[sdiy] The magic of ARM

Rainer Buchty rainer at buchty.net
Sat Aug 10 19:14:26 CEST 2013


On Fri, 9 Aug 2013, Lanterman, Aaron wrote:

> Very interesting! So it's not that ARM is *technically superior* to 
> MIPS, or AVR32, etc? They basically nailed a better business model?

I would say this was the door opener, yes. During my Agere Systems 
times, we looked into several IP vendors, and ARM was the clear winner 
with regard to the licensing model.

Today, I'd say that its widespread use also creates further momentum, 
making ARM some sort of "de-facto standard".

As for your question "does the PPC have a future": as an IP core 
probably yes. As a dedicated processor manufactured by IBM, however, I 
don't think so. Unless they have some government contracts which 
cross-finances the further technological development, both of the 
architecture and the manufacturing process (or "tick" and "tock" in 
Intel terms).

The question is even whether ARM has a future. With embedded devices 
getting more and more performance-hungry, they require similar features 
like desktop CPUs. Intel seems to have learned now how to make their 
CPUs sufficiently power-efficient; OTOH, ARM would need to install 
features like out-of-order execution, speculation techniques etc. -- 
which all costs chip area and, implicitly, also power, therefore 
thwarting one key benefit of the ARM architecture.

You already see where this leads to when comparing high-performance ARMs 
with their performance-corresponding Intel counterparts. The gap is 
closing.

Rainer



More information about the Synth-diy mailing list