4 ms·

RISC-V. Internet replaced (and far surpassed) proprietary corporate and government communication systems such as X.25, Compuserve, ATT.net, AOL. Linux replace

RISC-V. Internet replaced (and far surpassed) proprietary corporate and government communication systems such as X.25, Compuserve, ATT.net, AOL. Linux replaced proprietary OSes. It is the biggest mobile platform (Android) and the vast majority of servers, MacOS is a parallel and largely compatible Un*x, and Windows is slowly being assimilated by WSL. RISC-V is going to replace proprietary instruction sets such as x86 and Arm. Never entirely, of course, just as Mac and Windows and IBM S/360 have never gone fully away. They just lose their relevance to most people, and especially new users. I was using internet in 1989 when New Zealand universities got the first 9600 bps full time connection to Hawaii, and a local BBS I was using (Actrix) got a connection to the local university. I was then able to telnet to BIX in the USA instead of using crazy expensive X.25. Usenet groups and email updated in real time. It was amazing. Limited then, but clearly the future. I was experimenting with MkLinux part time on my PowerPC Mac in 1996, and x86 Linux on a borrowed Pentium 60 a year before that -- I had a contract to connect an app on the Mac to a mysql database on Linux. I bought a machine, a Pentium Pro 200, specifically to run Linux full time at home, in 1997. Linux also was limited at the time, but clearly the future. RISC-V is the same now, and I've believed so since I bought my first board, a HiFive1, in December 2016. RISC-V is already well established in embedded uses, taking significant market share from Arm and virtually killing off everything else. In the last two years RISC-V has started to appear in cheap (close to Raspberry Pi price) SBCs and in 2024 in low end laptops. I was playing around with various internet access methods in the 90s, including a weird system with download via a satellite dish but outbound data via modem which I got in I think 1997 (IHUG SatNet). And playing around with Linux. Both started to feel really serious when I got both a cable modem and Ubuntu in 2004. I think RISC-V will hit that stage in the next 12-24 months.

brodouevencode 2y ago
I don't know that it will fundamentally reshape compute as strongly as the Internet reshaped communications, but RISC-V is such an appealing platform. It's so powerful given power demands and cost.
- brucehoult 2y ago
  The question was "have a massive impact", which is not quite the same thing as "fundamental". Though being the Last ISA would be pretty fundamental. Unless they are really really bad, ISAs don't die because their design got old. x86 has proven that, over and over. As has been observed by a number of people, the only flaw that is really fatal is not having enough address bits. RISC-V is probably the only 64 bit ISA today with a built in plan for 128 bits. The main reason ISAs die is because they are proprietary and have only one vendor that goes out of business, decides to concentrate on something else, gets too greedy, misses the Next Big Thing etc. Same with Un*x's, actually. Something owned by no one and everyone avoids all that, as long as it gains initial traction and has a governing person/body that keeps things sensible but can be bypassed/forked later if needed.

brodouevencode 2y ago

I don't know that it will fundamentally reshape compute as strongly as the Internet reshaped communications, but RISC-V is such an appealing platform. It's so powerful given power demands and cost.

brucehoult 2y ago
The question was "have a massive impact", which is not quite the same thing as "fundamental". Though being the Last ISA would be pretty fundamental. Unless they are really really bad, ISAs don't die because their design got old. x86 has proven that, over and over. As has been observed by a number of people, the only flaw that is really fatal is not having enough address bits. RISC-V is probably the only 64 bit ISA today with a built in plan for 128 bits. The main reason ISAs die is because they are proprietary and have only one vendor that goes out of business, decides to concentrate on something else, gets too greedy, misses the Next Big Thing etc. Same with Un*x's, actually. Something owned by no one and everyone avoids all that, as long as it gains initial traction and has a governing person/body that keeps things sensible but can be bypassed/forked later if needed.