@@mor4y in its origins, from what I've heard, ARM actually standed for Acorn RISC Machine (nowadays is Advanced RISC Machine), so yes, he's not talking about RISC-V specifically
@@geckoram6286 the original arm desktop (strong arm was a name for it, can’t mind if that was the machine or the CPU) is worth looking up for anyone who remembers win95/98 or early colour Mac days just to see what else was on offer. I think one of acorns claims to fame was either bill gates or Steve jobs was using one of their machines for a desktop machine for a couple of years during the development of some of their respective companies biggest products
This isn't a message to RISC-V. What infrastructure did Linux have in 1993? What he is saying is if you want to be pragmatic don't jump into something with no ecosystem.
Considering that a compiler already exists in RISC-V, a lack of "infrastructure" is almost irrelevant. Your favorite linux distribution and all your favorite free and open source software can simply be recompiled.
@@dannygjk I can think of many concrete examples there that wouldn’t be true. Can you elaborate why you think otherwise? Because you didn’t back your claim or at least relate it to personal experience.
I usually share linus' opinions but have to disagree on this one. An ISA is a product of an infrastructure not the other way around, and it's the infrastructure of x86 and aRM that currently blocks OSHW. The problem is not that we don't know how x86 works because they don't document it, it's that even if you reverse it or get the documentation from intel, you are just not allowed to make anything that's x86 compatible. We are not allowed to do something, not unable. Intel's infrastructure itself is built in a way that the result, the ISA is closed by definition. Lawyers will always have the bigger slice of the decision cake than open-source activists.
"Never, ever buy into something that doesn't have infrastructure." This comment almost borders on fearmongering. One would be inclined to believe that Mr. Torvalds in vehemently opposed to innovation. And in what alternate universe is the x86 architecture considered to be open? Intel and AMD must license IP to one another in order to build their current line of processors. If anything, I'm seeing a big push right now for open source hardware. You can't be secure in something if you're not allowed to look in the box.
This is like saying: don't buy into an operating system which doesn't have any (or an ecosystem of) applications. Understandable and also very practical way of looking at it. As an example Linus clearly is much more in the open source camp (better way of working together) than the free software foundation (free software for ethical and long term reasons).
Open as in they produce manuals which say exactly how they work and how to develop software for them. This is opposed to some microcontrollers and SoC's which require NDA's for the manuals, strict licensing and/or closed source chunks of firmware to configure the processor.
Why does this surprise you? The man is still working on a kernel 28 years after he released it. He might have been slightly less opposed to change in his teens, but then again... Linux was an Unix re-write with existing GNU infrastructure. Everything checks out.
What he probably means is the PC platform is the closest thing we currently have to open, since with the ARM computers the kernel has to be recompiled with proprietary blobs and with android phones such blobs don't even exist. Of course the management engines and signed firmware throw the whole 'open' deal out the window but it's a brick in the path compared to the alternatives.
@@autohmae That doesn't make sense in a Linux environment anyway. Compilers already exist for RISC-V. Your favorite linux distribution and all your favorite free and open source software can simply be recompiled.
The infrastructure around the architecture wouldn’t be there without people willing to build it in the first place. It’s the people who don’t give a damn about the opinions of ego programmers and innovate even if something is not popular. I started programming in Rust back in 2015, everybody talked down on it. Now it’s the most beloved language.
by who?? you just lied your way into believing rust is something useful just as previously many js-idiots did... those are temporary trends, not real languages...
> Now it’s the most beloved language. Only to people who deep down wanna feel "cool" or "different" (but won't admit it). The biggest, most important and most used tools are still written in C/C++/Python/Javascript. But you are right that it has come a long way, I won't refuse that... D, C++, hell even Zig are all much better than Rust, lol!
"It's really hard to get [necessary numbers of] people interested in small hardware [openness] when the existing small hardware doesn't encourage that type of loving."
I have read the old comments and cleary only a few people seem to have understood the connections, to see the full picture. 1.) Despite how it may seem, the uploader did not intend to mock nor to show RISC-V or Torvalds in a wrong light. Quite the opposite is true. Among his other videos there is one, where the uploader shows enthusiastically a project achieved on a RISC-V develoment board, emphasizing the usage of open source tools. But why the uploader didn't change the title, remove or reupload a cleared video after critical comments? The uploader seems to be absent from his channel for quite a long time now. His last video upload is from 3 years ago. So whatever happend to him, i hope, he is ok. 2:) Linus Torvalds is clearly not against Open Source nor does he recommend a proprietary ISA over an open source one. Clearly that Video Clip suffers from unlucky video editing, because the cuts seem to transport a twisted and contradictory message from Linus Torvalds. To see the whole picture please watch the full original video talk below, organised by arm supporters. RISC-V wasn't even a topic in that talk, it seems. He talks about the advantages of the PC infrastructure over other systems. He points out its openness and he obviously does not mean the x86 architecure but the PC infrastructure. The ISA and the Core are not very important in that context. And of course Torvalds likes other interesting ISAs. But he talks in the context and advantages of the open PC infrastructure. For example you are free to exchange computer parts, like CPUs, GPUs and Mainboards like you want, as long as the PC parts follow common standards. And Linus Torvalds is right. The PC gives you more freedom, more choice and more possibilities than any other current infrastructure. Because of that, many wished from SiFive to develop a RISC-V mini ITX form factor board, which could fit into a pc. And SiFive did it. Torvalds mentioned his dream machine at that time, which did not come true, because "all the things around the cpu were not there". Here at the latest people should grasp slowly, what Linus Torvalds actually means. He is not favoring the x86 ISA but points out the importance of the infra structure around it. What can you do with an open source ISA or any other different ISA, when there is only a very limitited infrastructure around it. Because of the many possibilities and compatibility around x86 it became so established. So whoever wants to achieve something similar, they have to plant that idea in an open infrastructure like the PC is with as many compatible connections and standard interfaces as possible. "It's not about the instruction set, it's about everything around it." (Linus Torvalds) He mentions the problems, small hardware and IOTs have in comparison to PCs, when it is about reaching out to the people. And he means arm in that talk and context. Despite smartphones with arm chips being widespread nowadays, arm did not reach out to the PC platform, which offers a rich variety of customizable hardware and OSs. Arm didn't realize its big potential on the PC-platform and this is indeed a disappointment. "Being compatible just wasn't as big of a deal for the ARM ecosystem as it has been traditionally for the x86 ecosystem. I've been personally disappointed with ARM as a hardware plattform, not as an instruction set, though i had my issues there, too. As a hardware platform, it is still not very pleasant to deal with." But don't understand that wrong either and watch the full video, where he explains it all in detail. For example Torvalds recently mentioned, that he would love to see Linux running on a M1 CPU as well. So thanks for your attention. Here is the full talk. Watch and enjoy wa Salam "Fireside Chat with Linus Torvalds & David Rusling at Linaro Connect (LAS16-500K3) th-cam.com/video/fuAebQvFnRI/w-d-xo.html
By his own words we should not have jumped on the linux wagon, since when it came it was so limited that people had to "borrow" infrastructure (ie reverse engineer drivers) since most infrastructure was locked down.
ปีที่แล้ว
Maybe this sayings are out of a bigger context we don't get, but yeah, sounds pretty pessimistic, like "we have to stay where we are, don't bring changes".
Yea, this video seems to be pure BS. Linus tends to sound bad when taken out of context and this video is doing just that. Show the whole damn thing, not just one or two sound bites
So how do you get infrastructure if nobody starts using the newer or better system? That answer was probably due to all the years of frustration trying to get linux used and just not having the infrastructure to make it happen. I remember back in the early/mid 90's getting a linux disc and think wow a free operating system (we only had msdos or windows 3.1 back then. But there was no infrastructure, no software, no nothing for linux back then so it was hard to implement for everyday use. Maybe if people put some effort in to risc-v like many did back in the early days of linux we would be able to utilise the huge benefits of a more streamlined technology? I wonder how Linus feels about risc-v and risc architecture now that big companies are starting to use it in their hardware?
I think the point was that RISC-V won't expand until it offers something more to entice a user base. But as long as all of the boards available are closed blob city, that's not going to happen. I wrote a more detailed post but TH-cam censored it. 🤷
@@johnmackenzie3871 When Linus started Linux, the PC infrastructure was there and it was open. Anybody could look at the specs for ISA (PC Bus), for BIOS etc. It was an open standard and it was already crushing the alternative architecture (his point on the Archimedes) because it was an open arena for company to compete and develop powerful and cheap products. Linus is absolutely right, the ISA (chip ISA) is not important. It's just a flag in your compiler. However all that is around determine if you are able to have thousands of people working on their free time to develop drivers and tools. Having RISC-V be open is great but if every RISC-V boards implements it's own little SOC, nobody will care.
I have been working on a platform infrastructure on top of Linux Kernel for several years ... And, although it began on X86, later X86-64, I have been running it "without changes" on ARM32 and ARM64 (Banana-Pi, Orange-Pi and, of course, several incantations of Raspberry machines). And, by my surprise, I was able to recompile it WITHOUT CHANGES with the RISC-V cross compiler on my Linux development box. What I nice thing to do ... now, I will acquire some RISC-V machines for testing them. About infrastructure, well .. but I have it. The machines, including RISC-V, work with networking, and they have audio and video, they use USB and disks so what? I can interchange ANY machine with ANY of these CPUs and a working module on them, and what I see is that I don't care about what CPU they have if they are powerful enough to fulfill my requirements. At the end, the base infrastructure I need to guarantee interoperation is the Operating System (Linux in my case), and that I can compile my C++ code with a compiler, that already exist, in any of these machines. For me, this is a bright future :-)
@@toto123456ish Along with that he wasn't talking about RISC-V, just architectures in general At this time I believe Risc-V didn't have any sort of computers, just microcontrollers, so a clickbait hungry TH-camr probably saw it and decided to upload the clip that you see in the video with the caption of "Message of Linus Torvalds to Risc-V" for views
He's talking about Plug and Play/ACPI. This is the thing that other architectures have always sucked at. Having to manually figure out a device tree for platform/machine X is lame. It's always been lame. This is why there is so much Arm eWaste.
Linus has already pointed out its opening is a advantages itself, and also importantly it is vital to make more and more people care about it, more and more people get involved, only this way can make it bigger and greater
The whole market for computers used to be much smaller back then. Comparatively, it was nothing next to x86 today. The sheer difference in scale is pretty overwhelming.
LOL, not only did he completely miss the ARM revolution with embedded systems and SBCs, but now RISC-V is becoming massive in the next big thing: AI. And sure enough, Linux is embracing both ARM and RISC-V in really big ways. x86 is sort of limping along with its inertia, but come back in 10 years and see where x86 is. Or ARM for that matter. RISC-V is where all the excitement is these days.
At the time this was a true statement. ARM in the mean time broke into Intels domain. Especially Apples M-Processors becoming more and more dominant. Qualcom follows currently with an architecture which seems to rival the M processors and is ARM based as well. They have all the important technologies on board on the desktop. On the server side it will still take 2-5 years but then Intel and AMD have a massive problem with ARM. On the low end Risc-V is more and more improving. Being min 5 years behind ARM it is developing in high speed. The advantage of Risc-V is the fact that several market players from Asia, Europa and the US are pushing the technology into several directions. Therefore it is very well possible that eventually Risc-V catches up with Intel,AMD and ARM and surpasses them. But that will not happen tomorrow. In my opinion we are speaking here about 10 years.
The instruction set definitely matters. Why do programs requiring high floating point performance avoid the x87 FPU? Obviously because x87 instructions lead to lackluster performance and are now even considered obsolete. Why is it that AMD added 8 additional integer registers, 8 additional SSE registers, and the instructions to access the new registers all to the x86 architecture? Why is it that Itanium is on end-of-life? Clearly instruction sets make a big difference.
Instruction set kinda matters, but not really. Your instructions are useless if there is no adoption, no infrastructure, and no reason to pick it up outside of niche uses.
In programming perspective, Instruction set indeed does not matter, programs are not written in CPU instructions direclly now days because they became too complex to do so, you also have higher level programmers that treat even low level languages like C or C++ like some archaic ancient programming language and avoid is like devil spawn, i can see that in indie game developer community. What makes platform popular is infrastructure.... and Linux knows this more then any other OS on desktop, making your software run on it (and im not talking here about WINE) is not really that big deal, distribution and getting any user base on it is complitly different story
Maybe I'm more interested in Stallman's opinion of Risc-V then. Pragmatic thinking is not very interesting long term, it's mostly useful in the short term. Linus has always felt like the Trotsky of the FOSS movement.
Video was uploaded 6 years ago. Me: "oh that's a long time ago". Realizes that even smartphones existed back than: "shit 6 years are nothing for me, I'm old".
Linus seems to still live in the Acorn Archimedes era. I wanted one too - and yeah... he was right then. When Moores law ruled. But in the years coming Moors law will end and "anyone" will be able to create clean slate hardware and port much useful software to it. When Acorn Archimedes was new, no one could just recompile "Linux" and have a fully fledge OS on it with apps from day one. The fact that there's Open Source for almost anything including creating hardware combined with Moores law ending makes this a totally different situation.
You're literally the only person on this page saying this. "Infrastructure" means nothing to linux users. Compilers already exist for RISC-V, so just recompile your favorite linux distro and all your favorite free and open software.
@@gregorymalchuk272 So what? Compilers don't matter (because ISA doesn't matter). You need drivers, you need hardware. Can I use RISC-V CPU with NVIDIA GPU? No? NVIDIA uses RISC-V internally for their GPU you say? But who cares? ISA doesn't matter. Deal with it.
He's right, tooling > instruction set. As for RISC-V, when the market inevitably gets flooded with proprietary incompatible implementations, people will be longing for the "good old days of X86, which was proprietary but standardized." Fanboys don't like to hear that, but its the truth. Mark my words.
The way RISC-V is supposed to work is that the core instruction set is locked with a range set aside for proprietary instructions to control specific hardware. If some manufacturer deviates from that, then it'll be no different from the hordes of proprietary chips that cover motherboards already. If they don't, they give a free gaggle of people posting updates to libraries so they don't have to do the work themselves. At least with RISC, you don't have to pay ARM, Intel or AMD a license on top of the already expensive silicon production, which lends itself to hobbyist-level chips showing up.
He actually said bubkes about Risc-V. He said that small complicated proprietary architectures won't get the necessary interest to develop to something large.
This video editing kinda sucked. We have no idea what they asked him and when did he say it was about RISCV?
The quote about the acorn computer, it was the last attempt at a risc desktop
I knew someone who had one, it was quite the machine back in the day
@@mor4y Is not ARM a risc architecture? The last apple macbooks and notebooks powered by Qualcomm all have ARM in them
@@mor4y in its origins, from what I've heard, ARM actually standed for Acorn RISC Machine (nowadays is Advanced RISC Machine), so yes, he's not talking about RISC-V specifically
@@geckoram6286 the original arm desktop (strong arm was a name for it, can’t mind if that was the machine or the CPU) is worth looking up for anyone who remembers win95/98 or early colour Mac days just to see what else was on offer. I think one of acorns claims to fame was either bill gates or Steve jobs was using one of their machines for a desktop machine for a couple of years during the development of some of their respective companies biggest products
This isn't a message to RISC-V. What infrastructure did Linux have in 1993? What he is saying is if you want to be pragmatic don't jump into something with no ecosystem.
He not RISC-V should just be dropped more like that it's not ready yet
Considering that a compiler already exists in RISC-V, a lack of "infrastructure" is almost irrelevant. Your favorite linux distribution and all your favorite free and open source software can simply be recompiled.
@@gregorymalchuk272 not true tons of projects will be poorly optimized and broken in subtle ways for many years. Including build systems, JITs etc...
@@thecount25 Optimization is not much of a factor when it comes to RISC.
@@dannygjk I can think of many concrete examples there that wouldn’t be true. Can you elaborate why you think otherwise? Because you didn’t back your claim or at least relate it to personal experience.
He didn't even mention RISC-V. How is this a message?
I usually share linus' opinions but have to disagree on this one.
An ISA is a product of an infrastructure not the other way around, and it's the infrastructure of x86 and aRM that currently blocks OSHW. The problem is not that we don't know how x86 works because they don't document it, it's that even if you reverse it or get the documentation from intel, you are just not allowed to make anything that's x86 compatible. We are not allowed to do something, not unable. Intel's infrastructure itself is built in a way that the result, the ISA is closed by definition.
Lawyers will always have the bigger slice of the decision cake than open-source activists.
I don't know whether you'd like this link, but here it is anyway.
wiki.mises.org/wiki/Without_Intellectual_Property
It is became complicated when the isa and the infrastructure is evolving at the same time
Sometimes you have to do bootstrapping, but the advantages in existing infrastructure are hard to beat.
"Never, ever buy into something that doesn't have infrastructure." This comment almost borders on fearmongering. One would be inclined to believe that Mr. Torvalds in vehemently opposed to innovation. And in what alternate universe is the x86 architecture considered to be open? Intel and AMD must license IP to one another in order to build their current line of processors. If anything, I'm seeing a big push right now for open source hardware. You can't be secure in something if you're not allowed to look in the box.
This is like saying: don't buy into an operating system which doesn't have any (or an ecosystem of) applications. Understandable and also very practical way of looking at it. As an example Linus clearly is much more in the open source camp (better way of working together) than the free software foundation (free software for ethical and long term reasons).
Open as in they produce manuals which say exactly how they work and how to develop software for them.
This is opposed to some microcontrollers and SoC's which require NDA's for the manuals, strict licensing and/or closed source chunks of firmware to configure the processor.
Why does this surprise you? The man is still working on a kernel 28 years after he released it. He might have been slightly less opposed to change in his teens, but then again... Linux was an Unix re-write with existing GNU infrastructure. Everything checks out.
What he probably means is the PC platform is the closest thing we currently have to open, since with the ARM computers the kernel has to be recompiled with proprietary blobs and with android phones such blobs don't even exist.
Of course the management engines and signed firmware throw the whole 'open' deal out the window but it's a brick in the path compared to the alternatives.
@@autohmae
That doesn't make sense in a Linux environment anyway. Compilers already exist for RISC-V. Your favorite linux distribution and all your favorite free and open source software can simply be recompiled.
The infrastructure around the architecture wouldn’t be there without people willing to build it in the first place.
It’s the people who don’t give a damn about the opinions of ego programmers and innovate even if something is not popular.
I started programming in Rust back in 2015, everybody talked down on it. Now it’s the most beloved language.
by who?? you just lied your way into believing rust is something useful just as previously many js-idiots did... those are temporary trends, not real languages...
> Now it’s the most beloved language.
Only to people who deep down wanna feel "cool" or "different" (but won't admit it). The biggest, most important and most used tools are still written in C/C++/Python/Javascript. But you are right that it has come a long way, I won't refuse that...
D, C++, hell even Zig are all much better than Rust, lol!
"It's really hard to get [necessary numbers of] people interested in small hardware [openness] when the existing small hardware doesn't encourage that type of loving."
Yep - it really matters what Microsoft does...
I have read the old comments and cleary only a few people seem to have understood the connections, to see the full picture.
1.) Despite how it may seem, the uploader did not intend to mock nor to show RISC-V or Torvalds in a wrong light. Quite the opposite is true. Among his other videos there is one, where the uploader shows enthusiastically a project achieved on a RISC-V develoment board, emphasizing the usage of open source tools.
But why the uploader didn't change the title, remove or reupload a cleared video after critical comments? The uploader seems to be absent from his channel for quite a long time now. His last video upload is from 3 years ago. So whatever happend to him, i hope, he is ok.
2:) Linus Torvalds is clearly not against Open Source nor does he recommend a proprietary ISA over an open source one. Clearly that Video Clip suffers from unlucky video editing, because the cuts seem to transport a twisted and contradictory message from Linus Torvalds. To see the whole picture please watch the full original video talk below, organised by arm supporters. RISC-V wasn't even a topic in that talk, it seems.
He talks about the advantages of the PC infrastructure over other systems. He points out its openness and he obviously does not mean the x86 architecure but the PC infrastructure. The ISA and the Core are not very important in that context. And of course Torvalds likes other interesting ISAs. But he talks in the context and advantages of the open PC infrastructure. For example you are free to exchange computer parts, like CPUs, GPUs and Mainboards like you want, as long as the PC parts follow common standards. And Linus Torvalds is right. The PC gives you more freedom, more choice and more possibilities than any other current infrastructure. Because of that, many wished from SiFive to develop a RISC-V mini ITX form factor board, which could fit into a pc. And SiFive did it. Torvalds mentioned his dream machine at that time, which did not come true, because "all the things around the cpu were not there". Here at the latest people should grasp slowly, what Linus Torvalds actually means. He is not favoring the x86 ISA but points out the importance of the infra structure around it. What can you do with an open source ISA or any other different ISA, when there is only a very limitited infrastructure around it. Because of the many possibilities and compatibility around x86 it became so established. So whoever wants to achieve something similar, they have to plant that idea in an open infrastructure like the PC is with as many compatible connections and standard interfaces as possible.
"It's not about the instruction set, it's about everything around it." (Linus Torvalds) He mentions the problems, small hardware and IOTs have in comparison to PCs, when it is about reaching out to the people. And he means arm in that talk and context. Despite smartphones with arm chips being widespread nowadays, arm did not reach out to the PC platform, which offers a rich variety of customizable hardware and OSs. Arm didn't realize its big potential on the PC-platform and this is indeed a disappointment.
"Being compatible just wasn't as big of a deal for the ARM ecosystem as it has been traditionally for the x86 ecosystem. I've been personally disappointed with ARM as a hardware plattform, not as an instruction set, though i had my issues there, too. As a hardware platform, it is still not very pleasant to deal with." But don't understand that wrong either and watch the full video, where he explains it all in detail. For example Torvalds recently mentioned, that he would love to see Linux running on a M1 CPU as well.
So thanks for your attention. Here is the full talk. Watch and enjoy wa Salam
"Fireside Chat with Linus Torvalds & David Rusling at Linaro Connect (LAS16-500K3)
th-cam.com/video/fuAebQvFnRI/w-d-xo.html
By his own words we should not have jumped on the linux wagon, since when it came it was so limited that people had to "borrow" infrastructure (ie reverse engineer drivers) since most infrastructure was locked down.
Maybe this sayings are out of a bigger context we don't get, but yeah, sounds pretty pessimistic, like "we have to stay where we are, don't bring changes".
RISC-V content ZERO!!!
Yea, this video seems to be pure BS. Linus tends to sound bad when taken out of context and this video is doing just that.
Show the whole damn thing, not just one or two sound bites
So how do you get infrastructure if nobody starts using the newer or better system? That answer was probably due to all the years of frustration trying to get linux used and just not having the infrastructure to make it happen. I remember back in the early/mid 90's getting a linux disc and think wow a free operating system (we only had msdos or windows 3.1 back then. But there was no infrastructure, no software, no nothing for linux back then so it was hard to implement for everyday use. Maybe if people put some effort in to risc-v like many did back in the early days of linux we would be able to utilise the huge benefits of a more streamlined technology? I wonder how Linus feels about risc-v and risc architecture now that big companies are starting to use it in their hardware?
I think the point was that RISC-V won't expand until it offers something more to entice a user base. But as long as all of the boards available are closed blob city, that's not going to happen.
I wrote a more detailed post but TH-cam censored it. 🤷
@@KiraSlith Fuck youtube
Exactly what I thought. I wonder why he made Linux when there was more infrastructure around Microsoft ... It's the same thing.
Well said. Pretty insane how Linus doesn't even realize that had he applied his own logic to Linux, it wouldn't be where it is today.
@@johnmackenzie3871 When Linus started Linux, the PC infrastructure was there and it was open. Anybody could look at the specs for ISA (PC Bus), for BIOS etc. It was an open standard and it was already crushing the alternative architecture (his point on the Archimedes) because it was an open arena for company to compete and develop powerful and cheap products. Linus is absolutely right, the ISA (chip ISA) is not important. It's just a flag in your compiler. However all that is around determine if you are able to have thousands of people working on their free time to develop drivers and tools. Having RISC-V be open is great but if every RISC-V boards implements it's own little SOC, nobody will care.
I have been working on a platform infrastructure on top of Linux Kernel for several years ...
And, although it began on X86, later X86-64, I have been running it "without changes" on ARM32 and ARM64 (Banana-Pi, Orange-Pi and, of course, several incantations of Raspberry machines).
And, by my surprise, I was able to recompile it WITHOUT CHANGES with the RISC-V cross compiler on my Linux development box. What I nice thing to do ... now, I will acquire some RISC-V machines for testing them.
About infrastructure, well .. but I have it. The machines, including RISC-V, work with networking, and they have audio and video, they use USB and disks so what? I can interchange ANY machine with ANY of these CPUs and a working module on them, and what I see is that I don't care about what CPU they have if they are powerful enough to fulfill my requirements. At the end, the base infrastructure I need to guarantee interoperation is the Operating System (Linux in my case), and that I can compile my C++ code with a compiler, that already exist, in any of these machines.
For me, this is a bright future :-)
That video was 6 years ago, the landscape changed since then.
@@toto123456ish Along with that he wasn't talking about RISC-V, just architectures in general
At this time I believe Risc-V didn't have any sort of computers, just microcontrollers, so a clickbait hungry TH-camr probably saw it and decided to upload the clip that you see in the video with the caption of "Message of Linus Torvalds to Risc-V" for views
He's talking about Plug and Play/ACPI. This is the thing that other architectures have always sucked at. Having to manually figure out a device tree for platform/machine X is lame. It's always been lame.
This is why there is so much Arm eWaste.
Linus has already pointed out its opening is a advantages itself, and also importantly it is vital to make more and more people care about it, more and more people get involved, only this way can make it bigger and greater
Gosh that editing... :(
The IBM 360 was the dominant computer at one point in time. There was a huge infrastructure in place.
The whole market for computers used to be much smaller back then. Comparatively, it was nothing next to x86 today. The sheer difference in scale is pretty overwhelming.
And It's still in use and can run Linux! IBM Z backwards compatible to 1964!
Thats a 100% software oriented pov
Anyone remember the transmeta crusoe?
I used to have a flybook based on it
Nice click-bait title.
Anyone know of the source. I want to listen the whole talk so I can get the context.
LOL, not only did he completely miss the ARM revolution with embedded systems and SBCs, but now RISC-V is becoming massive in the next big thing: AI. And sure enough, Linux is embracing both ARM and RISC-V in really big ways. x86 is sort of limping along with its inertia, but come back in 10 years and see where x86 is. Or ARM for that matter. RISC-V is where all the excitement is these days.
At the time this was a true statement. ARM in the mean time broke into Intels domain. Especially Apples M-Processors becoming more and more dominant. Qualcom follows currently with an architecture which seems to rival the M processors and is ARM based as well. They have all the important technologies on board on the desktop. On the server side it will still take 2-5 years but then Intel and AMD have a massive problem with ARM.
On the low end Risc-V is more and more improving. Being min 5 years behind ARM it is developing in high speed. The advantage of Risc-V is the fact that several market players from Asia, Europa and the US are pushing the technology into several directions. Therefore it is very well possible that eventually Risc-V catches up with Intel,AMD and ARM and surpasses them. But that will not happen tomorrow. In my opinion we are speaking here about 10 years.
Should no one have built apps for Linux in the early 2000s then? Because Linux definitely didn't have the infrastructure to support server apps.
The death of the Motorola 680x0 family of microprocessors confirms what Linus said.
What does he mean by CPU infrastructure?
where is the message?
The instruction set definitely matters. Why do programs requiring high floating point performance avoid the x87 FPU? Obviously because x87 instructions lead to lackluster performance and are now even considered obsolete. Why is it that AMD added 8 additional integer registers, 8 additional SSE registers, and the instructions to access the new registers all to the x86 architecture? Why is it that Itanium is on end-of-life? Clearly instruction sets make a big difference.
Instruction set kinda matters, but not really. Your instructions are useless if there is no adoption, no infrastructure, and no reason to pick it up outside of niche uses.
In programming perspective, Instruction set indeed does not matter, programs are not written in CPU instructions direclly now days because they became too complex to do so, you also have higher level programmers that treat even low level languages like C or C++ like some archaic ancient programming language and avoid is like devil spawn, i can see that in indie game developer community. What makes platform popular is infrastructure.... and Linux knows this more then any other OS on desktop, making your software run on it (and im not talking here about WINE) is not really that big deal, distribution and getting any user base on it is complitly different story
Maybe I'm more interested in Stallman's opinion of Risc-V then.
Pragmatic thinking is not very interesting long term, it's mostly useful in the short term.
Linus has always felt like the Trotsky of the FOSS movement.
Video was uploaded 6 years ago. Me: "oh that's a long time ago". Realizes that even smartphones existed back than: "shit 6 years are nothing for me, I'm old".
I really loved that part were Linux said something about RiscV and it wasn't just 'eduardo', if that is your real name, click baiting us
...oh wait
flipper zero is open
But its pro-russian useless BS
Linus seems to still live in the Acorn Archimedes era. I wanted one too - and yeah... he was right then. When Moores law ruled.
But in the years coming Moors law will end and "anyone" will be able to create clean slate hardware and port much useful software to it.
When Acorn Archimedes was new, no one could just recompile "Linux" and have a fully fledge OS on it with apps from day one.
The fact that there's Open Source for almost anything including creating hardware combined with Moores law ending makes this a totally different situation.
You're literally the only person on this page saying this. "Infrastructure" means nothing to linux users. Compilers already exist for RISC-V, so just recompile your favorite linux distro and all your favorite free and open software.
@@gregorymalchuk272 It's a long time since I wrote that comment, but yes ... I think that was a part of my point.
@@gregorymalchuk272 So what? Compilers don't matter (because ISA doesn't matter). You need drivers, you need hardware. Can I use RISC-V CPU with NVIDIA GPU? No? NVIDIA uses RISC-V internally for their GPU you say? But who cares?
ISA doesn't matter. Deal with it.
@@RenamedChannel finally a person with a brain
He's right, tooling > instruction set.
As for RISC-V, when the market inevitably gets flooded with proprietary incompatible implementations, people will be longing for the "good old days of X86, which was proprietary but standardized."
Fanboys don't like to hear that, but its the truth. Mark my words.
The way RISC-V is supposed to work is that the core instruction set is locked with a range set aside for proprietary instructions to control specific hardware. If some manufacturer deviates from that, then it'll be no different from the hordes of proprietary chips that cover motherboards already. If they don't, they give a free gaggle of people posting updates to libraries so they don't have to do the work themselves.
At least with RISC, you don't have to pay ARM, Intel or AMD a license on top of the already expensive silicon production, which lends itself to hobbyist-level chips showing up.
He actually said bubkes about Risc-V. He said that small complicated proprietary architectures won't get the necessary interest to develop to something large.
Arm is way more open than X86. IBM made X86's infrastructure.
Wonder what his thoughts are about apple silicon's M1