I really, really want to like these guys - I have been following Forth, Charles Moore, and colorforth since first reading up on Forth in 1996 or so.

However, the site is a dog's breakfast, and despite clicking around I cannot seem to locate an actual binary to download and run colorforth on Windows. Where / when can I buy an eval board? How do I connect it to ethernet, or USB?

Further, it seems to be a company where you have a lot of smart guys who are retired, and they are just putzing around without a burning desire to really take over the market they are targeting.

BTW it seems each core only has a very limited amount of RAM on it, 64 "words" which means 32 bits on PCs and 18 bits (???) on the g18 core CPUs.

Sorry about the title, I couldn't fit in who Moore is and have it make sense. I wanted to say: "Chuck Moore, the inventor of Forth, announced GreenArrays new CPU: 144 cores, capable of 100 billion ops/sec"

Very naïf question from a software-only guy: given that such a chip exists, how is one supposed to USE this kind of thing? Like, maybe plugging some kind of card on a PC? Or is it just interesting for hardware manifacturers?

Anyone ever worked with one of his cpus?

Wow. As an undergrad (equivalent - I live in Brazil), the best project I did (the one I am most proud of) was a stack-based CPU that ran something very close to Forth.

This is really, really cool.

Some basic spec comparisons:

A Intel i7 900 series processor has a die size of 296 mm^2, with a process size of 45 nm.

The GA144 has a die size of 21 mm^2, at a process size of 130 nm, which is a couple generations old. If the i7 900 series was manufactured at 130 nm, it'd be 2488 mm^2.

So, it's smaller. A lot smaller.

I'd compare MIPS too, but that would be worse than useless, since it implies that it actually means something, which is far from the truth. Comparing raw instructions per second values between different architectures means nothing.

(Plus, Intel have passworded their datasheets!)

But it really doesn't look good. The Greenarray guys might manage some deep magic that makes their processor twice as fast per transistor, but it's still 118 times smaller. There's just not enough silicon!

100 billion instructions per second is what.. fast than the current Intel i7s? Am I reading this right?