A Clash of Architectures: ARM vs. x86

RE: A Clash of Architectures: ARM vs. x86

clem57 — Mon, 22 Dec 2014 18:33:08 GMT

Good read showing how things change but no change is apparent to the end user. The battle will continue with no clear winner for some time in the middle. I bet ARM/ATMEL are MCU winner and X86 is the high end winner!

Clem

RE: A Clash of Architectures: ARM vs. x86

D_Hersey — Thu, 06 Nov 2014 17:09:34 GMT

I think ARM and Intel uPs are coming from really different places. The Intel uPs have their etiology in the desktop, Acorn aside. Only their new, deprecated versions lack the pipelining and speculative execution that make using them in a real-time context so difficult. ARM needs to make coding to their devices easier. It shouldn't take twenty CMSIS function calls (or so it seems) to gun-up some section of the I/O.

RE: A Clash of Architectures: ARM vs. x86

DAB — Wed, 05 Nov 2014 18:35:27 GMT

A lot of your issues have been around for a long time.

The differences between different computer architectures can make an impact in certain applications, but in general, most user devices can be accomplished in any architecture.

I still think that the software support issues drive most of the issues. Speed at any price comes with a high price and limited tools.

A general purpose processor like the X86 line was intended for the 80% solution market. It does most things good enough. This design philosophy enable the chip to dominate the broad PC market in cost and performance. Special applications might benefit from other architectures, but the end product needed to have a wide appeal to make the change cost effective.

Today, the software costs dominate. So you have wide support with those processors with extensive software support.

Software compatibility still matters. I am running software today that was written fourteen years ago. It still does what I need done, but it runs faster on the new machine.

Always do a broad cost/benefit analysis before you choose an architecture. Look at all of the issues, especially support and development tools.

It usually takes a huge issue to beat the GP off the shelf device. Especially if you are going to produce your end product in any quantity.

DAB