Oddbean

IMHO far too early. RISC-V processors aren't competitive yet. I suspect it is like this: 1. RISC-V instructions are simple and have short decode paths, meaning clock cycles could be faster, but 2. Higher frequency clocks create hard to deal with electromagnetical effects, so they can't do that 3. Instead they get low power, which is good for the phone, but not if the CPU is too slow. 4. They also can scale up to lots of cores 5. SiFive has done a lot of multi-issue out-of-order work to optimize their chips, but still not competitive for phones I think, and that kind of thing might lead to Spectre-like bugs. There probably will need to be some kind of hybrid.

+10 years or so for software optimizations to catch up to the new arch. The amount of work done on both the software side and the hardware side to make these things performant and power efficient is insane.

I don't have the technical knowledge about it as you apparently have. But it seems promising seeing that there are a few SBC in the market already working. There's also the pinetab 2 which is a quad-core @1,5Ghz. Not superfast, but yes lacking a lot of software support yet. Also the Linux phones running on arm lack a lot of battery optimization, but it's very interesting how fast some problems were solved like standardization of OS images by using tow-boot. A problem android wasn't able to solve in more than a decade.

I hadn't seen tow-boot. Was familiar with u-boot though. This is interesting: https://wiki.pine64.org/wiki/PineTab-V I have a SiFive HiFive Unmatched running ubuntu. I tried to write low-level code to do what u-boot should do on that hardware (wind up the PLLs, flip some switches, etc, the stuff the HW manual says must be done at power on) and then print something to the screen. But it just printed garbage to the screen and I never figured out why my code wasn't working... and then I discovered nostr and my life changed.