Linked by Thom Holwerda on Wed 11th Mar 2009 17:23 UTC
Hardware, Embedded Systems To further prove that analysts' projections are just informed guesses, two major analyst firms just presented a completely different outlook on the netbook market and where it is going. Even though both project major growth, one of them sees a very bright future for non-Intel netbooks, while the other sticks with Atom.
Thread beginning with comment 352718
To read all comments associated with this story, please click here.
Just look at history
by ElCabri2 on Wed 11th Mar 2009 22:29 UTC
ElCabri2
Member since:
2009-03-11

There's never been a long term benefit of a different ISA (Instruction Set Architecture). Ever. Look at Apple's transition to Intel, look at the failure of Transmeta, look at the failure of Intel itself making IA-64 prevalent on workstations. Sooner or later, the dominant ISA prevails. The performance and feature advantages of the products using different ISA at a certain generation are ironed out two or three generations later. There's nothing that stops x86 and it's extensions from implementing any feature, any characteristic that you can think of, just as there is nothing that prevents the English language from expressing whatever idea or concept that you want, even if this idea or concept was originally formulated in another language.

All what these companies want is to make money for these two or three generations worth of 10-20% marketshare that they can get.

Reply Score: 1

RE: Just look at history
by lemur2 on Thu 12th Mar 2009 00:12 in reply to "Just look at history"
lemur2 Member since:
2007-02-17

There's never been a long term benefit of a different ISA (Instruction Set Architecture). Ever. Look at Apple's transition to Intel, look at the failure of Transmeta, look at the failure of Intel itself making IA-64 prevalent on workstations. Sooner or later, the dominant ISA prevails.


This is only true in a market where software is exchanged in binary executable form, rather than in source code form.

Currently, there exist large software collections in either form. Having the source code available for at least one very large software collection makes alternative ISAs viable in the market.

The performance and feature advantages of the products using different ISA at a certain generation are ironed out two or three generations later. There's nothing that stops x86 and it's extensions from implementing any feature, any characteristic that you can think of, just as there is nothing that prevents the English language from expressing whatever idea or concept that you want, even if this idea or concept was originally formulated in another language. All what these companies want is to make money for these two or three generations worth of 10-20% marketshare that they can get.


While it is true that x86 can implement new features, it is not true that it can afford to get rid of its legacy overhead. The simple fact that the vast majority of the x86 software base is not exchanged in source code form means that for its continued existance, the x86 architecture MUST maintain strict binary application compatibility stretching back 20 years.

Ouch.

That is a significant burnden to carry.

Meanwhile, architectures that do not rely on strict binary compatibility, and instead "trade" in the huge and ever-growing base of software available as source code, are far more nimble. Without the burden of legacy binary compatibility, chips utilising this software base can readily be built with far greater power efficiency.

In many markets, "bang per buck" is not the primary deciding criteria any more. It is "bang per watt" that is more important in many devices, especially embedded or battery-powered devices. Legacy binary application compatibility is becoming ever less and less important.

Reply Parent Score: 5

RE[2]: Just look at history
by Delgarde on Thu 12th Mar 2009 01:27 in reply to "RE: Just look at history"
Delgarde Member since:
2008-08-19

This is only true in a market where software is exchanged in binary executable form, rather than in source code form.


Which is, face it, all markets. I sit here writing this on a 100% open-source desktop - a RedHat distro, running Gnome, running Firefox. And not a single one of those packages was built by me from source code - all of them binaries provided by someone else.

Open source code certainly makes it easier to port to a different platform, but if you think an ARM netbook will be a success just because people can build any package for it, well, no. Binaries are what the users need, and binaries are what someone is going to have to provide.

Reply Parent Score: 2

RE: Just look at history
by major86 on Thu 12th Mar 2009 09:18 in reply to "Just look at history"
major86 Member since:
2008-04-21

Do not forget that ARM is SOC's god compared to Intel. Maybe Atom is some unreal marvel, but chipsets that come with it are certainly not.
/major86

Edited 2009-03-12 09:20 UTC

Reply Parent Score: 1

RE: Just look at history
by Lennie on Fri 13th Mar 2009 00:13 in reply to "Just look at history"
Lennie Member since:
2007-09-22

Ths is mostly because of what lemur2 already said and because of the dominent position of Microsoft.

Reply Parent Score: 1