Yeah I really don't think so since it is ten times slower than the pi3. Forget about being ready for the mainstream the hardware still needs to make massive leaps in performance before it is even considered ready for the mainstream.
Kinda like with Bitcoin / cryptocurrency too, there needs to be a use case / killer app besides just evangelism to get adoption to the mainstream / next level. Maybe the tighter coupling of open-source custom instructions and the open-source Linux kernel can provide that. Kinda like the hope that RISC-V will be beneficial for the “AI” bubble/craze.
I can't believe Chris speaks at a normal pace. I suppose he's presenting style is just the style for the type of videos he wants to make. His video style reminds me of old/vintage BBC science and technology programmes.
to add some info here is how a full date would be shown English "5th august 2024" Chinese "2024年8月5日" key 年 = year 月 = month (also moon) 日 = day (also sun)
@@AffectionateLocomotive It will be similar to ARM with good power and great power efficiency. ARM is also a RISC processor. Best of all RISC V is open source unlike ARM.
Thx for the video but I have some remarks. 1. the speed in YT is resulting from bad GPU drivers which is a commonly known problem with Imagination GPUs. If you would have used a Radeon RX550 in the PCIe slot, it would have looked much different! 2. a main difference between the K1/M1 chips is, that they support vector. That's why it is crucial to have software (even benchmarks) built with vector support which most aren't (or using version 0.7 of the standard but this one has 1.0!) Other than that it was a nice and informative review.
Oh, didn't even realize the RVV support in K1 was different! Honestly I had to dig through a lot of articles for the differences, Spacemit's own website doesn't have much data on the English version :/
Wishing RISC V the best of luck. I loved my time working on SunOS/Solaris/SPARC. I loved that Sun had published much of the Solaris/SPARC that it could be used in University classes. Hopefully that role could be assumed by RISC V and Linux.
@@Alfred-Neuman This. It probably needs another 5 years. Ebay just offered me a Lenovo PC for $35. It's easily 10 times as powerful as this machine. With a smaller foot print too.
SPARC was multisourced by design, and used in many other places, e.g. ERC32 for space. It also had some high end open source implementations published. MIPS was more popular in architecture classes last I checked, though, and is also used elsewhere (like the PIC32 rebrand). It kind of seems like abandoning either is fad driven, like the adoption of Java.
11:02 that's the character for month. in chinese and japanese dates are written with the characters for year, month, and day (eg 2024年8月1日), so it is kind of like the way we use slashes when writing the date in english, but 8月 on its own is also the chinese name for august, so i think that's probably supposed to be "August 1"
It seems like a new architecture would be a great opportunity to get out from under the absolutely crushing weight of technical debt we're currently trying to live under. Every OS and all the browsers (essentially OSs in themselves) are filled to the brim with a mindboggling amount of cruft. 240p full screen video playback was possible on a 66mhz 486 completely in software mode. A 1ghz machine of any kind should be smashing 1080p, and I'm sure it would if it wasn't running on bloat on top of bloat.
Well... Most likely the Chromium was installed through apt and might have a lot of disabled flags at compile time. Maybe Jeff can try compiling it from sources? 😂
I don't know why, but Chris Barnatt constantly surprises me! He really has an extremely good grip on probable future trends, and the videos he does around that subject are fascinating. Far better than me and I've been in IT for decades. RISC-V is a very intriguing trend, and will hopefully be developed by the open source community
That is why since the Pentium Pro Intel is using RISC cores and just keeping it CISC compatible using a translation layer breaking those instructions down to smaller parts - unsure when AMD made the change. That is why at the same level of performance (seemingly) CISC CPUs are as efficient as RISC (i.e. ARM) CPUs.
I think the thing that I'm possibly most hyped about with Risc V is the SIMD extensions of all things. There's so much code I run on a daily basis which is dependent on them, but retooling it for specific architectures can be a massive pain when you get into bigger projects, so I absolutely love the idea of a sensible, variable bit width instruction set made with a certain degree of hindsight and forward planning built in. I would love to see some coverage of Risc V SIMD!
I think you're spot on with the ARM effect when it comes to building software. By now, RISC-V is just another entry in a list, not a new dimension, in any toolchain that can compile to x86 and ARM. Any developer knows that "support something other than X" is worlds apart from "add Z support, but we already have X and Y support".
I'm wondering what utility Gentoo will have with it's compiling focused package management for risc-v considering it's such an edge case use right now. Maybe compile a working desktop and image it across a group of similarly set up computers?
Chris is very on point about the lack of software compatibility (kernel, drivers) with ARM. Two years later, Rock 5B is still hit and miss with official and community images. Heck, I believe Haiku has better RISC-V support than it does ARM64. My ARM SBCs aren't doing anything atm, because the support just isn't there yet. It's the one thing Raspberry Pi seemingly got right, everyone else has gotten it wrong.
@@Level2Jeff hate to break it to you but that coat of arms is likely just made up by some website. Trying to make money. Coat of arms aren't really used by families and are sparingly used by monarchs or dukes etc.
I watched Explaining Computers video as well, one of my Sunday line up faves of youtube! Will be fantastic to see this developed for main stream use. Great video!
So far the GPU support from Imagination Technologies for the RISC-V chips still has to produce some real results. From that perspective, the GPU support for the Raspberry Pi is mature, and now that Collabora has released the OpenGL driver (Panthor) for the RK3588, ARM is still ahead of RISC-V. And nobody seems to know when Imagination Technologies is going to upstream the GPU drivers.
I keep hearing they will... and then delays... so still holding out some hope, but I don't think most of us would give Imagination the benefit of the doubt!
Isn't Imagination really bad when it comes to GPU drivers for ARM and MIPS Linux? The CPU vastly underperforms and can't even compare to ARM in efficiency. I think that Apple's M-series blow this out of the water in both performance efficiency.
Panthor for Rk3588 is still a dumpster fire, to the point you are still better off running panfork. And no Vulkan yet. And i brought a OPI 5A a year and a half ago... They are far from being a good example. But i do agree about Raspberry PI.
Did not expect to see Chris here. Very friendly person, like to watch his videos too :) I'm interested in both how ARM but also RISC-V progresses, really.
Great presentation! It was interesting to hear Christopher speak at full speed. He usually, and tactfully, speaks more slowly during his videos. Hoping these AIO mobos mature into suitable alternatives for x86. I was thinking ARM might do that, but it hasn't turned out that way yet.
@@shadowarez1337 Those are well supported by current up to date drivers on linux. From his previous videos the older (gcn architecture) AMD cards are what tends to have the best compatability on non-x86 platforms on linux when it comes to dedicated gpus as well - so if any gpu would work that would be one of the more likely ones to work.
@@olnnn thank you then I may have a project I can work on and a use for these treasured GPUs so far 5 have gotten the liquid metal treatment with the conformal coating to help prolong them.
@@hfw3 Wendel for LVL1 actually has ways of doing this with rAMD GPUs in Linux but it's not automatic like it is with Nvidia what I'm wondering is when can we will see wee Hardware encoding as a better option everything I'm seeing in programs shows hardware Encoded/Acceleration is lower quality for some reason is think have hardware would only prove quality not make it worse 🤔
The switch to ARM and RISCV that is concerning me... Its all SOC, none of the chips are socketable, and the manufacturers are betting on that being normalised. Unless theres a pushback now, before these become mainstream, we are sleepwalking into a world where we get no upgradability at all
Nice video Jeff, love Chris and his work as well, really happy to see some movement with RISC-V, Framework should help to accelerate things even further with the upcoming release for a RISC-V upgrade board for their 13" laptop.
A studio thought for ergonomics: Set up a dual screen display for this (10:30) area so that you can show off what you're connecting while also being able to read it well. :)
I am considering adding a little cart with a larger screen I can look at in front of camera so I don't have to do weird non-ergonomic setups just to show the screen and keyboard/mouse at the same time!
Great video, however the talk about RISC-V getting to the desktop faster than ARM is missing some vital historical information. The ARM CPU was originally designed to be a computer CPU and first appeared in the Acorn Archimedes line of desktop PC’s in 1987 running RISC OS. This was 2 years after the ARM CPU was completed.
I think it's important to mention that SpacemiT is a fabless manufacturer which relies on Chinese homegrown fabs to actually manufacture the chips. They don't use high-end fabs in the US or Taiwan, most Chinese fabs are still slowly working towards < 20nm, 14nm is available in limited numbers, so it makes sense that these CPUs aren't going toe to toe with modern ARM CPUs. The Raspi 3 SoC is a decade old ARM chip, but made on TSMC 40nm. It probably makes more sense to compare these RISC-V parts to the ARM offerings from companies like Rockchip that use similar fabs. I'm sure Chinese domestic chips will get a lot better, but I think we'll need to see a RISC-V SoC on a high end TSMC process to really see what its capable of.
“Walk With Me” by TrackTribe, my new favorite groove band now. Thanks for this! I’m tempted to put on The Meters and go back in time with that wooden case. Actually, yes. Yes, I will. Do all yourselves a favor and dig “Look-ka Py Py” by The Meters, swim in the groove my mod, techie friends.
Interesting video thanks Jeff & Chris! Watching this on a DC-ROMA 2 with the SpacemiT K1 CPU and it can handle the video at 360p without stuttering but falls over at 480p. This is with the Ubuntu 23.10 that the laptop shipped with, wonder if the stuttering is actually because the software is not optimized yet? I have a Debian image for the laptop too and am hoping to try other operating systems if you have any recommendations? Also, yes definitely interested in a test of the PCI slot with a GPU (and also with other accelerator or NVMe cards perhaps?).
@@JeffGeerling Thanks I'll put Fedora as the next one on the list when the postman delivers some nice new micro SD cards. I must admit to being quite surprised at how well the machine works...I mean everything "just works" albeit a little slowly at times. I have my heart set on trying to port some kind of BSD or Solaris/Illumos based OS to it...why not run the most obscure OS on the most obscure CPU anyway?
I have been watching RISC-V development and think that it has a lot of potential. It took ARM quite some time to get where it is. I think the future is bright! Thanks for the video.
I mean, powerpc is still being used in the (albeit niche) line of servers being produced by IBM, they sold their x86 branch to lenovo along with thinkpads but their Powerpc servers are still produced inhouse. I imagine porting low level software from one risc architecture to another is probably easier than porting from cisc so this might help in some very specitic cases
@@reghawkins73 It is a common misunderstanding that RISC simply means fewer instructions, despite the coining of that particular acronym, itself the result of needing to label a particular design philosophy in a concise way. What RISC actually entails is more involved, but you can save us the emojis and go and read the early papers by the likes of Patterson if you are really interested in the topic.
I love when two CCs I follow co-op for a video, also nice that this offering gives (what looks like) a pcie x8 slot compared to the x1-4 slots you usually see in these lower power boards.
Nice video fully demonstrating RISC V. Watching the trouble playing video made me remember I have a G5 Mac with 10.5 that plays video off the net as well or better. But if they double its speed each release, those issues will quickly disappear.
Nice to see the Legendary Chris Barnatt on your channel. Noted his observations re the ARM eco system, where Raspberry Pi is the exception to the norm and whilst it could be argued that RPIs arent right up there with the fastest boards, the software and education behind RPIs have revolutionised the whole SBC genre.
IMO if Risc-V matures faster, it's because it benefits from effort previously put into making ARM work. People are now set up to work with multiple ISA, so adding one more is less of an issue.
It'll be years before RISC-V is a viable platform at this rate. Its an order of an magnitude too slow for even basic compute tasks and can't really do everyday tasks at all. Competition is good don't get me wrong but RISC-V is so far from being competition that it might as well not even be considered at all. Sure the software side needs a lot of work but the hardware side seems to need even more work. Risc-V is ten times thats right ten times slower than even an old arm platform like the Pi3 and it'll need another decade of work at this rate. Apple's M-Series of chips is really the way to go right now.
It's not like ARM was the first or even third processor architecture switch. Porting entire systems across processor architectures is at least as old as Multics, the predecessor of Unix (for which it is a main characteristic). Macs have gone from 68k to PPC to x86 to amd64 to ARMv8. Linux traversed a different widening path, x86 first, then 68k and onto all sorts of others (alpha, mips, sparc, ppc, arm, etc), all prepared by GNU.
and less efficient than the current Pi. And lacking in compatibility. Life's a journey, obviously, but "mainstream" this ain't, as they made clear in the video.
@@twistedtxb Primarily about software support and optimisation, I think. I'm not ready to try Risc-V yet, because you're right to say that the performance is poor compared to other architectures currently, but I think that's about adoption and support and not the architecture itself. Hoping for some big performance steps over the next year or so.
@@twistedtxb Well, there is also the 64-core Milk-V Pioneer. I'm hoping we are going to see the Milk-V Oasis (16-core) by the end of this year. The Oasis will also have much faster cores.
Getting Frigate up and running on this thing along with hardware acceleration using that Risc-V GPU sounds like mad scientist stuff, haha Good video btw
I think people need to stop simping for RISC. Especially click-bait titles like this don't help at all. Instead they hurt any chance RISC ever had. Right now it's a toy for a very tiny niche of tech-enthusiasts. It has a long way to go to be relevant at all for the mainstream.
The overall point of this video is to show how RISC-V (collectively) is at a state on the Arm SBC side where Raspberry Pi was around the Pi 2/Pi 3 era, where it went from quirky and horribly slow hobbyist board to 'a thing that actually makes sense in some use cases'. It's certainly not a board I'd see someone install in their gaming PC (but neither is a Pi), but the software and community/soft parts (which most Arm SBC makers lack) is already solid. If the hardware can catch up, with some faster and more efficient RISC-V core designs, Arm is going to have a reliable competitor for things above the microcontroller level.
RISC-V, yes. But RISC architectures have been picking up steam way back when the PowerPC was first introduced, where everything from the IBM servers to the iMacs and PowerBooks, and even the GameCube and the entire 7th generation of consoles, ran on PowerPC. And of course ARM practically revolutionised portable computing with modern smartphones and tablets, and now desktop (including laptops here) and server computing. There were plenty of other RISC architectures other than RISC-V. And even Intel and AMD eventually decided to create modern x86_64 processors that were actually hybrid RISC and CISC processors.
I just don't like the direction this is headed for video encoding. Software encoding is the best method for the highest quality at a given bit rate. Not having instruction sets for advanced math is going to make them slow, so they're pushing hardware encoders. The video we get on our physical media is going to look like ass pretty soon. This push is fine for laptops and desktops, but they need to quit pushing it on all sectors of computing.
I was just searching if a riscv mini itx board exists after watching explaining computers rock 5 itx video. Then your video came up in my feeds. Unbelievable timing.
The designers of the Atari Transputer would be proud! The Atari Transputer was a very early attempt at making a practical RISC computer based on the INMOS Transputer. It was parallel processing and even had a complete MEGA ST acting as an I/O processor. It was amazing tech for the time, but too ahead of it's time as well. And now the RISC V is coming out, and maybe it is time for that long forgotten technology that promised to revolutionize computing in the late 80s early 90s to mature and start a new revolution. I wish you and the designers of the RISC V the best of luck!!
40 Years ago, I worked at a software company and we had to support so many different machines and architectures so i completely sympathise with the Risc V design commitement. The slow board really reminds me of loads of early computers and Inmos Transputers that had so much potential but, hey, software takes time! Oddly today is the 50th Anniversary of CP/M :o)
So here for this, great video. I still remember how much (much!) younger me loved the Acorn Archimedes my long retired headmaster father brought home over summer holidays :-)
Many people say RISC is the future, RISC will change everything... well 30 years ago people also said and thought that as well. Does anyone still remember the Acorn Computer Ltd. (BBC Micro, RiscPC... and so on)? Anyway, just like in the 90s a Reduced Instruction Set Computer (RISC) does make a lot of sense when you need very high efficiency and low manufacturing cost. It doesn't make much sense when you need versatility and PCs (desktops) are very versatile machines when it comes to what people need them for.
Yes, I used Acorn's stuff, including their ARM-based machines that random people will now repeatedly claim affection for, while pumping out the usual abridged narrative about ARM. What you're overlooking is that while ARM became successful in the embedded market, gradually working its way up through mobile to where it is now, that route to success (which wasn't in any way guaranteed, contrary to that abridged narrative) was perhaps the only way ARM would survive in a very competitive market. However, Acorn, like other microcomputer manufacturers, originally wanted to go upmarket and make business machines and workstations, but they didn't pull that off in a timely fashion, and their architectural choices limited how competitive their ARM-based products were. You need to compare the ARM to SPARC, MIPS, PA-RISC, POWER and other RISC architectures to see that ARM - the company - had to move a lot faster than they did to keep up. Digital did give ARM a bit of a performance injection with the StrongARM, but that was too late to save Acorn (and, arguably, Digital for other reasons). But all of the mainstream RISC architectures were quite capable of powering "versatile machines", largely because that is precisely what they were designed for: workstations and personal computers. What then happened was that AMD and Intel adopted various RISC principles, enabling them to keep up in terms of performance and eventually leverage their fabrication, volume and cost advantages, gradually driving traditional RISC products away from mainstream computing. So, contrary to what you are claiming, RISC was the future and did, in fact, change everything.
I've been using my Visionfive 2 with an RX 580 and gentoo for a while now. I have to say, it feels WAAAY snappier than it did before. I'm not quite sure if it's due to compiling everything specifically for my hardware or because a modern graphics card brings a lot of stuff to speed up everyday tasks (video decoders, for example), but I'd definitely give a dedicated GPU a go. I haven't got my Jupiter yet, so I don't know the process to rebuild the kernel with AMDGPU and using mesa is like, but if it's anything like the Visionfive 2, a graphics card really upgrades the experience considerably. TLDR: Give the graphics card a shot.
Thanks for showing your hands on experience with this. We are indeed living in interesting times for open source hardware. Framework just released a video with Nirav Patel explaining their new RISC-V partner-created mainboard. This is such an exciting development! It's still for development, but can be built into their laptops which are consumer focused products. It seems consumer-time is even closer that we thought. Will you try to get your hands on one of those mainboard? Maybe you could try and drag race this motherboard vs the Framework RISC-V mainboard.
Ah interesting. Even though I'm not a systems engineer or kernel developer, Im so excited for this recent development in RISC-V. Happy to see there's open competition in this space. All this healthy competition between system manufacturers and more hype in the media, is why we'll see a consumer ready RISC-V device sooner than we think. This is why I'm subscribed to you Jeff. Thanks for the video 🙌
One thing about Bianbu you really need to run updates. It’s slow out of box. It runs much faster off m.2 and updates. I’ve been working for over a month getting Ubuntu 24.04 on my BPI-F3 and Lichee pi 4A laptop. I want to build everything from the beginning. I would love to work with others trying to accomplish the same tasks.
Nice to see Chris make a guest appearance =) I hope we get to see small and cool boards with at least 2x 2.5GbE ports and more and place for NVME wifi cards.
Nice crossover with Chris, a "Chrisover".
Jeff, did you get to meet Mr. Scissors? I've always wanted to meet Mr. Scissors.
🌚
@@geographicaloddity2 he was out for ice cream together with Stanley the Knife.
A Chris-cross you might say
Found the dad
Seeing you and Chris from EC in one video... my nerd heart could barely take it. As for Risc-V, I can't wait to see what it offers as it matures.
Nice Chrisover episode.
@@danielreed5199 OMG! lol :D
Interesting to see Chris talking more off the cuff.
Yes chris or as i like to call him Older Sheldon
RISC-V going mainstream is the new year of the Linux desktop.
Surely it'll be this year!
Yeah I really don't think so since it is ten times slower than the pi3. Forget about being ready for the mainstream the hardware still needs to make massive leaps in performance before it is even considered ready for the mainstream.
@@fastedwarrior7353maybe for NAS, high end fridge, etc, not as PC
Kinda like with Bitcoin / cryptocurrency too, there needs to be a use case / killer app besides just evangelism to get adoption to the mainstream / next level. Maybe the tighter coupling of open-source custom instructions and the open-source Linux kernel can provide that. Kinda like the hope that RISC-V will be beneficial for the “AI” bubble/craze.
@@seanwieland9763 except crypto/ai is an entirely useless tech that only serves to burn the planet
I did not expect Explaining Computers anywhere outside his channel
I can't believe Chris speaks at a normal pace.
I suppose he's presenting style is just the style for the type of videos he wants to make. His video style reminds me of old/vintage BBC science and technology programmes.
If you wanna continue down this Risc-y rabbit hole, framework also has a risc-v mainboard in development that could be worth looking at
no ty😂😂😂
But does it support the RVA22 and RVV1.0 profiles?
It's actually a third party board from a Chinese company, and Framework is only cooperating with them
@@rexsceleratorum1632 and who do you think makes ther other mainborads... hint it's not framework building 'em in there basement.
I speak a smidge of Mandarin. The "slash character" at 11:08 translates to "month", so it does say August :)
In fact, that character 月 literally means moon and unsurprisingly is used for month as well. And it's used where I live in Korea as well as Japan.
yue. i recodnised the 八 which means 8.
I'm really excited to see you trying to get that GPU working on RISC-V!
fsf would say to switch to risc-v software rendering lmao
@@gamagama69 they're extremist anyway
Eyyy! Explaining Computers cross over! I really like that Chap!
for 11:07, it mostly means month so translates to "8th month"
月 (month)
so for august they would just show that as "8 month"/"8月"
This person right here ❤ High quality comment 👍🏼🔋
to add some info here is how a full date would be shown
English "5th august 2024"
Chinese "2024年8月5日"
key
年 = year
月 = month (also moon)
日 = day (also sun)
8th moon
I'm watching this RISC-V video on my RISC-V system. It's the Banana Pi BPI-F3, running Bianbu, with Chromium. Indeed not fast, but it works.
What does ris. Risc-V specialises in?
Like x86-64 specialises in power and support.
And ARM In efficiency?
@@AffectionateLocomotive Open Source ISA. So anybody is allowed to desgin and make RISC-V CPU's, withou licensing.
@@AffectionateLocomotive It will be similar to ARM with good power and great power efficiency. ARM is also a RISC processor. Best of all RISC V is open source unlike ARM.
@nikhilpatil7218 so, but better?
Thx for the video but I have some remarks.
1. the speed in YT is resulting from bad GPU drivers which is a commonly known problem with Imagination GPUs. If you would have used a Radeon RX550 in the PCIe slot, it would have looked much different!
2. a main difference between the K1/M1 chips is, that they support vector. That's why it is crucial to have software (even benchmarks) built with vector support which most aren't (or using version 0.7 of the standard but this one has 1.0!)
Other than that it was a nice and informative review.
Oh, didn't even realize the RVV support in K1 was different! Honestly I had to dig through a lot of articles for the differences, Spacemit's own website doesn't have much data on the English version :/
@@JeffGeerling no problem....community is here for you :)
Chris Barnatt and Jeff in the same video! Life is good!!!
Wishing RISC V the best of luck. I loved my time working on SunOS/Solaris/SPARC. I loved that Sun had published much of the Solaris/SPARC that it could be used in University classes. Hopefully that role could be assumed by RISC V and Linux.
Open Solaris lives on in the illumos project. There is even a port to at least one RICS-V board.
It's struggling to play a RUclips video at 360p?
I like the concept but I can't really say I am very impressed by this thing...
@@Alfred-Neuman This. It probably needs another 5 years. Ebay just offered me a Lenovo PC for $35. It's easily 10 times as powerful as this machine. With a smaller foot print too.
SPARC was multisourced by design, and used in many other places, e.g. ERC32 for space. It also had some high end open source implementations published. MIPS was more popular in architecture classes last I checked, though, and is also used elsewhere (like the PIC32 rebrand). It kind of seems like abandoning either is fad driven, like the adoption of Java.
Jeff Gerling AND Explaning Computers in one video... Cool :-)
You tease! Of course we want to see GPU support on riscV!!!
Yes Indeed even though its going to tantamount to hot garbage due to lack of cpu performance and drivers.
Can it run Crysis? 😂
11:02 that's the character for month. in chinese and japanese dates are written with the characters for year, month, and day (eg 2024年8月1日), so it is kind of like the way we use slashes when writing the date in english, but 8月 on its own is also the chinese name for august, so i think that's probably supposed to be "August 1"
Yes, you're right.
(It's actually quite strange though to say "8月1" instead of "8月1日". That is actually a crude translation from English.)
That's awesome you had Explaining Computers on.
It seems like a new architecture would be a great opportunity to get out from under the absolutely crushing weight of technical debt we're currently trying to live under. Every OS and all the browsers (essentially OSs in themselves) are filled to the brim with a mindboggling amount of cruft. 240p full screen video playback was possible on a 66mhz 486 completely in software mode. A 1ghz machine of any kind should be smashing 1080p, and I'm sure it would if it wasn't running on bloat on top of bloat.
It’s probably a driver issue. The board Chris reviewed was able to play 1080p video.
Now imagine making arcade games by Sega and others on it. Consoles too.
Well... Most likely the Chromium was installed through apt and might have a lot of disabled flags at compile time. Maybe Jeff can try compiling it from sources? 😂
I don't know why, but Chris Barnatt constantly surprises me! He really has an extremely good grip on probable future trends, and the videos he does around that subject are fascinating. Far better than me and I've been in IT for decades.
RISC-V is a very intriguing trend, and will hopefully be developed by the open source community
"RISC architecture is going to change everything" - Acid Burn
RISC is good
That is why since the Pentium Pro Intel is using RISC cores and just keeping it CISC compatible using a translation layer breaking those instructions down to smaller parts - unsure when AMD made the change.
That is why at the same level of performance (seemingly) CISC CPUs are as efficient as RISC (i.e. ARM) CPUs.
Technicolored rainbow
@@belzebub16 that's how all cpus work
Hack the planet.
I think the thing that I'm possibly most hyped about with Risc V is the SIMD extensions of all things. There's so much code I run on a daily basis which is dependent on them, but retooling it for specific architectures can be a massive pain when you get into bigger projects, so I absolutely love the idea of a sensible, variable bit width instruction set made with a certain degree of hindsight and forward planning built in.
I would love to see some coverage of Risc V SIMD!
It's so trippy to hear Explaning Computers to speak off script!!! Such a cool dude though. :-) :-) :-)
It's nice seeing Explaining Computers on your channel! They are such an amazing resource for stuff like this
Chris from EC? Can't wait for the Mr. Scissors cameo!
or Stanley the Knife...
I was waiting for "And that (pause) is it!!" 😢
I think you're spot on with the ARM effect when it comes to building software. By now, RISC-V is just another entry in a list, not a new dimension, in any toolchain that can compile to x86 and ARM. Any developer knows that "support something other than X" is worlds apart from "add Z support, but we already have X and Y support".
I remember how hard it was to get x86 + Arm with so many toolchains 5-10 years ago. So much simpler now.
@@JeffGeerling good thing you've not been around when we went fromm x86 to AMD64 that was a pain and a half.
I'm wondering what utility Gentoo will have with it's compiling focused package management for risc-v considering it's such an edge case use right now. Maybe compile a working desktop and image it across a group of similarly set up computers?
Nice to see you tying in with Chris...
Chris is very on point about the lack of software compatibility (kernel, drivers) with ARM. Two years later, Rock 5B is still hit and miss with official and community images. Heck, I believe Haiku has better RISC-V support than it does ARM64. My ARM SBCs aren't doing anything atm, because the support just isn't there yet. It's the one thing Raspberry Pi seemingly got right, everyone else has gotten it wrong.
I love your Dutch soccer shirt!
Hup Holland Hup!
@@JeffGeerling You have Dutch family? Or is it a viewers present? Geerling is a Dutch surname, but you likely know that by now :)
With his surname, I'm certain he has some family in NL or something ?
@@autohmae My personal website also has my family's coat of arms at the bottom of every page. We are quite happy with our Dutch heritage!
@@Level2Jeff hate to break it to you but that coat of arms is likely just made up by some website. Trying to make money. Coat of arms aren't really used by families and are sparingly used by monarchs or dukes etc.
This has been a long time coming! I'm going to have to get my hands on one of these and put it through its paces 😁
I watched Explaining Computers video as well, one of my Sunday line up faves of youtube! Will be fantastic to see this developed for main stream use. Great video!
Super cool to see you and Chris from EC discussing technology. I enjoy his videos equally as much as yours!
So far the GPU support from Imagination Technologies for the RISC-V chips still has to produce some real results. From that perspective, the GPU support for the Raspberry Pi is mature, and now that Collabora has released the OpenGL driver (Panthor) for the RK3588, ARM is still ahead of RISC-V. And nobody seems to know when Imagination Technologies is going to upstream the GPU drivers.
I keep hearing they will... and then delays... so still holding out some hope, but I don't think most of us would give Imagination the benefit of the doubt!
Back when androids came with gpus from Imagination Technologies, they did the same stuff - not releasing proper drivers/blobs.
Isn't Imagination really bad when it comes to GPU drivers for ARM and MIPS Linux? The CPU vastly underperforms and can't even compare to ARM in efficiency. I think that Apple's M-series blow this out of the water in both performance efficiency.
It's already open and mainstream.
Panthor for Rk3588 is still a dumpster fire, to the point you are still better off running panfork. And no Vulkan yet. And i brought a OPI 5A a year and a half ago... They are far from being a good example.
But i do agree about Raspberry PI.
Those cases with wood are really nice.
Cool interview with Chris. Been following him for a long time too
Yes! We must have the graphics card video. I suspect it will not go well, but at least it is a baseline for future comparisons.
Did not expect to see Chris here.
Very friendly person, like to watch his videos too :)
I'm interested in both how ARM but also RISC-V progresses, really.
Good timing with Chris's Rock5 itx desktop build video!
It was a lucky coincidence!
Great presentation! It was interesting to hear Christopher speak at full speed. He usually, and tactfully, speaks more slowly during his videos. Hoping these AIO mobos mature into suitable alternatives for x86. I was thinking ARM might do that, but it hasn't turned out that way yet.
The Jeff Geerling channel is changing too. New studio. Higher production quality. Haven't seen Red Shirt Jeff in ages.
Business insurance mandates he limits time in studio :D
He stuck back in for an Intel-related video though :O
@@JeffGeerlingheh, poor Intel, today's favorite punching bag for everyone on the internet :)
Can these use Android TV? Could be the hardware to iron Mike punch Nvidia outa the tv console market.
@@shadowarez1337Google is porting Android to RISC-V.
The cross over we never knew we wanted.
Two of my favorite tech channels on this platform teaming together? Awesome!
I am so excited on this. Framework recently has a RISC-V one too. Maybe we would see it catch up to RPi 5 levels of performance in a few years.
I want a Framework Phone with a different OS than Android and iOS with the language apps to get off their asses to support it.
Jeff and Chris in the same video, that's awesome!
Thank you.
If developer's are the target then space MIT should publish register level info. So far, their spec sheet is a bullet list of features.
Of course you have to put the gpu in it!
Wonder if AMD would have compatibility drivers one day for the Fury X Nano 🤣picked up a bulk of 10 for $60 a piece.
@@shadowarez1337
Those are well supported by current up to date drivers on linux.
From his previous videos the older (gcn architecture) AMD cards are what tends to have the best compatability on non-x86 platforms on linux when it comes to dedicated gpus as well - so if any gpu would work that would be one of the more likely ones to work.
@@olnnn thank you then I may have a project I can work on and a use for these treasured GPUs so far 5 have gotten the liquid metal treatment with the conformal coating to help prolong them.
So… how far away from Plex running on Linux with transcoding support using a discrete graphics card are we?
@@hfw3 Wendel for LVL1 actually has ways of doing this with rAMD GPUs in Linux but it's not automatic like it is with Nvidia what I'm wondering is when can we will see wee Hardware encoding as a better option everything I'm seeing in programs shows hardware Encoded/Acceleration is lower quality for some reason is think have hardware would only prove quality not make it worse 🤔
I'm thrilled to see RISC-V making progress and BONUS having Jeff and Christopher in the same video. It's a 2-fer!
RISC-V may no longer be a RISK to adopt... I'll walk myself out
That was a Risky joke. I'm glad it only took 5 attempts to get right 😂
I will yeet myself out. 😅
Get outta here, Boooo, Jk
Don't beat yourself up, someone had to say it!
But question is: will it play RISC?
It's still a RISK. No socketed RAM means when a chip dies, the whole board goes in the garbage can, not happening.
I love that you did a collab with Chris! I love Explaining Computers
The switch to ARM and RISCV that is concerning me... Its all SOC, none of the chips are socketable, and the manufacturers are betting on that being normalised. Unless theres a pushback now, before these become mainstream, we are sleepwalking into a world where we get no upgradability at all
The software walled gardens are getting pretty green so you know what’s next
Nice to see explaining computers! These boards are exciting to see. If you could try pcie cards in the board that would be great!
Your videos always drop at midnight here - and they always remind me that I should be asleep. Damnit.
Like clockwork!
Nice video Jeff, love Chris and his work as well, really happy to see some movement with RISC-V, Framework should help to accelerate things even further with the upcoming release for a RISC-V upgrade board for their 13" laptop.
A studio thought for ergonomics: Set up a dual screen display for this (10:30) area so that you can show off what you're connecting while also being able to read it well. :)
I am considering adding a little cart with a larger screen I can look at in front of camera so I don't have to do weird non-ergonomic setups just to show the screen and keyboard/mouse at the same time!
@@JeffGeerling you could even make a DIY prompter, so that you can look at the camera at the same time.
That is like the most beautiful PC case I've seen. Bet it would look really nice with that like 1960's speaker covering kinda thing over the front too
Fractal also makes a few cases with fabric-they do have a nice aesthetic!
That pc case is like a 1970s speaker- are we coming back around??
There were some great styles, and some terrible styles, that came from the 70s :)
@@JeffGeerling what model of PC case is this ? looks like a Fractal Design or something stylish like that
@@dju1999 Yes, it's the Fractal Design North case
Thx for keeping us updated on where risc V is at! :)
Great video, however the talk about RISC-V getting to the desktop faster than ARM is missing some vital historical information.
The ARM CPU was originally designed to be a computer CPU and first appeared in the Acorn Archimedes line of desktop PC’s in 1987 running RISC OS. This was 2 years after the ARM CPU was completed.
ARM will always be Acorn Risc Machine
Chris and Jeff in one video! This is peak RUclips, folks. It can't get any better.
15:13 - yes
It's really interesting seeing the similarities/differences between this and Explaining Computer's recent video, too!
It was very lucky that we both were working on an ITX build in the same week!
@@JeffGeerling It happened by chance! :) A happy co-incidence.
Nice vid! Love Chris, he's such a font of knowledge and pragmatism in the techtuber space. Have him on more if he wants!
Two of my favorite channels in one video.
"going mainstream" seems a bit optimistic when this thing gets handily thrashed by my dual core Sandy Bridge ThinkPad from the Obama administration...
....kids
And my 2009 Gateway DX4831-01e with a Core i3-530
You don't need to run GTA-5 on everything
My former gov Thinkpad dual core i3 2nd gen pimps this shit.
I'm still using my Sandy bridge workstation. That quad channel memory is bloody nice trick its a shame they stopped using.
I think it's important to mention that SpacemiT is a fabless manufacturer which relies on Chinese homegrown fabs to actually manufacture the chips. They don't use high-end fabs in the US or Taiwan, most Chinese fabs are still slowly working towards < 20nm, 14nm is available in limited numbers, so it makes sense that these CPUs aren't going toe to toe with modern ARM CPUs. The Raspi 3 SoC is a decade old ARM chip, but made on TSMC 40nm. It probably makes more sense to compare these RISC-V parts to the ARM offerings from companies like Rockchip that use similar fabs. I'm sure Chinese domestic chips will get a lot better, but I think we'll need to see a RISC-V SoC on a high end TSMC process to really see what its capable of.
7:40 Now this is computer-building music!
“Walk With Me” by TrackTribe, my new favorite groove band now. Thanks for this!
I’m tempted to put on The Meters and go back in time with that wooden case. Actually, yes. Yes, I will.
Do all yourselves a favor and dig “Look-ka Py Py” by The Meters, swim in the groove my mod, techie friends.
Happy to see you both with Chris !! I follow you both guys 😊😊
Interesting video thanks Jeff & Chris!
Watching this on a DC-ROMA 2 with the SpacemiT K1 CPU and it can handle the video at 360p without stuttering but falls over at 480p. This is with the Ubuntu 23.10 that the laptop shipped with, wonder if the stuttering is actually because the software is not optimized yet? I have a Debian image for the laptop too and am hoping to try other operating systems if you have any recommendations? Also, yes definitely interested in a test of the PCI slot with a GPU (and also with other accelerator or NVMe cards perhaps?).
Could be, and Fedora seems to be okay as well (Fedora V-force images for Fedora 41). Definitely worth testing!
@@JeffGeerling Thanks I'll put Fedora as the next one on the list when the postman delivers some nice new micro SD cards. I must admit to being quite surprised at how well the machine works...I mean everything "just works" albeit a little slowly at times. I have my heart set on trying to port some kind of BSD or Solaris/Illumos based OS to it...why not run the most obscure OS on the most obscure CPU anyway?
@@glenwalker4093 Then you should port TempleOS 😂
I have been watching RISC-V development and think that it has a lot of potential. It took ARM quite some time to get where it is. I think the future is bright! Thanks for the video.
RISC V is a cool and very important project. But RISC V as a desktop or server CPU is a meme and will be for a while yet unfortunately .
But it's the year of the Linux Desktop!? :D
@@JeffGeerling :D
I mean, powerpc is still being used in the (albeit niche) line of servers being produced by IBM, they sold their x86 branch to lenovo along with thinkpads but their Powerpc servers are still produced inhouse. I imagine porting low level software from one risc architecture to another is probably easier than porting from cisc so this might help in some very specitic cases
The collaboration we've all been waiting for. Now, try it out as a NAS and through Colten into the game. =)
PowerPC was running a RISC desktop before WIndows had sound in them "multimedia" monitors
Arm & PowerPC processors use risc architecture your mixing up architecture and instruction sets😂
@@reghawkins73He isn’t… RISCV is a RISC isa, but they are not the same.
@@xanderplayz3446 I know there not the same that's what I said 😂
Arm risc ISA , powerPC Ric's ISA have different assembly commands, risc ISA just means less instructions on the CPU 😂
@@reghawkins73 It is a common misunderstanding that RISC simply means fewer instructions, despite the coining of that particular acronym, itself the result of needing to label a particular design philosophy in a concise way. What RISC actually entails is more involved, but you can save us the emojis and go and read the early papers by the likes of Patterson if you are really interested in the topic.
I love when two CCs I follow co-op for a video, also nice that this offering gives (what looks like) a pcie x8 slot compared to the x1-4 slots you usually see in these lower power boards.
Nice T-Shirt, hup Holland hup laat de leeuw niet in zijn hempie staan
Nice video fully demonstrating RISC V.
Watching the trouble playing video made me remember I have a G5 Mac with 10.5 that plays video off the net as well or better.
But if they double its speed each release, those issues will quickly disappear.
"RISC, RISC Architecture is going to change everything. Yes, RISC is good "- Acid Burn and Crash override, Hackers the Movie 1995 😂
Nice to see the Legendary Chris Barnatt on your channel. Noted his observations re the ARM eco system, where Raspberry Pi is the exception to the norm and whilst it could be argued that RPIs arent right up there with the fastest boards, the software and education behind RPIs have revolutionised the whole SBC genre.
IMO if Risc-V matures faster, it's because it benefits from effort previously put into making ARM work. People are now set up to work with multiple ISA, so adding one more is less of an issue.
It'll be years before RISC-V is a viable platform at this rate. Its an order of an magnitude too slow for even basic compute tasks and can't really do everyday tasks at all. Competition is good don't get me wrong but RISC-V is so far from being competition that it might as well not even be considered at all. Sure the software side needs a lot of work but the hardware side seems to need even more work. Risc-V is ten times thats right ten times slower than even an old arm platform like the Pi3 and it'll need another decade of work at this rate. Apple's M-Series of chips is really the way to go right now.
It's not like ARM was the first or even third processor architecture switch. Porting entire systems across processor architectures is at least as old as Multics, the predecessor of Unix (for which it is a main characteristic). Macs have gone from 68k to PPC to x86 to amd64 to ARMv8. Linux traversed a different widening path, x86 first, then 68k and onto all sorts of others (alpha, mips, sparc, ppc, arm, etc), all prepared by GNU.
Chris is my fav, next to Jeff's Dad. Good to see him here.
The burning question with this is, "can it run Crysis?" 🤣
Heh so far definitely no!
Love it Jeff! This all makes me want to get off the couch, pull up my sleeves, and do some development. 😀
Yeah, slower then a raspberry pi 3 line puts things in to perspective.
and less efficient than the current Pi. And lacking in compatibility. Life's a journey, obviously, but "mainstream" this ain't, as they made clear in the video.
I'm quite shocked how far behind it's still is vs contemporary architecture
@@twistedtxb Primarily about software support and optimisation, I think. I'm not ready to try Risc-V yet, because you're right to say that the performance is poor compared to other architectures currently, but I think that's about adoption and support and not the architecture itself. Hoping for some big performance steps over the next year or so.
@@twistedtxb Well, there is also the 64-core Milk-V Pioneer. I'm hoping we are going to see the Milk-V Oasis (16-core) by the end of this year. The Oasis will also have much faster cores.
So can anyone tell me what the advantage of RISC-V SoC if it can't even come close to ARM or X86 in performance and/or efficiency?
Happy to see 2 of my favorite RUclipsrs collaborating
I still don't understand why would anyone buy risc-v system. There is no upside other than being enthusiastic about it.
Getting Frigate up and running on this thing along with hardware acceleration using that Risc-V GPU sounds like mad scientist stuff, haha
Good video btw
I'm still not used to PC cases looking like dehumidifiers with no drive bays....
Are you used to the beige ones or the RGB versions with the need of sunglasses? 🤔😁
@@JohnADoe-pg1qk heheh more like on wheels with 8 5.25" drive bays ...
I just built a pc with that exact same case! It’s beautiful on my matching wood floor
I think people need to stop simping for RISC. Especially click-bait titles like this don't help at all. Instead they hurt any chance RISC ever had. Right now it's a toy for a very tiny niche of tech-enthusiasts. It has a long way to go to be relevant at all for the mainstream.
The overall point of this video is to show how RISC-V (collectively) is at a state on the Arm SBC side where Raspberry Pi was around the Pi 2/Pi 3 era, where it went from quirky and horribly slow hobbyist board to 'a thing that actually makes sense in some use cases'.
It's certainly not a board I'd see someone install in their gaming PC (but neither is a Pi), but the software and community/soft parts (which most Arm SBC makers lack) is already solid. If the hardware can catch up, with some faster and more efficient RISC-V core designs, Arm is going to have a reliable competitor for things above the microcontroller level.
RISC-V, yes. But RISC architectures have been picking up steam way back when the PowerPC was first introduced, where everything from the IBM servers to the iMacs and PowerBooks, and even the GameCube and the entire 7th generation of consoles, ran on PowerPC.
And of course ARM practically revolutionised portable computing with modern smartphones and tablets, and now desktop (including laptops here) and server computing.
There were plenty of other RISC architectures other than RISC-V. And even Intel and AMD eventually decided to create modern x86_64 processors that were actually hybrid RISC and CISC processors.
15 years ago I saw many of the same comments about ARM.
How can I dislike a comment
I just don't like the direction this is headed for video encoding. Software encoding is the best method for the highest quality at a given bit rate. Not having instruction sets for advanced math is going to make them slow, so they're pushing hardware encoders. The video we get on our physical media is going to look like ass pretty soon. This push is fine for laptops and desktops, but they need to quit pushing it on all sectors of computing.
I was just searching if a riscv mini itx board exists after watching explaining computers rock 5 itx video. Then your video came up in my feeds. Unbelievable timing.
The designers of the Atari Transputer would be proud! The Atari Transputer was a very early attempt at making a practical RISC computer based on the INMOS Transputer. It was parallel processing and even had a complete MEGA ST acting as an I/O processor. It was amazing tech for the time, but too ahead of it's time as well. And now the RISC V is coming out, and maybe it is time for that long forgotten technology that promised to revolutionize computing in the late 80s early 90s to mature and start a new revolution. I wish you and the designers of the RISC V the best of luck!!
40 Years ago, I worked at a software company and we had to support so many different machines and architectures so i completely sympathise with the Risc V design commitement.
The slow board really reminds me of loads of early computers and Inmos Transputers that had so much potential but, hey, software takes time!
Oddly today is the 50th Anniversary of CP/M :o)
So here for this, great video. I still remember how much (much!) younger me loved the Acorn Archimedes my long retired headmaster father brought home over summer holidays :-)
If there is any additional interview footage on the geopolitical discussion I would love to see it!
Many people say RISC is the future, RISC will change everything... well 30 years ago people also said and thought that as well. Does anyone still remember the Acorn Computer Ltd. (BBC Micro, RiscPC... and so on)? Anyway, just like in the 90s a Reduced Instruction Set Computer (RISC) does make a lot of sense when you need very high efficiency and low manufacturing cost. It doesn't make much sense when you need versatility and PCs (desktops) are very versatile machines when it comes to what people need them for.
Yes, I used Acorn's stuff, including their ARM-based machines that random people will now repeatedly claim affection for, while pumping out the usual abridged narrative about ARM. What you're overlooking is that while ARM became successful in the embedded market, gradually working its way up through mobile to where it is now, that route to success (which wasn't in any way guaranteed, contrary to that abridged narrative) was perhaps the only way ARM would survive in a very competitive market.
However, Acorn, like other microcomputer manufacturers, originally wanted to go upmarket and make business machines and workstations, but they didn't pull that off in a timely fashion, and their architectural choices limited how competitive their ARM-based products were. You need to compare the ARM to SPARC, MIPS, PA-RISC, POWER and other RISC architectures to see that ARM - the company - had to move a lot faster than they did to keep up. Digital did give ARM a bit of a performance injection with the StrongARM, but that was too late to save Acorn (and, arguably, Digital for other reasons).
But all of the mainstream RISC architectures were quite capable of powering "versatile machines", largely because that is precisely what they were designed for: workstations and personal computers. What then happened was that AMD and Intel adopted various RISC principles, enabling them to keep up in terms of performance and eventually leverage their fabrication, volume and cost advantages, gradually driving traditional RISC products away from mainstream computing. So, contrary to what you are claiming, RISC was the future and did, in fact, change everything.
The "TO VICTORY" Dalek poster is Fantastic!
I've been using my Visionfive 2 with an RX 580 and gentoo for a while now. I have to say, it feels WAAAY snappier than it did before. I'm not quite sure if it's due to compiling everything specifically for my hardware or because a modern graphics card brings a lot of stuff to speed up everyday tasks (video decoders, for example), but I'd definitely give a dedicated GPU a go. I haven't got my Jupiter yet, so I don't know the process to rebuild the kernel with AMDGPU and using mesa is like, but if it's anything like the Visionfive 2, a graphics card really upgrades the experience considerably.
TLDR: Give the graphics card a shot.
Thanks for showing your hands on experience with this. We are indeed living in interesting times for open source hardware. Framework just released a video with Nirav Patel explaining their new RISC-V partner-created mainboard. This is such an exciting development! It's still for development, but can be built into their laptops which are consumer focused products. It seems consumer-time is even closer that we thought.
Will you try to get your hands on one of those mainboard? Maybe you could try and drag race this motherboard vs the Framework RISC-V mainboard.
It looks like that will be based on the JH7110, which is a bit slower than the Spacemit M1/K1, but it's still an okay chip for devs.
Ah interesting. Even though I'm not a systems engineer or kernel developer, Im so excited for this recent development in RISC-V. Happy to see there's open competition in this space. All this healthy competition between system manufacturers and more hype in the media, is why we'll see a consumer ready RISC-V device sooner than we think. This is why I'm subscribed to you Jeff. Thanks for the video 🙌
One thing about Bianbu you really need to run updates. It’s slow out of box. It runs much faster off m.2 and updates. I’ve been working for over a month getting Ubuntu 24.04 on my BPI-F3 and Lichee pi 4A laptop. I want to build everything from the beginning. I would love to work with others trying to accomplish the same tasks.
Nice to see Chris make a guest appearance =) I hope we get to see small and cool boards with at least 2x 2.5GbE ports and more and place for NVME wifi cards.
Jesus showed up? :)
@@followthetrawler typo, or was it.. 😁😂
Yeaah!! Milk-V! I am so much looking forward to the Milk-V Oasis. I will use it for programming instead of my x86_64 pc.
Build montages are really the most undervalued tool in the engineer's toolbox.
I can really see a bright future ahead for RISC-V. :)
Also, CHRISTOPHER BARNATT, MY BELOVED!!! 🥺❤