Slashdot Mirror
Why Mobile Innovation Outpaces PC Innovation
Sandrina sends in an opinion piece from TechCrunch that discusses why mobile systems are developing so much faster than the PC market. The article credits Intel with allowing hardware innovation to stagnate, and points out how much more competitive the component vendor market is for smartphones. Quoting:
"In PCs, Intel dictates the pace of hardware releases — OEMs essentially wait for CPU updates, then differentiate through inventory control, channel / distribution and branding. Intel and Microsoft win no matter which PC makers excel — they literally don't care if it's Asus, Dell or HP. In the smartphone world, it's the opposite. Dozens of component vendors fight each other to the death to win designs at smartphone OEMs. This competitive dynamic forms an entirely different basis for how component vendors approach system integration and support. Consider Infineon, which supplies the 3G wireless chipset in the iPhone. In order to stay in Apple's graces, Infineon must do everything necessary to help the hardware and software play well together, including staffing permanent engineers in Cupertino or sending a team overnight from Germany. Do you think Intel does this for Dell?"
If Intel was holding everyone back with your proposed CPU and Chipset conspiracy, don't you think that would just prime the market for AMD to pair up with VIA or someone and just wreck Intel?
AMD tried hard. They introduced 64-bit x86-compatible CPUs. And Microsoft wouldn't support them until Intel caught up. On the other hand, Microsoft supported the Inanium until 2004.
Because PCs have a headstart of decades?
It's like asking why China can have growth rates of over 10% while "Western" countries only get 1-3%. It is very hard to improve if you're already close to technical and physical limits and any made improvement won't look as impressive. Handhelds will soon enough hit the same walls that Desktop Systems currently try to tear down.
The "dominance" is the x86 instruction set. Intel and Microsoft have locked us in; AMD is just a second source for chips that use that instruction set.
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
Incorrect.
AMD introduced a 64-bit/32-bit hybrid CPU as a competitor of the Itanium for the server market. Opterons were and still are quire successful in that market, especially with the new g34 socket and 12-core processors (up to 48 cores per server and no tier-BS - all processors can run 1-4 SMP configuration) Microsoft viewed AMD's technology as *superior* to Itaniums because it allowed for seamless migration from 32-bit to 64-bit platform. Microsoft essentially *told* Intel that they will only support *one* 64-bit CPU and that will be the AMD instruction set. Intel had no choice but to incorporate AMD's instruction set into their processors.
Microsoft doesn't care if AMD or Intel catch up to each other as long as their software runs on those processors. They didn't "wait" for Intel to catch up. It simply took many years to migrate Windows from 32-bit code to 64-bit clean code. There was XP 64-bit, but how many people used that? Hell, lots of people didn't even get 64-bit Vista because of perception that if you don't use more than 4G of RAM you don't need it. Actually, all modern machines should be running 64-bit OS only - simplified address space management and increased register count makes it a no-brainer.
If you want an example of a company that still fails and fails hard at 64-bit software, it would be Adobe. They recently dropped support of the 64-bit plugin. Not sure, maybe they are still "waiting for Intel to catch up"?
Don't forget that the mobile market gets to take advantage of knowledge and research done for the server/desktop market. Sure, there's new tech going on in there, but it's the whole trickle down approach, too. The mobile market is *catching up* to the desktop market, so there's a lot of acceleration just from using all of the prior knowledge. Building multi-core processors isn't easy and how many mobile phones do you know that are sporting them? Zero that I know of. And what about Intel's turbo processing (dropping cores and overclocking the remaining cores when not needing as many cores), how long do you think before a mobile phone will have that technology?
The innovation in a lagging area (mobile) seems faster only because the innovation has already been researched in the leading area (servers first and consumer second). It takes longer to figure out something the first time than it does to figure out how to make it "smaller" (smaller in the sense that it is for the mobile market, it may be a smaller die footprint or power footprint or whatever).
I point to this fallacy:
Consider Infineon, which supplies the 3G wireless chipset in the iPhone. In order to stay in Apple’s graces, Infineon must do everything necessary to help the hardware and software play well together, including staffing permanent engineers in Cupertino or sending a team overnight from Germany. Do you think Intel does this for Dell?
Dell is not comparable with Apple in this case. Apple develops the operating system software for the iPhone. Intel also has permanent engineers at Microsoft, just like Infineon has engineers at Apple. Microsoft develops the operating system software for the PC. Intel also funds many Linux driver developers, and has staff working specifically on Linux support.
There are multiple x86 vendors including Intel, AMD, VIA. The reason there is not more competition is that Intel exploits network effects leveraged by their market monopoly which lead to the current situation. It used to be at a time that the chipset was manufactured by different vendors than the CPU. This enabled more rapid progress in some cases (e.g. ALI and VIA had a chipset with onboard 3D graphics long before other vendors). This is no longer the case. In fact it seems chipsets are becoming increasingly irrelevant as more things get integrated in the same chip. Intel is starting to include the graphics card and high speed I/O in the processor chip. Eventually the chipset will be today's equivalent of a slow I/O south bridge. Perhaps it will even vanish completely.
Another reason that mobile devices will not leave the PC industry behind is that Intel has superior manufacturing prowess. Historically Intel has had inferior chip design capabilities: the 8086 was inferior to the 68000, the 486 was inferior to many RISC processors, the Pentium Pro was inferior to the Alpha, etc. None of this mattered because Intel had the ability to deliver in volume and price where its competitors could not. The Pentium Pro, for example, had similar integer performance to Alpha because it had superior manufacturing, even if the hardware design was worse. Today Intel enjoys a healthy manufacturing process lead over all their competitors. It is a matter of time until they develop a specific chip to attack the smartphone market, like they developed Atom to counter the rising MID market, or Centrino to counter Transmeta years before.