Please see the disclaimer.
In my next post, I talk about a new chip design that uses some ideas from the Mill architecture.
Unfortunately, even if the design I have in that post is feasible, especially commercially, I cannot build it. That is because Mill Computing, Inc, the creators of the Mill architecture, have filed for, and been granted patents on, some aspects of its design, including some aspects that we are putting in this chip design.
This means that if we were to build this design, we would be infringing on their patents, so we can’t do that without a license.
This is extremely upsetting.
Before I heard of the Mill architecture, I was of the opinion that patents have no place in the software and computing industry, and the Mill has convinced me even more of that.
The reason is that, even if there are good ideas in the Mill, they can’t be used by anyone for about 20 years. (The patents expire in March 2035.) Of course, Mill Computing can use them, but they are the only ones, unless they license their intellectual property (IP). This means that no one can use their ideas in architectures of their own design, like the one we are designing now.
Not only that, but Mill Computing says that they are interested in becoming a hardware company first and foremost, since companies that actually make chips, like Intel, tend to make more money than those that license IP, like Arm.
That would be fine, except that Mill Computing does not seem to be capable of actually producing chips! They have had sims for about a decade, but they haven’t actually taped out anything, as far as I am aware.
Not only that, but their stated purpose of being a chipmaker contradicts the fact that they have patents in the first place! If they really wanted to make chips above all else, they would not really care about filing patents.
“But Gavin, they just want to protect their property!”
If that were true, they would have just made their designs in secret and then built chips. They would have been protected by several things:
- The cost to build a new fab, especially one to build a radical design like the Mill is cost prohibitive for anyone who would want to copy them.
- Even if an existing chipmaker, like Intel, could build a new fab, they would not do so because they have chips that are popular and cost relatively little to update.
- Even if someone at another chipmaker were forward-thinking and saw the benefits of the Mill, they would have to produce tools, like compilers and such, for customers to use the new chip.
All of those factors means that even if the patents were not protecting the ideas, the chance of the ideas being stolen is small.
“But Gavin, they are producing the tools that you say that other chipmakers would need to produce!”
They started in 2004. And they still don’t have the basic tools, like a C compiler, libc, a C++ compiler, libc++, a BIOS, and a basic operating system!
So if they have been going since 2004, 16 years, and don’t even have the basic tools for their chip, much less an actual chip, I would presume that they are another decade away from having a chip!
As an aside, when I was looking for a job, I was desperate, so I was somewhat interested in working for Mill Computing on their compilers. I emailed to apply for a job, thinking that I could get paid for it, even if it was a minimal salary.
Nope. They said that I would have to put in sweat equity, and when I said that I could only spare 2-3 hours a week for that, they said that wasn’t good enough.
This means that, as of when I was emailing them, they didn’t even have the funding to pay salaries! There is no way they have the funding to tape-out a chip.
In that time, no one can use any idea of theirs that has been patented, which means that, with respect to their ideas, the computing industry is frozen in place; it cannot progress.
That is why I hate their patents; the patents are actively stopping some of the progress of the industry!
In my opinion, even if they are trying to become a chipmaker, they will fail, and if they fail at that, they will try to license their IP. By that time, it may be so old that it would expire in a few years anyway, which means that everyone will just wait for it to do so.
And I would like to put a final nail in their coffin.
If I succeed in building a business, I will design a chip with their ideas before their patents expire, and as soon as they do expire, I will start building the chip.
I hope this makes their IP worthless because that might convince them to actually build a chip! I hope it will be the only thing they can do to survive.
But even if it kills them, maybe the IP will be sold to a company that will release them into the public domain, removing the dam of progress those patents have created.