disclaimer: I’m not a Mill employee, I’m just summarizing what I’ve found out from a bit of recent reading
They’re at least 1-2 years from having something that the rest of us can really mess with, outside of attempting to verify that what they’ve currently put out could work via an online compiler. Due to a change in how patents were awarded (from first to invent to first to file), the company had to shift gears from developing the CPU tech in stealth to getting patents filed on all of their relevant advances so that they stand a chance in the current market.
@angersock will probably feel a bit snarky about this post, but I’m bullish on the Mill CPU bringing something pretty revolutionary to the table. It’s just going to take 3-5+ years before we’ll see any actual fab of the chip, since the startup has been trading time for money until this past year.
I can see this potentially becoming a big deal on either servers or phones, since those are both markets where MIPS/Watt matter a lot, and where superscalars are starting to peter out. Moore’s law is slowing, though new chips are still markedly faster than tech from 5-6 years ago (as I’ve been made rather brutally aware on my aging laptop), trying different ideas like this is where we’re going to see a lot of advancement in the next 25 years, barring a change from silicon to some other computing media.
The other reason I’m bullish on Mill being realized is that the company has not been avoiding scrutiny once they had enough to show people. If a properly solid rebuttal to the arch exists, I’m not aware of it. I’ve seen a lot of people suggest that because the Mill has shades of the Itanium in it’s design that it’s likely to have a similar ill fate. Nobody has given a conclusive argument as to why, however.
What Mill Computer Inc have is rather crazy by conventional standards, but they seem to be focusing on the right problems (Tool chain, porting Linux Kernel, LLVM integration, flexible arch so they can find a niche) with the right attitude (asking for scrutiny and welcoming skeptics). They still have quite a few hurdles to overcome, but I’m excited to see just how far they’re able to take this arch. If in 15-25 years it have ends the dominance of undefined behavior and makes context switches much better, that alone would be a huge win.
Interesting. Thanks for the update/info! I wasn’t able to find any news / updates from the team after 2015 or so. If you found a email list or update stream that isn’t dead let me know.
What’s the current status here? I’m excited about this project.
disclaimer: I’m not a Mill employee, I’m just summarizing what I’ve found out from a bit of recent reading
They’re at least 1-2 years from having something that the rest of us can really mess with, outside of attempting to verify that what they’ve currently put out could work via an online compiler. Due to a change in how patents were awarded (from first to invent to first to file), the company had to shift gears from developing the CPU tech in stealth to getting patents filed on all of their relevant advances so that they stand a chance in the current market.
@angersock will probably feel a bit snarky about this post, but I’m bullish on the Mill CPU bringing something pretty revolutionary to the table. It’s just going to take 3-5+ years before we’ll see any actual fab of the chip, since the startup has been trading time for money until this past year.
I can see this potentially becoming a big deal on either servers or phones, since those are both markets where MIPS/Watt matter a lot, and where superscalars are starting to peter out. Moore’s law is slowing, though new chips are still markedly faster than tech from 5-6 years ago (as I’ve been made rather brutally aware on my aging laptop), trying different ideas like this is where we’re going to see a lot of advancement in the next 25 years, barring a change from silicon to some other computing media.
The other reason I’m bullish on Mill being realized is that the company has not been avoiding scrutiny once they had enough to show people. If a properly solid rebuttal to the arch exists, I’m not aware of it. I’ve seen a lot of people suggest that because the Mill has shades of the Itanium in it’s design that it’s likely to have a similar ill fate. Nobody has given a conclusive argument as to why, however.
What Mill Computer Inc have is rather crazy by conventional standards, but they seem to be focusing on the right problems (Tool chain, porting Linux Kernel, LLVM integration, flexible arch so they can find a niche) with the right attitude (asking for scrutiny and welcoming skeptics). They still have quite a few hurdles to overcome, but I’m excited to see just how far they’re able to take this arch. If in 15-25 years it have ends the dominance of undefined behavior and makes context switches much better, that alone would be a huge win.
I too hope that it works out–it’d be great to see a new and novel architecture!
I’m just a bit worried about their getting of good compiler support.
See? No snark. :)
I mean, compiler support is going to be a challenge, though since they are developing one themselves that should at least be a start.
Interesting. Thanks for the update/info! I wasn’t able to find any news / updates from the team after 2015 or so. If you found a email list or update stream that isn’t dead let me know.
https://millcomputing.com/topic/mill-computing-in-2017/
That’s the most recent I’ve seen, combined with the marked visual update.
It’s been some time since I last visited their site, and it looks like they’ve finally updated it to be a bit easier to navigate