Hacker News Reader

Comment by NortySpock | Hacker News Reader

Why will ARM succeed where MIPS faded into obscurity? (Well, relative obscurity. I see they are still used in media devices and cheap routers.)

https://en.m.wikipedia.org/wiki/MIPS_Technologies

View on HN

ARM has already succeeded. There is no "will" here. Most microcontrollers in the world are ARM. Most application processors in the world are ARM. It spans from the cheapest to the highest-performing parts. Nearly every smartphone in the world is ARM.

AArch64 is very, very late to the game, and has a lot to prove (beyond mobile).

The older architectures were much earlier (MIPS64 1991, SPARC64 1995, Itanium 2001, AMD64 2003, AArch64 2011).

Yes, AArch64 runs the top supercomputer, and there is enthusiasm; don't mistake this for market share.

In the embedded arena, Arm (the instruction set) would appear to be pretty replaceable. Arm and MIPS competed for the same business, Arm and RISC-V can do the same. There is nothing magical about any of them.

Arm (the corporation) can make some money for themselves by coming up with increasingly cool Arm cores to sell (a mission Nvidia could have aided immediately and with high impact, but... nevermind) but the instruction set itself is not especially important anymore. There's a lot less path-dependency for companies in the embedded arena than there is among, for example, desktop computer companies.

Arm was one of the CISCiest of the RISC processors at a happy medium in instruction complexity that gave them all the simplicity of regular sized instructions while still getting the job done in fewer instructions than a more ideological RISC ISA.

Some in depth look at various features here:

https://userpages.umbc.edu/~vijay/mashey.on.risc.html

Also, getting the job done with fewer instructions meant requiring less RAM in embedded roles where it thrived before entering the market from below in classic disruptive innovation style, the same way that x86 entered the server market. The other RISCs were mostly high end workstation or server chips suffering from low volumes. I actually ported some software from an SGI MIPS box to x86/Linux once. It was a nice system and I see why people liked it but it just couldn't keep up.

>that gave them all the simplicity of regular sized instructions while still getting the job done in fewer instructions than a more ideological RISC ISA.

Thumb2 in 32bit ARM was competitive in code density.

AArch64 code density is abysmal, dramatically worse than RV64GC.

well, it's by far the most common CPU architecture in the world.

my thoughts exactly.

>(Well, relative obscurity. I see they are still used in media devices and cheap routers.)

It's actually gone; MIPS owners have abandoned MIPS to focus on RISC-V.

Kind of a shame because MIPS is one of the more approachable ISAs if you want to learn assembly

RISC-V was designed to to be a teaching ISP. A minimal RISC-V is not so different from MISP. I don't think you lose much.

> RISC-V was designed to to be a teaching ISP

Do you have a source for this? RISC-V always seemed like it was a production ISA from the get-go

"RISC-V is a new instruction set architecture (ISA) designed to support computer architecture research and education."

"Provide a realistic but open ISA that captures important details of commercial generalpurpose ISA designs and that is suitable for direct hardware implementation."

"Be simple to subset for educational purposes and to reduce complexity of bringing up new implementations."

https://www2.eecs.berkeley.edu/Pubs/TechRpts/2011/EECS-2011-...

"Yunsup Lee, Krste Asanovi´c, David Patterson, and I conceived RISC-V in the summer of 2010 as an ISA for research and education at Berkeley. Two of our research ventures, RAMP Gold [92] and Maven [60], had just wound down. These projects were based around the SPARC ISA and a lightly modified MIPS ISA, respectively, and we sought to unify behind a single architecture for the next round of projects."

https://www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-...

I can't remember if I heard it in a Jim Keller interview or somewhere else, but I think students also helped initially design the ISA for RISC-V also.

I remember reading this somewhere a few years ago as well. RSIC-V was the "this is an educational ISA, but we're gonna see how far we can take it"

It is an educational isa that just so happens to have the highest code density among 64bit architectures.

There's been a few changes since its inception in academy and the ratified standard unprivileged spec[0].

[0] https://riscv.org/technical/specifications/

Its literally how it started. They wanted a modern ISA for their course. Then they did a '6 month project' to develop it. That turned out to be RISC-V. Then companies would literally just adopt it and be angry at them doing changes and such.

Then they realized that they need to push this more aggressively, establish a foundation and create a real standard.

You can listen to the 'State of the Union' talks from the early conferences if you want a source.

Once you actually look, you'll likely be surprised by how much RISC-V resembles MIPS.

ARM has been here before. I owned stock in Arm, before it sold.

While it's indeed good to have a healthy skepticism towards these things, there's a major difference between the two: a) MIPS has been acquired too many times by a company that can lock others out (SGI is definitely the Nvidia of today outside of consumer stuff), and b) MIPS, while have been present in game consoles and workstations, didn't really have the visibility ARM has today. They didn't have any consumer devices labelled with "MIPS chipset inside!", even though it was present in game consoles.

> Why will ARM succeed where MIPS faded into obscurity?

Because for the mass consumer market needing decent compute power, there is only ARM and x86-64 left, everything else has gone downhill in actual usage - and ARM is way easier to implement with a lot more vendors available than the x86-64 world which is basically Intel vs AMD.

The only competition that remains is Loongson for Chinese government purposes, but no one outside of China and maybe Russia will touch that stuff voluntarily.

What about OpenPOWER? It's held its own in the supercomputer race and I don't think IBM is going to stop using it anytime soon.

The POWER architecture family, since the Apple migration to x86 many years ago, is only relevant for supercomputers and mainframes which is why I didn't include it in the list.

Russia has it's own CPU, Elbrus, which is slow, features x86 emulation and VLIW architecture (i think).

It's meant to be used in areas of critical infra and national security, every major nation that is not joint with US at the hip is thinking or planning chip independance.

Otherwise your economy could be rouined because Donald Trimp 2 had too much cocaine in the morning and tweeted something.

MIPS had a few architectural eccentricities, like branch delay slots, that were tuned to the original design to make it fast.

These eccentricities eventually became a liability.

https://www.jwhitham.org/2016/02/risc-instruction-sets-i-hav...

what are you talking about? ARM is every where used by all the tech companies.

ARM is on mobile, data center and desktop/laptop.

Reply on news.ycombinator.com