The open source RISC-V architecture and how it might eventually take on x86 and ARM

[Retro]

This is full nerd from the subject right down to the presenter of Explaining Computers and his presenting style.

EC does some analysis of the product and world market to predict that this architecture will end up being a viable third alternative to x86 and ARM in the future. I won't spoil it for you, just watch the video to find out.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Retro]

@Geffers this should be right up your street.

 

[Arantor]

Haven't watched the video - it's late and I should go to bed. But as someone who's been watching the journey for a while, someone who followed the original RISC tech on the Archimedes with interest for a viable alternative to the original Wintel era dominance... I think it's got some work to do.

ARM's been at this for a while - since the days of the Archimedes in fact! - back when they were Acorn Research Machines, and they were always at the front of what everyone was doing. I think Commodore gave them a good run for their money technically but ARM got their foot in the door with the educational sector in a way Commodore never did, and that's kept them in good stead: building low-power, solid dependable workhorse designs.

That's the thing RISC-V now has to figure out: where it wants to be in the market. I'd *love* there to be a good alternative to Intel/AMD/nVidia/Arm because competition is a good thing, but it's a complicated journey. On the one end you have Intel basically owning the high CPU end of the market, with nVidia getting in on the GPU market, vying with AMD (and what was formerly ATI). But AMD's gear is still ultimately x64 style at present because it's what people want to run.

Then you have Arm, which basically came out of nowhere with the smartphone generation with Qualcomm and Apple building out Arm chip-based gear. It comes in at a market segment where pure computational grunt isn't the only factor - you need decent performance with a (much) lower power consumption and (ideally) a much lower thermal envelope for mobile gear. No-one is going to stick a device in their pocket that is designed to cope with up to 105 degrees C in its interior. Intel stuff just doesn't play nice on mobile, because they sank all their money into fighting Moore's Law and winning the bigger numbers performance wars. And Arm stuff is widely used enough that it's worth investing in the toolchains to build for it - and of course even the performance argument is no longer what it was, with Apple's M1 range being surprisingly competitive.

So that's RISC-V's challenge: the high performance sector of the market at one end dominated by Intel/AMD, the low power/reasonable performance angle dominated by Arm, before you look at the ASIC and SoC specialised gear. What does RISC-V want to be? What does it do better than its rivals? Because that's what it has to be to take on the throne, it has to do something better (faster/cheaper/whatever).

Because if it's going to otherwise be another Transmeta or another Cyrix (or even Motorola), it might as well give up already; there's no shortage of also-rans even in the processor market. Being technically impressive just isn't enough any more.

 

[Retro]

Haven't watched the video - it's late and I should go to bed.

Nah, stay up, be wrecked for tomorrow, it's totally worth it!! But seriously, do give it a watch when you get a minute. It's very well presented by an extreme nerd. His explanations are very clear, in particular. I think it will answer at least some of the questions you've asked.

I also come from the Acorn era, having owned several 8-bit and 32-bit computers. I've owned several 32-bit Acorns so know what they're like and that A310 in the computer shop felt like a rocket powered car when I tried it out waaay back when it was relesased in 1987. Alas, I couldn't afford a grand for it then, still very expensive even now. My latest is a StrongARM Risc PC and that really flew for its time too. You've never seen a Basic listing fly past so fast!! btw, I still have them all... quite the collection now even though I no longer use any of them.

As for the low power thing, those visionaries at ARM foresaw the future very well at the time, around 1990. I remember going to one of their presentations shortly after they were set up where they played up the low power, low heat virtues of the ARM architecture and its use in portable devices that they were going to concentrate on.

I remember feeling disappointed that they weren't just aiming for outright performance, whichI think they'd have done well in given how efficient and streamlined that ARM architecture is (I've programmed it in assembler - epic). Now, 30 years later they're in billions of mobile devices and are the Intel of that huge market segment, so they got their product positioning right. I just regret not investing in the company at the time as I'd have been wealthy by now, sigh.

btw ARM used to stand for Acorn RISC Machine, but when the CPU division was spun off it became Advanced RISC Machines. There was a third renaming, but I can't remember what it was now. Wikipedia to the rescue if we wanna check.

 

[Arantor]

Huh, could have sworn one of the incarnations of the IP chunk of the business was called Acorn Research Machines (the chip was definitely Acorn RISC Machine) but I was convinced the original name of the original ARM chunk of the business was that. Age-related memory degradation, eh?

I was in the ZX camp until 1990 then went Amiga and was disappointed with the legacy there as you can imagine. The Archimedes was a thing I saw in school and loved since it seemed to continue after Commodore’s implosion, and I was never a fan of what the PC market was doing in that era. Still not really, those pioneers had a better feel for what hobbyists want to do with a platform than anything you see now.

Everything around 1990 was still passively-cooled but it was mostly Intel pushing for ever more transistors and the fight against Moore’s Law (in the “work harder not smarter” approach) that lead to things needing gargantuan heatsinks and the like.

I think RISC has some interesting avenues it can explore but no matter how good it is *technically*, it has to differentiate itself from its rivals on something other than pure technical wizardry. If it can bring a new position in the market that neither Intel or Arm can capitalise on, it’ll absolutely be worth watching.

I just remember the excitement of Transmeta back in the day as an Intel-killer…