I do not blame them, just by looking the history ,if some companies are extremely monopolistic and greedy e.g. Microsoft, and now Apple...and asking to much for their goods, services, they direct calling a competitor to destroy their hegemony e.g. AT+T and the ip phones. And to note something to one previous comment:"Remember how successful DoD was in mandating use of ADA? This too will implode. No worries man." The marketplace ultimately rules." 1) China is not the US, 2)China is the largest market with the size of it's middle class which is = the size of the whole US population,3) it is not impossible that they will learn from the mistakes happened elsewhere -e.g.US - and make a better system.
Didn't Bloomberg speculate about a week ago that MIPS was for sale?
China, with its bags of cash, could simply pick that one up (watch the USG try to block the sale) and own the IP as well as any existing licenses.
I would think the available tool chain would be a strong influencing factor. It's hard to make a new hardware architecture. But it's probably harder to make a decent optimizing C compiler (and related tools) for it. An architecture with an existing and mature GCC implementation would make a lot of sense.
Don't they already have their own CPU architecture - the Longmarch CPUs? This was a copy of the MIPS ISA. This is also apparently what they use in their supercomputers. They will soon realize that defining an ISA is the easy part. Getting folks to design and implement competitive CPUs around that ISA and the ecosystem around it is what makes an ISA succesful.
How about the OpenSparc architecture that was made open source by Sun? Apparently you can download the RTL for free. China likes free.
While I can understand why China prefers independence, to jump ahead they need more than an ISA, the latter almost being irrelevant today because of the heavy software layers.
What they need is the whole eco-system of associated EDA tools (generate ISA from specs), software development tools (compiler, simulator, RTOS, ...) and programming models. In the end, the determining factor will be fast they can crank out new applications, migrate to the next semicon technology at lowest power consumption and highest performance. I will take years before all that is in place.
Blog Doing Math in FPGAs Tom Burke 24 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 ...