You do need feedback, lots of it, but not necessarily from customers. It is not a social media platform on the web using the hottest new framework. Your feedback mostly consists of numbers from measurements. That's why they built a simulator first (and it has apparently been running for quite some time). That's also why they did work on a compiler that they did get something useful out of (they have since switched their compiler approach to making an LLVM backend instead).
But getting to a point where you can run it in silicon (even just as an FPGA) is a long, tough slog. Especially if you want decent/realistic speed. Getting good numbers on power use is also both hard and expensive.
And it's all worth nothing if they don't have the patents. And many of the innovations need the other innovations to be useful or even just make sense.
I met both Ivan and a German guy who works on the compiler late last year and have corresponded a bit with both. What I have seen so far fits very well together... but of course there are still many pieces that they haven't shown (and that I haven't seen either).
You don't build bridges like that, do you? Or skyscrapers?
What's the minimum viable product for something like this:
http://en.wikipedia.org/wiki/Great_Belt_Fixed_Link
http://en.wikipedia.org/wiki/Millau_Viaduct
You do need feedback, lots of it, but not necessarily from customers. It is not a social media platform on the web using the hottest new framework. Your feedback mostly consists of numbers from measurements. That's why they built a simulator first (and it has apparently been running for quite some time). That's also why they did work on a compiler that they did get something useful out of (they have since switched their compiler approach to making an LLVM backend instead).
But getting to a point where you can run it in silicon (even just as an FPGA) is a long, tough slog. Especially if you want decent/realistic speed. Getting good numbers on power use is also both hard and expensive.
And it's all worth nothing if they don't have the patents. And many of the innovations need the other innovations to be useful or even just make sense.
I met both Ivan and a German guy who works on the compiler late last year and have corresponded a bit with both. What I have seen so far fits very well together... but of course there are still many pieces that they haven't shown (and that I haven't seen either).