Embedded x86: Birthday Blessings And An ARM Up
This blog post references my cover story ‘Embedded x86: Keystone Of Your Non-PC Design?‘ in EDN’s May 29, 2008 edition. It’s one of a series of web addendums to the print writeup.
It was 30 years ago (in three days), that Sgt. Intel taught the x86 to play.* Happy (early) birthday to you. And many more…
Key to the x86 architecture’s longstanding dominance, of course, has been its ascendancy at the expense of other microprocessor approaches. In that light, although my recent cover story was x86-centric and didn’t therefore didn’t discuss other architectures in any detail, the ARM spectre was certainly lurking between the lines. ARM is clearly in Atom’s gunsights; this is no surprise to ARM and its various licensees, who inquire as to my desire for a comment from them anytime anything Atom-related gets publicly burped by Intel.
The bulk of ARM’s current success derives from the cellular phone market and I agree with ASUS’ chairman’s comments last week; primarily from a power-consumption standpoint, ARM’s probably safe here for at least another turn or two of the x86 product and process screw. However, as is probably no surprise to most of you, mobile phones’ voice communication capabilities are over time becoming a decreasingly important percentage of the entire package’s feature set, with computing-centric functions proportionally coming to the forefront. In that light, with all due respect to ARM and its partners, they’re in for a firefight beginning at 32 nm. And, with past history as a guide to the probable future, the outcome looks bleak for ARM.
The Siren songs of inexpensive and fast-evolving silicon, abundant experienced x86 programmer talent and low-to-no cost development tools are nearly impossible to ignore. I realize that by now the cellular industry has assembled a formidable library of ARM-targeted code. However, that software’s increasingly being created using high-level languages and is therefore capable (in a relative sense, compared to the assembly language alternative) of being ported to a different CPU family. Witness Apple’s broadening of OS X and its various apps to x86 and (yes, I know) ARM in recent years. And witness the plethora of already-available x86-targeted software for the beyond-telephony applications that’ll be dominant in handheld devices in the future.
See what I mean? For more on this topic, I’ll direct you to the following additional-reading recommendations:
- x86 Evolution Still Driving the Revolution
- Mr. Stokes’ treatise ‘RISC vs. CISC in the Mobile Era‘ (and accompanying Slashdot commentary and discussion)
What do you think; has ARM jumped the shark, or are its best days yet to come?
Brian Dipert commented:
Bonkers commented:
Brian commented:
