essentially all competitors, across a wide spectrum of platforms; this is the product of both allowing rather high clock speeds, as well as what is generally an unrivaled performance-per-clock; both real-world and synthetic benchmarks show that ARM consistently places behind even Intel's Atom CPU in PPC; and in turn, the Atom is by far the worst x86 CPU currently made in that respect, with most Phenom, Core, and even Bulldozer CPUs averaging the range of up to quadruple the PPC, particularly in the floating-point operations that are increasingly more critical than integer. (where the difference is closer to merely double)
Another factor I hinted at there is scalability: this is where we see Intel and AMD having some problems in trying to push into ARM's market. This difficulty is NOT any sign of any general superiority of ARM over x86, but rather, that ARM is superior at working at very low TDPs, and hence why we only see Atom and Bobcat typically competing only down to the 5-10w range. This is double-edged: performance-wise, we're seeing that ARM can't scale up very well at all; the fastest CPUs I've seen announced from them only hit 2.0 GHz, which doesn't quite compare to a modern 1.0 GHz x86 (non-Atom) CPU, which means that the WEAKEST non-Atom CPUs Intel and AMD are putting out for general consumer sales is more powerful than the most potent ARM CPU around.
You could make a pretty safe bet that one couldn't just clock ARM faster to compensate: it'd have to reach speeds past 6.0 GHz just to compete with the lowest-end CPUs found in laptops, which are, from Intel, running at TDPs of the neighborhood of 17w for two-core chips. Already, that puts the per-clock performance of core i7 close to that of ARM. And ARM's TDP does not evenly scale with TDP: each extra 100 MHz requires more wattage than the previous increment: even if an ARM Cortex COULD be scaled up to such multi-gigahertz speed, its TDP would definitely cease to be impressively low; it would likely go into the 40w+ range, which would simply not cut it; on a performance-per-watt figure, it wouldn't match up with *ANY* x86 CPUs: it'd have no place in laptops, desktops, or servers.
That's why ARM remains where it is: for phones and tablets, the performance expected and required isn't remotely near what it is for even normal laptops. However, it's just fine for such ultra-portable gadgets. By the same token, existing x86 CPUs can fulfill the needs for PCs just fine, but don't scale down too well. In other words, the day ARM replaces x86 on desktops is the day people stop using desktops: since why would you want the weaker power of an iOS or Android device with the bulkiness of a PC?
Actually, VIA does as well; they acquired it, I believe, through their purchase of Cyrix, the only other remaining manufacturer of x86 CPUs.
Thinking ARM would replace x86 is like thinking the Toyota Prius would replace all cars, vans, and trucks just because it has better mileage. As I mentioned above, ARM can't give you the power for serious gaming. It can't even yet quite match a gaming console.
To put it another way... Why would you want something with the power of an antiquated Pentium 4 for your desktop, even if it does offer a lower TDP? And even for the non-enthusiast audience, why would you want such a desktop, when you could get comparable performance out of a smaller, more sleek phone, tablet, or ultrabook?
Moot. If you can port most of your x86 windows activities to ARM (and arm 64 bit) then ARM doesn't need to support x86.
It would be arguable that for transitioning to ARM/mobile Windows it would be easier if ARM did have x86 support, more specifically if you could just install whatever apps you needed to an ARM based Windows 8 machine, mobile or desktop.
But anyone considering walking away from x86 windows to embrace ARM based Windows will have as much heartache as moving from any previous version to the next. Will app devs support that migration? Not traditionally, and not in the short term either.
You can still run software written for Windows 3.1 on Windows XP. I don't know about later, as we couldn't try – the floppies had given up, not to mention the motherboard didn't have a floppy connector. So I have to disagree with that. Also, I hope you realise that a lot of software depends on external libraries which the developer has no control over – hence it isn't quite as simple as just compile your software for ARM rather than x86.
You are so wrong. Intel has the most advanced fab and they're expert at the manufacturing process. In fact, one of the key reasons Intel keeps beating AMD is they have superior fab and AMD is always one generation behind playing catch up. Even if Intel has an average design, their manufacturing process can overcome their shortfalls. When Intel have a good design (like Core 2 Duo until now the i series), AMD has no way to catch up, because they couldn't beat Intel in either way. The only time AMD ever beat intel was during the Athlon vs Pentium 4 era.
FYI, Apple don't do manufacturing. Samsung made the chips for Apple in their iPhones and iPads.
For Intel and ARM, look at it this way...Intel is performance king, needing to simplfy their designs to be power efficient....ARMS focus on power efficient and works to improve performance. ARM dominate the desktop? Not going to happen EVER. They probably can't beat AMD at desktop performance, let alone Intel! But it is possible Intel makes headways in the mobile market.
Not quite that simple. VIA had to renegotiate the terms of their x86 contract when they purchased Cyrix. The licensing agreement is non-transferrable. All ARM would be permitted to do is continue to support and produce VIA's current line of processors.
You forgot to mention that most of ARM's "partners" are licensing their manufacturing tech from either IBM or Intel.... If Intel stops licensing their manufacturing tech to TSMC....most of ARM's "partners" would be completely screwed.... The HKMG "manufacturing technology" that TSMC is using for ARM, AMD and nVidia chips....is licensed from Intel. The Tri-Gate transistors that it was mentioned that ARM is looking at for their next processors....once again, licensed directly from Intel. Given that so much of the manufacturing technology is licensed from Intel....it wouldn't take much for Intel to bring ARM down.