Discussion in 'Article Discussion' started by CardJoe, 16 Aug 2010.
it's now ARM Holdings
Maybe they'll buy up a foundry and call it "ARM and Hammer"
I can see this being awesome on a number of fronts, especially for things like laptops and tablets with a small and fixed hardware base (think Apple). a 1.5 GB dual core chip is enough to run any application 98% of people will ever want to run, and coupled with a specialized version of linux or ChromeOS and an app store sort of format, you could have a real winner here.
How hard is it to program for ARM or to port x86 software to ARM?
yes, exactly. ARM Holdings sells licenses for the ARM instrution set to vendors (like Qualcomm and Via) who make the chips.
To stick with my original analogy, ARM Holdings getting an x86 license would be like Sony selling XBox Arcade titles in XBox 360 format on the Playstation Network Store: Desireable for those that want the extra architecture, but useless due to the incompatability in its current form...
ARM is a very clean architecture compared to the legacy cruft x86 has collected over the years. It's RISC so it has plenty of registers, comes with a compact instruction set for embedded systems where data storage space is at a premium and even has things like Jazelle, meaning a hardware Java VM straight in the chip (optional module).
While there are some differences between different ARM designs/chips, it's generally fairly consistent and aside from the need to align data (something x86 scoffs at despite the huge performance penalty), relatively easy to port to and to design new software for.
Well, consindering today's impressive power gating technology, as well as other power saving features, the way you need to look at it is thus; two cores doing nothing is going to draw just as much (or as little power) as one core doing nothing, and while two cores doing something is going to draw more power, it is also very likely to do it faster. The power draw will be higher, but not for as long, relatively speaking.
The performance per watt should increase, which is what we need. Power draw is irrelevant if you have to draw it for much longer.
Separate names with a comma.