Actually intel have tried to introduce non-x86 RISC architecture in 1990s; i860 and i960 family. They are truly innovative at the time (860 was first VILW, 960 was first superscaler in the market). 960 achieved some success in embedded application (860 was not a great success), but eventually overtaken by cheaper and higher-performance competitors (MIPS, PPC and ARM). After all, brand new architecture was not juicy for intel, compare to legacy x86, where they had definitive advantage in stable market. Well, those are things of past. It may be the time to intel to consider if they should stick with shrinking PC market or to make gamble on brand-new-architecture one more time.
It appears that people have fogotten that ARM and other "embedded" processor vendors are all the desktop/PC losers against Intel.
Intel didn't go to embedded market because PC market was very profitabl. Now it is the face-off time.
Can anyone be specific about what Intel is offering in terms of architectural innovation? ARM used the approach of elegant simplification right from the start, and to my knowledge haven't diverted from that very much. It is a virtuous circle, although one might expect diminishing returns at some point.
Intel have innovated their processes and silicon design, but their architectures seem to need all the help they can get from the boys in the basement. I never understood why they couldn't make a go of RISC, they spun out their ARM licensing experiment to Marvell I believe.
Maybe corporate Intel can't get over the 'not invented here' hurdle, or maybe Intel's architecture department doesn't contain any software gurus? I think they should look at what they do well (silicon), and what they do clumsily (architecture) and get focussed, either with some proper partnerships or invest in a proper architectural skunkworks (like Acorn Computers did when they created ARM).
Mind you, Intel's peripheral (chipset) architecture teams are pretty good. They have had to be, innovating while carrying all that legacy stuff.
A few things to consider, first when competition exists ... the market creates more innovation. Intel has fab technology to push products into a regime that is leaving most ODM companies behind. So if we all wish to enjoy those cutting edge products into the future, I would be cheering for Intel to keep the spirit alive. Otherwise we will all be stuck with our mobile devices running apps on slow coventional CPU products.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...