RISC-V CEO seeks 'world domination' by winning over the likes of Intel The CEO of RISC-V's governing body says she wants to nothing less than "world domination" for the rising open-source processor technology, but to do that, the nonprofit needs buy-in from a variety of organizations, even those steeped in dominant, proprietary architectures, such as x86 giant Intel. In an interview this week …
I had a real problem when I first skimmed through this. About halfway through it began to dawn on me that "Redmond" is not what Vulture Central usually means by the word. Especially as seeking to dominate the world by getting in bed with Intel is very much in character.
I will avoid certain obvious puns which might be construed as sexist, though in which direction I am less sure.
"...but it's obsolete today and there's no real need for it."
No, maybe not but is it costing you anything?
What is it with people, new to the forum, moaning about what other people do with their own time, money and effort?
People do this sort of thing because they want to and can do, what harm is it doing?
I think that the more of this sort of thing that goes on the better. If it doesn't interest you, then ignore it.
I'm merely suggesting they could make the world a better place by not wasting their talents on obsolete stuff nobody cares about.
If they want to do if anyway, I have no quarrel with that, I merely think it's a shame. There's still so much that can be improved.
Much of the appeal of old stuff is that it's possible to comprehend every aspect of the system. New stuff you're stuck behind API's and abstractions upon abstractions to achieve anything; and is frankly "work" rather than fun. I have lost any and all interest in trying to maintain software on modern, unstable platforms whose dependencies jump every 30 seconds.
There are ways to avoid that problem of course, chiefly, avoiding so called rapid application design environments...
Old stuff will always hold an interest because of exposure of the interior, and, is a very useful learning tool.
The DevTerm is an existing product that uses a Raspberry Pi Compute Module as the CPU. They've just substituted a CM3-compatible board with a RISC-V module, which they also sell separately for $28 for if you already have an ARM-based DevTerm or want to use the RISC-V module on another CM-compatible motherboard.
If Intel is on board then Microsoft will be as well. After all, they have to keep the WINTEL monopoly flying high, don't they?
This gives MS the perfect excuse to ignore Windows on ARM from now on. That looks like it is a dead duck.
If their OS assets won't run on silicon from the likes of Apple (and others) then they and Intel can raise a huge glass of something nice and say 'Job Done'.
"The job is not done until Lotus won't run'
Yes folks, we have been here before.
If this fails then heads will roll in Intel and MS but not the top people naturally.
Microsoft's products have been running on ARM for a while now... There's even been a 3rd party ARM-Windows laptop announced recently, from Lenovo - the first in a line of ARM laptops to come, no doubt, in response to Apple's M1 silicon. This is not the route that Microsoft want to go though - they're looking to DAAS as the future : something that's ISA agnostic. My guess is that someone will produce a processor with a translation layer similar to Apple's Rosetta.
We're still a way off RISC-V consumer-level computing, I think, but it's more to do with software than hardware : FOSS OSs boot on RISC-V silicon (with some struggle), but there's a lot of work still to do to rebase the necessary apps. https://lists.riscv.org/g/software/message/174
My prediction though is that Microsoft will lose market share because of the uptake of RISC-V - particularly in China, and in Chinese made RISC-V products being exported : the consumer doesn't really seem to find the idea of DAAS appealing.
I don't think Microsoft will lose market share to RISC-V, at least not anytime soon. My guess is that ARM will lose market share to RISC-V in the microcontroller market for web cams, refrigerators, smart light bulbs, and smart whatevers. These controllers don't have the software compatibility issues that a PC/laptop has, since they are fixed, single-purpose devices.
This post has been deleted by its author
The WinTel monopoly is based on x86 and that will never, ever change.
They've tried switching to ARM several years ago. If failed miserably. They tried switching to Itanium too with even worse results.
I don't see them switching to RISC-V having a different outcome. The duopoly is simply based on the enormous amounts of software written for the ISA and the unwillingness of ISV's to port it to another.
Apple have shown the way on how to migrate one arch to another. Rosetta in it's latest incarnations works very well indeed for X86-64 on ARM. Much better than the lagfest that was Rosetta for PPC to Intel.
It is not merely compatible but of comparable or better performance. In my limited sample size of comparing performance of Reason 11 running on an M1 Mac Pro with a 2013 intel mac pro.
Itanium tanked for, amongst other reasons, a failing of compilers to deliver on promises. Availability was never wide enough to rely on customers to roll their own tools to improve the platform. To say nothing of the hardware being a steaming brick. Intel had an ARM project around the time of the 486. The world wasn't ready for that, not in the face of Windows 3.1, and nor was emulation practicable on hardware of that time.
Rosetta in the more convoluted Windows space would be tougher than on Apple but certainly not impossible, and bypasses the recompile everything problem neatly; until the next version when your vendor will want to say they have re-engineered for it.
X86 has made drastic improvements of 40 years in terms of numbers of instructions complete per clock cycle. When talking about frequencies north of 3GHz; executing marginally more instructions per cycle makes an enormous difference; and so ARM represents a jump over what X86 can do. Moving arch is not as difficult as it once was, and the performance advantages cannot be ignored. Let's face it - once you have a working assembler, you are not far off a working C compiler; and then in turn you can build practically anything from that.
Programmers motivated to learn assemblers are rare; and perhaps even more so on hardware still in it's evolutionary / developmental stages. They are the difference between a platform becoming viable or not. Apple recognised this, and took advantage of what it had already done in the mobile space to realise it on desktop.
If Intel doesn't do it, someone else probably will. See Chromebook (for all it's foibles). Or AMD on X86-64 vs. Itanium.
This gives MS the perfect excuse to ignore Windows on ARM from now on. That looks like it is a dead duck.
Ironically it is Raspberry Pi themselves who are more proactive in discouraging running Windows on a Raspberry Pi than Microsoft.
It isn’t a decision for Raspberry Pi. When MS decided to support Raspberry Pi with the IoT version of Windows, it was welcomed by Eben, who made some comments about MS Windows validating Pi as a proper computer.
It is Ms that would have to do the work to support api, and it is MS that have chosen to go with a sap dragon. For their ARM version.
This post has been deleted by its author
Swiss neutrality only goes so far.
Switzerland has, for example, joined the EU in sanctioning Russia over the invasion of Ukraine, and frozen many Russian accounts.
Switzerland is neutral when it chooses to be so and moving a company there won't protect it from sanctions if pressed hard enough. Doubly so for anyone contributing to said company from outside Switzerland.
All this gained traction while ARM was hanstrung by Softbank's rule, under which there's been all sorts of bad decisions. But what I think one of the worst among them is the fact that Softbank had ARM invest heavily on the Internet of Things sector.
Well, yes IoT is a thing but was it a right decision a company like ARM, who supplies the best in class architectures in the mobile area, focused on cheap, low power, small processors, sacrificing a chance to increase shares in desktop and server? ARM under Softbank even bought a company called Treasure Data, a California-based company processing IoT data, which btw was actually founded by the Japanese.
As for RISC, they totally get it. That's why they're investing heavily on the high power areas. They rightly believe that even as ARM now being selected by Apple computers and cloud operators like AWS, ARM's indecision for the last five years has rewarded them an invaluable chance.
What the academics don't seem to understand is that a company doesn't just switch between ARM and RISC-V based on raw performance figures on a like for like CPU. You have an entire ecosystem to deal with, tools and support and software. The don't switch due to the aesthetics of an ISA.
Someone like Apple can pull this off in 5-6 years as a political ploy to rid themselves of the accountants and boomers at Intel chucking out fusion reactors to their laptop customers. They own their entire ecosystem. Once someone has decided to ditch x86 for ARM or ARM for RISC-V, it is not about the ISA, that is a decision made in a board room years before any designs are even on paper.
I think the benefits of switching from x86 to ARM/RISC-V are far greater than switching between RISC-V and ARM. Now that the big boys are locking in to their own ARM implementations it could leave RISC-V very few opportunities to show itself off.
The one problem I see with RISC-V is, ironically, its openness. It allows unfriendly nations which are on the sanctions list to keep up with Western technology simply by implementing homegrown RISC-V parts. If the West moves towards RISC-V nations like China, Russia and even North Korea will benefit and this may not be something that's in our interest.
Maybe I'm corresponding a weird angle, but has anyone even thought about this?
China and Russia are political problems, not technical problems. You can't restrict tech based on which random country the West has annoyed/been annoyed by this week.
In any case, designing a processor ISA isn't actually all that hard. Pretty much any CompSci undergrad could do it...
In the end only politics matter.
If anyone can produce their own ISA why haven't China and Russia done so? Well, they have, but they haven't been very successful with it because they don't have the industrial and technological capacity of the West. They can only copy and mimic, at best, not produce something that's truly innovative or substantial.
By banning them they'll start to look like the old Soviet Union in two decades or so, where they only produced backwards stuff that the West had long since abandoned, like vacuum tubes (valves in British English).
To be truly inventive your mind needs to be free and that can only happen in a free society, which China (and Russia, and North Korea) isn't.
This is why truly groundbreaking inventions are (mostly) done in the West. Authoritarian societies can only come up with small variations or improvements on something that was invented elsewhere.
To be truly inventive your mind needs to be free and that can only happen in a free society
I don't believe that's true; one only needs to feel free enough to be truly inventive. If you don't want or need a so-called freedom which a nation won't allow you won't ever miss it. It then becomes more about having access to resources.
One can argue the freedom of the west, which allows polarisation and deep division, terrible social injustices, hatreds and disadvantages, ultimately holds a country back more than authoritarianism imposing a unity of direction does.
Each has its pros and cons; the trick is finding the perfect balance between the two extremes.
Even if the CCP started to wage a war against us right now, there's nothing we can do about them using ARM architectures. ARM is in the hand of Softbank, a Japanese investment company highly exposed to Chinese assets, and more importantly they've sold off a majority stake in the Chinese venture to companies bearing names like "China Investment Corp", etc.
So in the case of ARM, as of now we don't own it and we don't control it in one of the "unfriendly nations" and its biggest market. Looks like a bleak scenary if we were to depend the West's future in semiconductor on the company, isn't it?
RISC-V is indeed open source and basically everybody can use their ISA but usually when you design a chip, you don't start it all over again from scratch. And the biggest library developer of RISC-V right now is American company Si-Five. Not a UK company and it employs a lot of Asian talent but at least it's based in the US.
I'd call it a tie comparing the risk you're worrying about. Would like to see the Gov nationalise ARM but could it be realised?
I'm suggesting that ISA that's open could benefit nations that don't have our interests at heart, to put it mildly.
Whether it's a tie is something that's very much in the air right now. If the tech belongs to private companies situated in the West or friendly nations we can at least pressure them into denying this tech to unfriendly nations.
A good example is the export ban of EUV lithography machines from ASML (Dutch) to mainland China. Had their tech been open-source the Chinese would simply have rolled their own.
Given all you've said recently on the news about Arm-China:
"looks like SB/ARM already conceded additional chunk of ARM China to the Chinese investors"
"virtually zero prospect of the Chinese venture getting included in the expected IPO..."
"Wu himself has ... a 16 percent stake in the JV."
"ARM has been totally wrecked since the delisting/acquisition"
and
"So it seems ARM's Chinese business is gone and so is at least a quarter of ARM's annual revenue. Adding insult to injury, the Chinese Arm is already making a move to become a formidable rival to ARM in the region, by having been actively developing their own IP"
... which I thought was very insightful, thank you...
I'm not at all convinced that advocating the nationalization of Arm Ltd is good for the UK. At best it would be the acquisition of a - in my opinion - way-overvalued asset, and at worst it would be that PLUS a UK trade dispute with China starting with the suing of Chinese OEMs for using Arm IP. (I don't see the current US-China dispute a) being effective or b) being of much benefit to the US).
If the UK were interested in investing in the industry (and I agree with you, that it hasn't shown much appetite yet) it would be better off "Doing an India" : invest in fabbing, look at developing IP-free chips, and get some mentoring going.
I don't see a CPU ISA as free speech.
By allowing them to use our tech we're enabling them to compete with us, both economically and militarily. Preventing them from doing so has no impact on free speech.
Besides, free speech or true democracy isn't a given in those aforementioned countries. Why should we allow them the benefit our our independent justice system when our companies on their markets are beholden to the fickleness of their kleptocracy?
As far as I see it they have no rights here and we can do with them as we please or what benefits us the most. Only true and equal democracies can have the benefit of fair treatment here.
China, Russia, and N Korea can (and do) produce X86 ISA or ARM ISA parts as well. Hell I could probably cobble together some (painfully slow, single threaded) silicon using discrete parts that that would honor an X86 ISA. RISC-V is an ISA... Instruction Set Architecture, ie a list of low level primitive commands about slinging data between registers and memory and maybe an I/O bus if it's not memory mapped.
Making silicon CPU's that properly implement the ISA, and offer performance, and pipelining and branch prediction, and ... That is a whole other ballgame. The best 'western technology' would actually be pretty much in Taiwan at this point in the far East but I digress.
It's the implementation of the ISA that matters, not the instructions themselves.
I would have to agree. Intel are unsurprisingly in the Risc-V game only for what they can get out of it. I suppose the fact that they are not a charity and want to maintain long term dominance of at least one sector of computing is all the explanation they need.
They can see it coming, RISC V is already winning the hearts and minds battle and is starting to snowball. It will be used in education and develop just like ARM did, stealing ground behind ARM as ARM steals ground from Intel. It is ARM's next move that is going to be really interesting because they can't go on like they are forever.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
