This CPU is FREE! - Milk-V Pioneer with RISC-V

  Переглядів 1,225,952

Linus Tech Tips

Linus Tech Tips

День тому

Your business deserves a website! Create one for free at odoo.com/LTT
Who doesn’t like a free CPU? The answer is a bit more complicated than you’re imagining. This computer might look small and unassuming, but inside it is a motherboard you’ve never seen before, a 64-core processor unlike anything you currently own, and the technology behind it could have far reaching geopolitical implications in the very near future. It is the subject of trade sanctions, stock dumps, and could lead us into the next technology-fueled cold war. But can it game?
Discuss on the forum: linustechtips.com/topic/15456...
Check out the Milk-V Pioneer: milkv.io/pioneer
Purchases made through some store links may provide some compensation to Linus Media Group.
► GET MERCH: lttstore.com
► GET EXCLUSIVE CONTENT ON FLOATPLANE: lmg.gg/lttfloatplane
► SPONSORS, AFFILIATES, AND PARTNERS: lmg.gg/partners
► EQUIPMENT WE USE TO FILM LTT: lmg.gg/LTTEquipment
► OUR WAN PODCAST GEAR: lmg.gg/wanset
FOLLOW US
---------------------------------------------------
Twitter: / linustech
Facebook: / linustech
Instagram: / linustech
TikTok: / linustech
Twitch: / linustech
MUSIC CREDIT
---------------------------------------------------
Intro: Laszlo - Supernova
Video Link: • [Electro] - Laszlo - S...
iTunes Download Link: itunes.apple.com/us/album/sup...
Artist Link: / laszlomusic
Outro: Approaching Nirvana - Sugar High
Video Link: • Sugar High - Approachi...
Listen on Spotify: spoti.fi/UxWkUw
Artist Link: / approachingnirvana
Intro animation by MBarek Abdelwassaa / mbarek_abdel
Monitor And Keyboard by vadimmihalkevich / CC BY 4.0 geni.us/PgGWp
Mechanical RGB Keyboard by BigBrotherECE / CC BY 4.0 geni.us/mj6pHk4
Mouse Gamer free Model By Oscar Creativo / CC BY 4.0 geni.us/Ps3XfE
CHAPTERS
---------------------------------------------------
0:00 Intro
1:31 What is this thing?
2:37 What is an Instruction Set Architecture?
3:39 Using the Milk-V Pioneer
4:56 RISC vs CISC
5:55 ARM and RISC-V
6:28 Steam with box64
7:25 Why not just use ARM?
7:58 Looking at the hardware
9:52 Who is behind this CPU?
10:41 US vs China
12:03 What does it mean for gamers?
13:04 ARM's Race
14:49 Euro Truck Simulator 2
15:56 The Future/Conclusions

КОМЕНТАРІ: 2 000
@LinusTechTips
@LinusTechTips 5 місяців тому
At 16:15 we refer to Imagination Technologies as a "GPU Startup". With a legacy dating back to 1985, Imagination Technologies is the brains behind the PowerVR GPU series among other achievements. Apologies for the mix-up in the video; they're definitely more than just a "GPU Startup."
@ThePortuguesePlayer
@ThePortuguesePlayer 5 місяців тому
They might as well be a startup, I'm still waiting for Linux drivers for their PowerVR SGX545 GPU since 2012! I need drivers to use the thing, man, come on!
@PrivateSi
@PrivateSi 5 місяців тому
It's RISC 'V' as in VEE for VECTOR as half the point of Risc VEE was to replace the bloated SIMD ISAs that forever balloon the number of instructions for each fixed sized vector it deals with. RISC VEE spec uses Variable length Vector Processing... Only those that don't research RISC VEE properly don't know this basic, fundamental platform fact.. ARM is Risc V's main target. Intel would rather sell Risc V than pay money to ARM when it finally kills X86 in the next 5 years. No way will it survive much past that fore new kit.
@aladdin8623
@aladdin8623 5 місяців тому
IT started to support open source drivers some time ago and recently contributed code as such to the kernel. Floss is the way to make products more sustainable. They seem to have learned their lesson.
@jspringer86
@jspringer86 5 місяців тому
Just downloaded an adblocker and came here to see if it works. It does. Made my YT account in '06 and just finally snapped after youtube is going crazy with ads recently. Hour long ads and the YT homepage now has x4 videos of the x20. Looked up which adblocker to use from Rossman and saw his video on you from a year ago. Cheers!
@ThePortuguesePlayer
@ThePortuguesePlayer 5 місяців тому
@@aladdin8623 I know for a fact that the drivers exist because there was a proprietary Linux based OS that had them. I don't remember what it was called. The problem is that those drivers are not in any of the modern OSes yet, even including Windows. This GPU released with an installer for its drivers made for Windows 7 32-bit and it won't work on literally anything else. Not Windows 8, not Windows XP, not even the 64-bit version of that same Windows 7, not any version or release of Linux. It's one of the biggest scams I ever fell in.
@yoshiman3692
@yoshiman3692 5 місяців тому
Its so strange to see real cpus running RISC-V, because being in university, and in a cpu design class we used Verilog HDL and FPGA's to implement the RISC-V instruction set from the ground up. Not saying it was easy though, so I have a huge respect for these guys (Chinese or American), but Verilog did a lot of the heavy lifting.
@terinrichardson6061
@terinrichardson6061 5 місяців тому
Literally just got done with a digital design class where I learned about it. This would not be a fun project lol.
@anurag9314
@anurag9314 5 місяців тому
@@terinrichardson6061 haha same. Got an exam about RISC-V architecture in a few days, needless to say, it's hard
@matthewbass8152
@matthewbass8152 5 місяців тому
Lol same we did this at Colby
@trs5127
@trs5127 5 місяців тому
To be fair, implementing it now that it already exists, and writing the instructions from scratch are obviously different things. But yeah, since it's RISC based, that means there are going to be few operating commands that you can write all instructions with. So it makes sense.
@conorstewart2214
@conorstewart2214 5 місяців тому
@@anurag9314 that doesn’t sound like a very useful exam or class. Rather than teaching you about RISC-V they should be teaching you about processor construction and architecture in general.
@yak27_0
@yak27_0 5 місяців тому
I'm studing Computer Engineering at a public polytechnic university and they recently switched to RISC-V as the primary real-world academic example for many courses that discuss ISAs, assembly, and computer acutecture. I watched this video with a SiFive dev board sitting on my desk.
@peterlin5857
@peterlin5857 5 місяців тому
Definitely good to see the knowledge learnt from school is applicable in real-world, instead of some proprietary or out-of-date things.
@kayakMike1000
@kayakMike1000 5 місяців тому
It's a great little ISA.
@russsobti
@russsobti 5 місяців тому
Cal poly?
@rafradeki
@rafradeki 5 місяців тому
@@peterlin5857 how about propertiary AND out-of-date? Becase thats how often it is
@nickiascerinschi206
@nickiascerinschi206 5 місяців тому
Lets goo i am a dev willing to write software for risc v
@personinousapraham3082
@personinousapraham3082 5 місяців тому
Student at UC Berkeley here! RISC-V was created by researchers here, and it's made our way down to the introductory curriculum. The final project of our lower division (required introductory) computer architecture course involves making a slightly simplified RISC-V processor out of basic logic gates! Super cool experience
@darsparx
@darsparx 5 місяців тому
That sounds so cool even if my overthinking would prevent me from finishing that 😂
@Erevos85
@Erevos85 4 місяці тому
20 years ago, one of my professors at university was one of the researchers of the original Berkeley RISC.
@rafaelflorez3880
@rafaelflorez3880 4 місяці тому
Can second this CS61C at UC Berkeley is a great class!
@Zarqus99
@Zarqus99 4 місяці тому
our intro class to computer architecture (EECS 112) at UC Irvine uses RISC-V as well. Next quarter I'm going to make a simplified processor as well!
@HungNguyen-to7dg
@HungNguyen-to7dg 4 місяці тому
We do RISC V here at HCMUT Vietnam too
@fedora
@fedora 5 місяців тому
We're glad to see that all the work our RISC-V team has been doing is showing, great video 💙
@mkunz-3548
@mkunz-3548 5 місяців тому
Fedora is awesome. Recently installed the asahi remix which is probably one of the best distros I have ever used
@harshsomvanshi8963
@harshsomvanshi8963 5 місяців тому
I use Arch, btw
@TemplePate01
@TemplePate01 5 місяців тому
@fedora awesome job everyone! 🎉
@pakxo.
@pakxo. 5 місяців тому
ok redhat
@TemplePate01
@TemplePate01 5 місяців тому
@@pakxo. Actually I'm a basic PopOS guy x3 but I'm all for every distro working to improve cause it just means more options if someone pulls an Ubuntu.
@h1Lu
@h1Lu 5 місяців тому
No, nothing is free. Except this fabulous segue to our sponsor!
@bmacd72
@bmacd72 5 місяців тому
Stop giving linus ideas lmao
@h1Lu
@h1Lu 5 місяців тому
@@bmacd72 why would I?
@RiverHolloway-wm5jf
@RiverHolloway-wm5jf 5 місяців тому
This is something he WILL say. Lol.
@smartduck904
@smartduck904 5 місяців тому
Skillshare
@h1Lu
@h1Lu 5 місяців тому
@@smartduck904 Skillshare is not free
@davidgoodnow269
@davidgoodnow269 5 місяців тому
*NOTE:* The same Linux kernel 6.2 that introduced RISC V support ALSO introduced *FULL SUPPORT* for Intel ARC video cards -- so yes, *THAT* is your first choice for video card!
@SianaGearz
@SianaGearz 4 місяці тому
There's any number of relatively powerful GPUs that have full Linux drivers, much nicer than the 500 series. There are two problems, one is x86 specific intrinsics in the driver, but that's a problem that presents itself clearly and is easy to solve. The other is that GPUs require extensive support for memory mapping from the PCI host chipset and the driver, which is specifically the problem why none of the nicer GPUs work on a Raspberry Pi or this for example. Solving this requires an INSANE level of insight, which is why it's been so slow to get tackled. I surmise Intel ARC is going to prove particularly quirky, given how much dependence on Resizable BAR it has shown. For the time being, GCN4 chips are the only ones that got any headway in solving this at all. The low performance also won't be helped much by going to a faster GPU, since the primary bottleneck is the CPU performance in this system.
@davidgoodnow269
@davidgoodnow269 4 місяці тому
@@SianaGearz Thank you for the explanations from what I presume to be your experience! I have never considered plugging a Raspberry Pi in to a 3070 Ti . . . . But what basis do you have for this being a low-performance C.P.U.? Back in the PowerPC G3 - Intel Pentium III day, R.I.S.C. C.P.U.s normally gave a 10% per hertz efficiency premium, that certainly remained true between the G4 and P 4. The 64-bit-pure G5, even with its decision tree and circuit paths neutered for export restrictions, smoked even the XEONs. The un-neutered G5 was a die-shrunk mainframe processor that required pre-payment and then a wait of two months to a year for authorization from the Department of Defense to release to ship . . . after the buyer application, a security clearance of anyone who might gain access to the computer, security inspection of the facility the computer would be housed in, vendor access, communications, etc. Just because it's only four cores at 2 GHz is not the limit . . . how many threads on each core? Decision tree branches on each thread? I don't know those answers, myself; I am *very* interested in the Milk V, but haven't e-mailed Sales@Milk because I simply don't have US$1,500 to spend right now. (I am in debt paying for my first year of law school, just because I was interested and could just barely manage the payments.)
@Psychx_
@Psychx_ 4 місяці тому
Wow, theres A LOT of false information here. Intel wants to merge a new kernel driver specifically for Xe graphics and newer. That'll land with kernel 6.8 at the earliest. The userspace part (mesa) will require this new driver to make all the Vulkan API features available that are needed for VKD3D to function properly and performantly. Also - Arc cards perform poorly without resizable BAR and Intel GPUs only support reBAR on Intel CPUs! Xe runs like molasses even on AMD x86 CPUs - so no, Arc wouldn't be the first choice, but one of the worst!
@davidgoodnow269
@davidgoodnow269 4 місяці тому
@@Psychx_ Huh. I put the wrong kernel number in, thanks for drawing my attention to that! I meant to type "6.7" because I read that as being fully supported in that kernel's information. But it has no information at all about the rest, and I consider that, "fully supported" information. If AMD doesn't have the memory buffer support, that would be a significant barrier to it . . . but this is not an AMD CPU, as far as I have seen mentioned. Who knows what memory management arrangements it supports? That would be down to on-chip PCI-E and memory address controllers and drivers. In RISC, those instructions are in the *drivers* whether the hardware circuits are in the CPU or on a separate controller. Reduced Instruction Set is in the name. Sometimes manufacturers lock hardware *out,* as Apple is infamous for doing (infamous among long-time Apple users, anyway), and HP and other manufacturers to a much lesser degree. If you saw the video on that final-version Voodoo 3D card, well, I *had* one that was PCI-X 64-bit (before PCI Express existed; I think 4x AGP had just come out. 64-bit is the square of throughout of 32-bit, and that "4x" is a multiplier of 33 MHz. In this case, though I didn't know it at the time, the GPU cores were only 32-bit and the 64-bit bus was divided between them. Probably why my driver never worked, eh?) for up to 100MHz bus, and I had an HPUX RISC workstation to put it in. What I didn't have is a copy of HP UniX to run on them to grab the driver .kext from, to import to the Linux I was running. I found a vendor to sell it to me, but didn't have the budget for my hobby. (It's not cheap, being corporate development and banking/industrial-market specific software; they don't blink at paying $10,000 for a single-seat program.) I tried for a couple of years to write a driver, but I was completely new to programming and . . . well, I didn't succeed. No one in the Debian Dev group had an example to work with, so since I had bought three of the HP boxes at an auction because they looked cool and my boss gave me that video card as he hadn't sold it, I shipped that box off to a Debian dev who had been interested and willing.
@Psychx_
@Psychx_ 4 місяці тому
@@davidgoodnow269 "If AMD doesn't have the memory buffer support, that would be a significant barrier to it (…)" It's the other way around - the Intel GPU does not fully support non-Intel PCIe root controllers, because it implemented reBAR in a way that's not compliant to the PCIe specification. AMD supports that spec just fine (as can be seen with AMD and NVidia GPUs, aswell as tons of network cards using reBAR just fine on AMD). The Arc GPU situation is independent from the CPU instruction set or driver level! It's either firmware stuff or, (more likely), directly tied to the silicon of the GPU!
@Felice_Enellen
@Felice_Enellen 4 місяці тому
Gotta say, as a gamedev who frequently has had to work directly with assembly language, RISC has probably been my favorite to work with, mostly because you're almost never shooting yourself in the foot by not using the esoteric instruction you didn't realize existed or works the way it does in order to do the thing you're doing manually with multiple other instructions. You just accept that you have a pretty good instruction set and you'll need multiple instructions to do most things, and that's fine if it means the CPU itself can _process_ instructions faster or with a much deeper pipeline, which is the goal with RISC.
@anonymoose4267
@anonymoose4267 3 місяці тому
The things you can do with Load Effective Address are absolutely hilarious, want to multiply a number by five? Use lea because it takes one instruction cycle instead of the multiply instruction which takes multiple xD. Also x86 has two instructions to help calculating sha256 checksums because why not have that built in to the CPU in every core. I'm sure those are called all the time (I am aware they are called pretty regularly, but not THAT often lol)
@IngwiePhoenix
@IngwiePhoenix Місяць тому
Do you think splitting those multiple instructions into a separate function would yield performance gains?
@COLAMAroro
@COLAMAroro Місяць тому
What do you mean you didn't knew the instruction GF2P8AFFINEQB ? smh my head
@OpenGL4ever
@OpenGL4ever 20 днів тому
@@anonymoose4267 There are filesystems like ZFS, that have checksums integrated. Thus they will have to calculate these when writing and reading data.
@anonymoose4267
@anonymoose4267 20 днів тому
@@OpenGL4ever yeah I'm not saying that checksums aren't useful but it just seems like an interesting feature to put in a CPU core. If I were to design it, I'd put stuff like the checksum into a separate chip (maybe still on the CPU, but as a component separate to the cores) so it reduces the complexity etc etc of the CPU. This does reduce the theoretical "checksum bandwidth" but it's incredibly rare to need to calculate multiple sha256 checksums at the same time unless you're running a server or something, in which case just build the functionality into an external card like a network card or something. Also note that I'm not a CPU designer and may not know what I'm talking about.
@AgentOrange96
@AgentOrange96 5 місяців тому
Had Nvidia successfully purchased ARM, I suspect RISCV would have gotten a huge surge in development and adoption. (From companies not wanting to deal with Nvidia) Which would have probably been the only upside to that acquisition.
@VFPn96kQT
@VFPn96kQT 5 місяців тому
China trying to replace US technology it has enough boost as it is.
@siedenburg1
@siedenburg1 5 місяців тому
But now after the acquisition arm announced that they want to produce self and it was planned (don't know if it's still planned) that you could license arm only if you use the chips in your product and not if you sell them to others, so a company like Qualcomm would be without arm
@formes2388
@formes2388 5 місяців тому
@@siedenburg1 Well then... Here comes the RISC-V Development train. But it will probably take a decade to get their.
@platin2148
@platin2148 5 місяців тому
In that sense it’s a SiFive IP combined with a Interconnect. Still amazing can’t wait to see something with more integrated io.
@Sal3600
@Sal3600 5 місяців тому
lol what economics is this😂😂
@sagerobot
@sagerobot 5 місяців тому
Bravo! This is is top content you guys. Seriously, everyone on the LTT team should be proud of this video, its a banger. This is the result of your improved workflow. Whatever happened to make this video happen double down on it. I love the amount of research that went into this, it shows.
@spodefollower
@spodefollower 5 місяців тому
Completely agree - this video was awesome. I learned something and it was entertaining too!
@qwerasdfzxcv98765
@qwerasdfzxcv98765 5 місяців тому
who is glazging
@omarops
@omarops 5 місяців тому
I totally agree but I was hoping they would show the actual CPU inside that machine not a picture from the internet.
@nevadaxelizabeth
@nevadaxelizabeth 5 місяців тому
yes, im impressed with the quality of this specific episode. stands apart from the usual LTT video.
@satyampatel3713
@satyampatel3713 5 місяців тому
@@omarops probably not yet allowed to do that.
@ap194
@ap194 5 місяців тому
Huge kudos to Tanner. And also props to LTT for giving him the ability to keep working on it. It's clear the current pace of videos is giving the team time to work on projects and videos so that they arrive in a more polished and finished form. Have there been times in the last little while that I've missed having a new LTT video to watch? Yeah, probably. But it's nowhere near the satisfaction that I get when I see the payoffs. It may have only been a post script on the video, but Euro Truck Simulator running felt huge and I hope these kinds of moments are continually possible. Way to go Tanner and the rest of LTT.
@CameronTacklind
@CameronTacklind 5 місяців тому
The "dust covers" on the SATA ports are probably to help pickup the part when assembling the mainboard. Having a flat top surface is easier for the vacuum pickup tools. They're just saving an assembly step by not removing them.
@Versette
@Versette 5 місяців тому
True! This is very common on some parts like, for example, female mini USB ports.
@WyvernDotRed
@WyvernDotRed 5 місяців тому
@@Versette huh, so that's what those patches of tape are for! I was wondering about these, as I couldn't figure out their purpose. I was thinking it had something to do with restricting the bending of the clips holding in the connector. It being for vacuum pick and place machines makes more sense though.
@Renis_
@Renis_ 4 місяці тому
My thoughts exactly. I've picked a fair few of those by hand working in the electronics industry
@benoitbenito3319
@benoitbenito3319 4 місяці тому
RISC V is an actual architecture whereas RISC is the instruction set is uses. Big difference lol
@themegjake4000
@themegjake4000 5 місяців тому
This is one of those LTT videos where I am far too ignorant to fully understand what Linus is saying half the time but they still manage to A. Keep me interested throughout the length of the video and B. Inspire me to read up a little bit on the subject and learn some things. I was especially interested in the geopolitical implications, since these sorts of things tend to get lost in the politics news cycle.
@sanctuary7324
@sanctuary7324 5 місяців тому
Shitty ELI5: Think of it like instructions you can give to people. RISC is like a set of instructions that tell someone to turn left, turn right, move forward. ARM has fancier stuff like, go to the room in front of you, go upstairs. x86, the stuff in a lot of computers, have really fancy instructions that includes the above plus things like "go to the house at the end of the street".
@dot_rich
@dot_rich 5 місяців тому
try putting it on x2 speed and laughing at chipmunk linus
@imsorryyourewelcome
@imsorryyourewelcome 5 місяців тому
He said, "There may soon be an industry-changing widespread adoption of an ultra-efficient open-source computing architecture, "open source" being the single most important phrase in this statement." With a far enough outlook over the timeline (involving the future of wearable electronics, the internet-of-things, and humanity's push towards reducing carbon emissions and increasing overall power efficiency) this is more like some *world*-changing stuff that's been slowly happening since 2010 and may now be on the verge of explosive progress. 👍
@he8535
@he8535 5 місяців тому
Linus should talk about box86+box64 not just one but it's not too big a deal
@davidgoodnow269
@davidgoodnow269 5 місяців тому
Look up Linux kernel 6.2, you should quickly find this as something newly supported by that kernel. That's how I found out about it about five hours ago! Truly the first new P.C. that I have wanted in several years.
@4RILDIGITAL
@4RILDIGITAL 5 місяців тому
This RISC-V stuff is some heavy tech. Gotta say, it's cool seeing how far it's gotten. That whole scene of running Steam, even though it was slower than a turtle, felt like catching a glimpse of the future for us gamer folks.
@cromefire_
@cromefire_ 5 місяців тому
I think a misconception as that arm made it easier to port to RISC-V because both are are risc, but what makes it easier is actually just that people are more used to supporting x64 and arm over the last years (especially on Linux) and as a developer it's then easy to just add a 3rd option. For modern developers the adjustment wasn't the instruction set, but just to generate 2 or more builds instead of one.
@xXRedTheDragonXx
@xXRedTheDragonXx 5 місяців тому
As a dev, I'm actually really excited for RISC-V. It's a really cool concept with not a ton or push behind it right now, but there's an immense amount of promise. I hope we'll see it go from a niche platform to something actually used by desktops/laptops in the next few years!
@nocturn9x
@nocturn9x 5 місяців тому
As a dev, I really don't care lol. x86 is fine and good, it's also basically RISC under the hood. The CISC vs RISC debate died in the 90s, and for a good reason. The only reason companies are funding riscv is because it will save them money eventually, let's not kid ourselves
@autohmae
@autohmae 5 місяців тому
The way things are going, ... seems like we'll be running WASM on all the devices so we don't need to deal with binary incompatibility....
@nocturn9x
@nocturn9x 5 місяців тому
@@autohmae lol true
@autohmae
@autohmae 5 місяців тому
@@nocturn9x I feel like the future might not look that dissimilar to the talk: The Birth & Death of JavaScript. After all, eBPF also exists.
@RobertWinkler25
@RobertWinkler25 5 місяців тому
​@@nocturn9xTell that to anyone who has to (or wants to) work with assembly. I actually like assembly and programming in RISC-V or MIPS is sooo much nicer than in x86.
@Turpentinefab
@Turpentinefab 5 місяців тому
You should 3d print a plastic bit to go in the screwdriver so you can use it to point to stuff on a pcb without risking a short.
@ArensLive
@ArensLive 5 місяців тому
that is absolutely genius
@EatYourUranium
@EatYourUranium 5 місяців тому
That will be $9.99
@TylerFurrison
@TylerFurrison 5 місяців тому
I doubt they'd sell it, it's a bit too niche to really do that@@EatYourUranium
@jimmyhopkins8305
@jimmyhopkins8305 5 місяців тому
​@@EatYourUranium + 20% premium
@herrkatzegaming
@herrkatzegaming 5 місяців тому
how about you do it and then put it on thingiverse than point him to it
@xero110
@xero110 5 місяців тому
Just a quick FYI for everyone. ISA or as it's referred to here, 'instruction set'. Does not matter when it comes to performance, it is all down to the microcode. Think of the ISA as the building blocks and the microcode as the blueprint. How well you optimize the blueprints determines how fast you build what you need.
@ausfoodgarden
@ausfoodgarden 5 місяців тому
Yes, with a Reduced Instruction Set it's far easier to optimize microcode. I used to work with Sun on the SPARC systems - Yeah I know, ancient.
@psiah9889
@psiah9889 5 місяців тому
I mean, simplifying the ISA simplifies the whole design, no? Fewer instructions means fewer paths to optimize. Not having to maintain backwards compatibility with x86 bugs that got elevated to features will help with that, too. Theoretically, you can fit the full performance of these most important instructions into a smaller chip that uses less power. But... x86 has a lot of history and momentum behind it. What we have is *heavily* optimized while your RISC-V chips are still largely in the exploratory phases. But... There's a reason Intel's Atom chips never *really* competed with Arm CPUs.
@user-bc7cb8uu7e
@user-bc7cb8uu7e 5 місяців тому
​@@psiah9889It really depends on the task at hand. Even if you can get better optimization of your simpler design, that doesn't mean the software runs in a more optimized way. Just look at how many instructions ARM has picked up over time. Went from not supporting division to supporting SIMD, crypto extensions, and so much more. On the opposite side, x86 is largely RISC under the hood with only the instruction decoder understanding all the instructions before promptly decoding them to simpler instructions that the processor can actually execute. I don't think there's a huge difference in complexity, which is why only a few companies actually design ARM cores (just Apple and ARM these days, I believe) and the rest of the SoC designers just license ARM's core designs.
@hubertnnn
@hubertnnn 5 місяців тому
Not true. ISA can affect performance, specifically in 2 ways: 1. Atomic operation, specifically synchronization. Thread synchronization when done without atomic operations can take hundreds of cycles since it requires trying and retrying to synchronize until it is successful. With atomic operations synchronization is guaranteed to be attempted once and completed successfully.That gives CISC CPUs some advantage in multithreaded applications that often synchronize. 2. Heat generation. A CPU core is generating heat from each circuit regardless if its being used or not. The reason is that all of those circuits are doing the calculations at the same time and all results except for one are discarded. The more instructions a CPU can handle, the more circuits a core needs and the more heat it is generating. And heat generation significantly limits max clock speed of the CPU and thus its max performance. First one give CISC advantage, while second one gives RISC advantage. And over time the heat issue became so limiting that RISC started winning with CISC, even though CISC was the performance winner for many years.
@BruceHoult
@BruceHoult 5 місяців тому
Standard RISC-V instructions until now don't need microcode or microops -- or at least 1 instruction = 1 microop. Future high end cores might combine 2 instructions into 1 microop sometimes (as recent x86 and Arm do for "cmp;bCC" but nothing in the market right now (and not this computer for sure).
@annoorange123
@annoorange123 5 місяців тому
LTT group is delivering on the promises that they made after the one week shutdown. content these days is more informational, while still having LTT entertainment. Great job guys, more videos like this!
@darcagn
@darcagn 5 місяців тому
Oh yeah, like the part at 16:15 where they call Imagination Technologies a "startup" even though they were founded in 1985 and their GPUs were used in the Sega Dreamcast and the original iPhone. I love that accuracy!
@genelinet1941
@genelinet1941 5 місяців тому
@@darcagn I'm not catching that in your timestamp.
@Tr4ns1st0r
@Tr4ns1st0r 5 місяців тому
@@darcagnI think you’re hearing things, buddy.
@bobbywibowo
@bobbywibowo 5 місяців тому
@@darcagn it's "useless" to cite only issues like that, since LTT has access to editing videos post-upload (yes they already got rid of that error at the time of this comment's writing). though they did announce it in the pinned comment so you aren't hallucinating
@mbsfaridi
@mbsfaridi 5 місяців тому
@@genelinet1941Check the pinned comment, they apologised.
@radornkeldam
@radornkeldam 5 місяців тому
Not all of RISC-V is free or open source. There's a baseline that is, of course, but then, many of the highperformance extensions you'll see popup will probably be propietary to one manufacturer or another, and it's yet to be seen whether any of them become free/open like the main spec, particularly those that eventually succeed in the market. It's probable there will be licensing costs to the most desirable extensions to RISC-V.
@AndyHerbert254
@AndyHerbert254 5 місяців тому
If Google could make RISC V good for their Chromebooks, and later on Android, maybe they could have something to actually compete with Apple since shooting their pixel lineup in the headphone jack.
@Tophatguy_vr
@Tophatguy_vr 5 місяців тому
First
@Spirrwell
@Spirrwell 5 місяців тому
I've been really excited about the future of RISC-V for a while. I really hope they break into the mainstream sometime in the next few years with all the major issues worked out.
@Tophatguy_vr
@Tophatguy_vr 5 місяців тому
First
@CanIHasThisName
@CanIHasThisName 5 місяців тому
In specialized hardware, quite possibly. But for general computing, I wouldn't expect it this decade, especially with all major issues worked out. One of the biggest factors for X86 is backwards compatibility. Right now, you can buy a 20 years old videogame and just play it without any fuss. Moving to a different architecture will have to contend with that.
@Spirrwell
@Spirrwell 5 місяців тому
​@@CanIHasThisName Apple did it with ARM. I imagine creating RISC-V chips that are specialized for x86 and/or ARM translation will have a big impact. As I understand it, Apple has specialized ARM chips that help it work better with the x86_64 translation. AFAIK Microsoft has been kinda stagnant in the ARM space because of a deal with Qualcomm. But I could be wrong. If Microsoft took RISC-V seriously and invested in the future like Apple did, we could see something very compelling. We can already see in this video that gaming IS possible. It's definitely not there yet. But the sheer fact that it can be done means that it will inevitably get better. Hopefully once the RISC-V instruction set gets standardized vectorization instructions, we'll see much better performance.
@nocturn9x
@nocturn9x 5 місяців тому
@@Spirrwell Apple did it with ARM because they already had experience migrating from a RISC architecture like PowerPC to an x86 one when they moved to Intel processors. And even then, compatibility is far from perfect. And ARM has been in development for a lot more time than RISCV, with a lot more resources behind it. So, yeah, not in this decade. Probably the next
@Spirrwell
@Spirrwell 4 місяці тому
@@nocturn9x These days moving an OS to a different CPU architecture isn't the hard part. It's the legacy software support that's the issue. The hardware acceleration aspect of the RISC-V platform will improve and performance will improve. Given the pace, I expect that to happen very quickly. I give it 5 years and we'll probably see a pretty compelling Linux desktop experience with RISC-V. Whether Microsoft or any other company jumps on the opportunity or not, who knows? But it's all happening really really fast. Much faster than I certainly would've expected.
@jody5661
@jody5661 5 місяців тому
those "dust covers" are for the PCB manufacturing process. it gives the pick-and-place machine something to grab and pick up the part with. that being said they are usually removed as one of the final steps of assembly of the PCB before it shipped out.
@aiden_3c
@aiden_3c 5 місяців тому
I remember seeing stuff about RISCV when it was first coming out. I was so excited for the ecosystem to develop and for Linux to get on it. Here I am now with a Milk V Duo on my desk, writing software for the Linux it came with.
@me2olive
@me2olive 5 місяців тому
This takes me back to the Transmeta Crusoe from 2000, a RISC CPU that converted x86 instructions to its own in its middleware. It's a pity it never caught on as that was a really interesting concept, no software emulation/translation required, and they even demonstrated swapping out the middleware layer for a Java Bytecode one.
@catchnkill
@catchnkill 5 місяців тому
Though I have not used a laptop computer using the processor, one of my colleague did own a Fujitsu laptop have the Crusoe processor. It was a tiny laptop with a long battery life. However it did not pull out a significant lead over the usual Intel processor laptop computers. Another major competition disadvantage of this Crusoe processor technology is that it needs extra RAM to work well. Notebook computer RAM chips were expensive that day and eventually it was caught up by competitors.
@Deinorius
@Deinorius 5 місяців тому
I would be really nice, if in the future there comes a RISC-V CPU that's able to do x86 instructions kinda natively. That's the only gripe I have with the probably coming transition from x86 to RISC-V. Game compatibility and efficiency. The latter especially is necessary for gaming handhelds like the Steam Deck.
@alfiegordon9013
@alfiegordon9013 5 місяців тому
​@@Deinoriusyeah, the big thing about the Crusoe wasn't the instruction translation, it was stuff that we now have on x86 cous, like dynamic frequency scaling
@TheReaper-fq6yv
@TheReaper-fq6yv 5 місяців тому
@@Deinorius Modern x86 cpus are risc in nature beneath the hood. The complex x86 instructions are broken down into simpler risc instructions. It is possible for a third party to design an x86-RISC processor but the problem is licensing. Only AMD and Intel are allowed to implement the x86 ISA.
@l10industries
@l10industries 5 місяців тому
No Transmeta CPU was EVER RISC. They were VLIW which quite frankly isn't anything close.
@kiwimagic9857
@kiwimagic9857 5 місяців тому
The "dust covers" at 8:43 are more likely to be pads providing a surface for the suction cups to grab onto during automatic pick and place assembly of the motherboard.
@kristiansims
@kristiansims 5 місяців тому
Yep. Same reason a lot of USB/HDMI ports have kapton tape on the top. Sometimes it’s not worth removing
@TaylorPassofaro
@TaylorPassofaro 5 місяців тому
I really liked this style of video! The explanations cut in with the showcase really made me feel like I was learning something here, and learning something cool! This is the style of LTT video that draws me in and engages me longer than most. I imagine that it took more work, but hopefully this type of "explanation / showcase" video continues.
@ruediix
@ruediix 5 місяців тому
As a note, for the past few decades, most CISC processors use an RISC Microcode with a CISC interpreter unit, creating a CISC on RISC hybrid machine. This is more complicated than regular RISC processor, but less complicated than designing a CISC processor for a particular instruction set.
@aggyz4509
@aggyz4509 Місяць тому
So could they theoretically keep the risc microcode and swap out the cisc interpreter for a risc interpreter? Meaning they would have incredibly fast riscv processors?
@Tall_Order
@Tall_Order 5 місяців тому
I'm definitely looking forward to this. Once the performance is improved, I may pick one up to use as a general purpose living room family work / research pc (with light gaming).
@zennyboi1
@zennyboi1 5 місяців тому
Sophon, what an ambitious name. Plus that teardrop logo. Its a reference to a supercomputer in a sci-fi novel (3 body problem)
@Thomas-lv9se
@Thomas-lv9se 5 місяців тому
You guys are amazing! This video is so well done and the combination of "fun" content and things you can learn from on the channel make it nothing less than the greatest channel on youtube. It's obvious your workflow has improved a lot and the quality and level of awesomeness have improved by at least 5 freedom eagles per Hamburger or whatever unit of measurement you North Americans use for awesomeness.
@10001000001
@10001000001 5 місяців тому
Two notes: - That's not a dust cover - it's a piece of plastic used to suck onto during Pick & Place assembly. - The greenspill on the CGI background scenes is quite bad.
@totallyuneekname
@totallyuneekname 5 місяців тому
Never thought I'd see Linus using Fedora. On a better-supported CPU, it's my OS of choice!
@dragon2knight
@dragon2knight 5 місяців тому
I like the idea of this, and I'm looking into getting one of these....but I'm not in a hurry. This needs a little bit more, shall we say, finesse, before I totally commit to it. This has really come far in the last few years though and I like this implementation. Thanks for showing it to the great unwashed Linus, it's pretty amazing!
@austinwhite3132
@austinwhite3132 5 місяців тому
I’m sticking with my Intel 9900K
@davidfernelz
@davidfernelz 5 місяців тому
This feels a bit worse than when i tried Linux out 10 or so years ago. Considering where Linux is at now, especially with the Steam endorsement... Yeah this is a lot closer than most would think looking at it. I completely agree with the overnight aspect is a big tech company commits.
@gamm8939
@gamm8939 5 місяців тому
@@davidfernelzI mean Linux is still a pain in the ass to use and is really more for people who actually have fun dealing with random nonsense than people who want or need to use their computer.
@Tall_Order
@Tall_Order 5 місяців тому
My thoughts as well. Once the performance is ironed out, I'd like one of these in my living room for family general purpose use.
@powerfulshammy
@powerfulshammy 5 місяців тому
Just get a OLED steam deck better than this also don't need a gpu more expensive than your scooter also cheaper than this
@saquial
@saquial 5 місяців тому
In University we ended up simulating a RISC-V core from scratch, writing code which describes how the components of the core dealt with the binary input and what did they output. We ended up loading assembly code to its memory, more specifically the code for 5!. The moment I saw 120 in the simulation tool was something else.
@whendidamfask
@whendidamfask 5 місяців тому
Sorry to break ur heart but 5! = 120
@saquial
@saquial 5 місяців тому
@@whendidamfask Yeah, my bad there, I just wake up, editing soon. (The processor did solve 5! correctly, I just forgot what 5! was lol)
@electrostatic4677
@electrostatic4677 5 місяців тому
​@@saquial lol😅😂
@MenkoDany
@MenkoDany 5 місяців тому
Imagination Technologies is NOT A GPU STARTUP LMAO. Linus you were alive in the 3dfx days they've been making GPUs along with Nvidia, 3dfx, Intel and ATI and others...
@dedpix3l
@dedpix3l 5 місяців тому
Loved this infographic format, a bit like essay and more in depth, would love more about x86 and or x64 history aswell as arm
@arokan327
@arokan327 5 місяців тому
You mentioned box64 for 64-bit applications... there's also box86, that came out even earlier, that works with x86 apps :D
@Mockedarchie
@Mockedarchie 4 місяці тому
I REALLY like the cut away to green screen Linus for descriptive stuff. It makes details a lot more clear without the clutter of a scene. I feel like it should definitely be used more often.
@TamayJentjens
@TamayJentjens 5 місяців тому
Guys I just want to give you a big ups for the amount of work that has been put into this video. And for the effort that you do to really see things from a more nuanced and complete point of view. I was skeptical in the beginning of your intentions when you said you wanted to move on to more quality Content. this to me is a clear statement that you do. thenk you for the work!
@ZypherGames
@ZypherGames 5 місяців тому
Honestly this was great to see, I had no idea RISC-V was so far along already! I agree with LMG, in 5 years i'd expect to see RISC-V alternatives to chromebooks and macbooks.
@localzuk
@localzuk 5 місяців тому
Imagination Technologies isn't a "GPU start-up". Its been around since 1985 (previously called VideoLogic), and has been creating the PowerVR line of GPUs pretty much since then. Its chips have been used widely in TV set-top boxes since the 90s.
@BruceHoult
@BruceHoult 5 місяців тому
Correct. And their CPUs don't use RISC-V, they're just trying (successfully) to sell them to companies making RISC-V SoCs e.g. JH7110, TH1520.
@tonyolmstead8282
@tonyolmstead8282 5 місяців тому
This is a hilarious coincidence. I just got in to risc v last month, and I’ve been reading up on cpu design and architecture ever since. I really want to see risc v take off
@Metal_Maxine
@Metal_Maxine 5 місяців тому
and now, when people ask what you're doing, you've got a more mainstream explanation to point people too.
@ShadyHero
@ShadyHero 4 місяці тому
I remember the last risc-v video (im pretty sure it was around the time of the 6 editors 1 CPU video with all the titan V's) it's exciting to see how far it's come!!!!
@nyanbadacc5228
@nyanbadacc5228 5 місяців тому
First really good ltt video in a while. Great job on this one! I would love to see more topics like this!
@CrystallisedEntertainment
@CrystallisedEntertainment 5 місяців тому
03:19 - If I could make a small correction here: ARM originally to stood for "Acorn RISC Machine" as Acorn Computers in Cambridge, UK originally invented ARM for their "Archimedes" computers in 1987. I believe in 1993 or 94 it was renamed to "Advanced RISC Machine" as it is known today. 😅 Changing subject - it's amazing to see how far RISC-V has come in 5ish years; can't wait to see how it goes in the future! 😄
@gorkskoal9315
@gorkskoal9315 Місяць тому
oooh 5 years it can...play a indie game at 15fps. wow. such speed. in 15...it'll play pong and chrisis core.
@Carstuff111
@Carstuff111 5 місяців тому
Honestly, as a player of both Euro Truck Simulator 2 and American Truck Simulator, both are not exactly very well optimized. The fact you got Euro Truck Simulator 2 running at all is insanely awesome! I don't have THE most powerful machine on the planet, but a Ryzen 5 5600X, 16GB of 3600MHz RAM and a RTX 2080 Super shouldn't have any trouble running either game at 2560x1440 with highest settings, and REALLY shouldn't have issues at 1920x1080, yet I still get dips and drops at both resolutions. Seeing it run at all on RISC-V is mind blowing. Edit: Also, I am assuming you ran Euro Truck Simulator 2 in OpenGL?
@zoltanberkes8559
@zoltanberkes8559 Місяць тому
Actually, most of the ideas behind the original RISC goes back to the sixties and Seymour Cray. He realized that CPUs spend most of their time to load operands from and store result to the memory. Then he designed a CPU which only had few functional units (ALU, Branching, Load/Store) and all instructions (except load and store) were register to register ones. The CPU was much faster. So far, I just described the original RISC, but it happened about 15 years earlier then guys at Berkeley started calling it as RISC. In his next CPU design, Seymour Cray even modified the controller unit to run multiple instructions in a time shifted parallel way: pipelined execution. He also let different functional units work on their respective tasks at the same time: superscalar execution. What else were discovered in the sixties? Almost everything: cache, virtual memory addressing, machine virtualization, resource isolation etc. The only thing I haven't seen any reference from the sixties is register renaming.
@gamerpaddy
@gamerpaddy 5 місяців тому
8:38 those SATA covers are not dust covers, they are needed by pick and place machines since they require a flat surface for the vacuum tip to hold on in the manufacturing process. you see these covers on pin headers sometimes too.
@jazoice
@jazoice 5 місяців тому
1:20 "most of its specs are pretty normal" *immediately mentions the 128 gb of ram*
@DeadRyGuy
@DeadRyGuy 5 місяців тому
Really great video! As much fun as a lot of the content is (house tech, home upgrade, etc..), this style of video has more substance. Seems like a fair bit of research and effort went into the writing, good stuff!!
@blackasthesky
@blackasthesky 5 місяців тому
I've worked on RISC-V related projects, and I am excited to see this being a thing. The only thing I can complain about is that it is a pain in the butt to work with RISC-V International as a third party.
@jeffwang543
@jeffwang543 5 місяців тому
I enjoyed the video, finding it quite interesting. However, I believe the portrayal of the current reality of RISC and CISC was somewhat oversimplified. While it's accurate to say that RISC is more power-efficient, it doesn't necessarily translate to RISC being inherently "faster" than CISC computers. The intricacies involved in this comparison are numerous, with a notable example being the state of the x86 architecture. In this context, many CISC instructions are broken down into RISC-like instructions, a process that demands additional power due to purpose-built hardware converting instructions at the outset. To sum it up, the speed of processors like the M1 isn't solely attributed to their RISC architecture.
@mmbb1645
@mmbb1645 5 місяців тому
They never claimed that RISC is inherently faster than CISC. Just that it has the potential to be equally as fast.
@Supermrloo
@Supermrloo 5 місяців тому
I know videos like these must take a lot of effort to make, but they’re by far my favourite! These are the kind of videos which made me jump ship to a CS major
@user-rd2ev9vw5d
@user-rd2ev9vw5d 5 місяців тому
The sata "dust covers" are there for PCB assembly so the pick and place machine has a large flat surface to get suction on.
@sorek__
@sorek__ 5 місяців тому
I am very excited to see RISC-V being adopted. As embedded engineer I can't wait for more RISC-V microcontrollers, having cheap and license free MCUs is a life saver and I am all in for supporting such endevours. We are not there yet in terms of raw processing power, but at some point someone will jump on this ship, once it start getting popular. RISC-V is equivalent to Linux in our world (well in some cases anyway).
@meneldal
@meneldal 5 місяців тому
The problem is having a free ISA means nothing if you have to license actual implementations (except I guess giving you more options than just ARM). Making an efficient core for an ISA takes a lot of dev time, and I'm not seeing the big companies offering that for free. Maybe there will be open source cores that can deliver good performance but that could take a while.
@mitcoes
@mitcoes 5 місяців тому
@@meneldal I think there will be RISC-V soon in NVMe modules and other devices, and probably later some free alternative to the first ones. I think I read WD was working on using new RISC-V microcontrollers.
@sorek__
@sorek__ 5 місяців тому
@@meneldal implementation is something that different companies will compete about but the standard itself will be free and this is really incredible part.
@tbuk8350
@tbuk8350 5 місяців тому
It's always great to see more coverage of the best CPU architecture.
@handlemonium
@handlemonium 5 місяців тому
Time to buy some Synopsys stock whenever the stock market dips next 😏
@AbrahamMusalem
@AbrahamMusalem 5 місяців тому
More videos like this please! Great job y’all.
@ragtop63
@ragtop63 5 місяців тому
I used to use a Sun Microsystems desktop back in the 90s. It was designed to run a desktop version of Sun Solaris. I installed a RISC port of Redhat Linux on it. It was leaps and bounds faster than a similar spec x86 desktop running the same OS.
@reteurg96
@reteurg96 5 місяців тому
This is by far one of the best videos released on this channel in recent months.
@SilverKnightPCs
@SilverKnightPCs 5 місяців тому
I love how this video was basically a tech quickie mixed with an awesome computer impression. More like this!!!!
@Nabee_H
@Nabee_H 5 місяців тому
This feels like a Techquickie and LTT hybrid video.
@Spirrwell
@Spirrwell 5 місяців тому
It does, I actually quite like it.
@Guybrushgg
@Guybrushgg 5 місяців тому
This was a great video. It was very interesting knowing more history/background about how companies and even governments "use" the chip industry. I'm waiting for the moment in 20+ years for a RISC cpu with quantum processing capabilities being released, if that ever happens :P
@NaClSandwich
@NaClSandwich 5 місяців тому
This is legitimately the best video I have seen this company output - and I have been watching for a looong time. Congrats whatever you are doing over there is working.
@PSwayBeats
@PSwayBeats 5 місяців тому
You know you got cash when you say 128 gigabytes of RAM is normal
@mostlyinept
@mostlyinept 5 місяців тому
Yeah I thought that was so weird. I figured he must have made a mistake lol
@thorwaldjohanson1263
@thorwaldjohanson1263 5 місяців тому
@@mostlyinept I mean it is a 64 core dev board with eec memory. The aim is not consumer but rather the server / compute / research market.
@SujeetRaj711
@SujeetRaj711 5 місяців тому
​@@thorwaldjohanson1263yea. But with benchmarks of less than 10% of S23 ultra (a mobile), with 10 times more RAM 😮
@psiah9889
@psiah9889 5 місяців тому
​@@SujeetRaj711Dev boards like this tend to vastly over implement ram, and, well, ram is cheap right now. Like... I think the Xbox 360 had like 512mb of ram but like 32gb on some of the dev boards.
@Pasi123
@Pasi123 5 місяців тому
@@SujeetRaj711 In Geekbench 6 a laptop Core2 Duo beats it in multi-core and especially in single-core. I don't think it actually ran on all of the 64 cores. Intel Core2 Duo P8700 (25W TDP): 366 single, 550 multi Intel Celeron N3350 (6W TDP): 290 single, 513 multi
@911delorean
@911delorean 5 місяців тому
I would love to see these in DIY home servers or on those premade NAS boxes or even on PIs.
@No-mq5lw
@No-mq5lw 4 місяці тому
There's the LiChee Pi 4a, which is the closest thing to a Pi RISC-V has right now. But it's not really in a true Pi form factor as it's much closer to a CM3 in function. That being everything is on a DDR SODIMM module complete with ram, cpu, and storage that can be broken out into whatever you want. Current carrier board is geared less towards what's expected of a Pi and more tablet or computer. There's also the Beagle-V Ahead that's more spiritually in line with a Pi, though it's part of the Beagle ecosystem.
@xavierjiang7112
@xavierjiang7112 5 місяців тому
Sun Microsystems actually made entire racks of RISC CPU with their SPARC CPU lineups. Although that was probably lost to history.
@davidgoodnow269
@davidgoodnow269 5 місяців тому
I *_just_* was looking at that, wanting one, a few hours ago! I found out about it when looking up Linux kernal 6.2 and went, huh, I haven't seen RISC since I lost my old HP/UX machine! I am currently wanting one *so bad!*
@takinabradley
@takinabradley 5 місяців тому
This was a great video! I really felt educated about the technology, and how it (could) really change the world. I've never felt so excited about an open-source technology. It really made me want to pick up some kind machine and dive into ARM/RISC-V assembly to learn more about it. I've also never really thought about the political effects of open technology like this. I've always felt very pro-open-source since I got into programming, but I've never really thought about the wide ripples such projects could have on the world around us. I can't help to feel somehow trepidatious, though, as this technology evolves- especially if it continues to evolve as quickly as Linus makes it feel like it might.
@PatArthurRules
@PatArthurRules 5 місяців тому
What sort of political effects are you concerned about?
@takinabradley
@takinabradley 5 місяців тому
@@PatArthurRules I think anyone would be wise to be wary of anything that might spark more tension between Western countries (particularly the US) and China. The US can control it's private tech companies, and use it's sway to enact sanctions on China (more have been added just this year), but it's a lot harder to control free ideas. There is no doubt China has manufacturing prowess. If the US really wants to keep more modern tech from China, and this tech evolves rapidly in the near future, I could imagine a tone shift in the political sphere. As pro-tech as I am, I think the evolution of this story could be a scary one. I'm not the most well-versed politically, but this feels like writing on the wall if a way isn't found to dissipate the that tension.
@PatArthurRules
@PatArthurRules 5 місяців тому
@@takinabradley Ahh, so what you're worried about is China getting the upper hand on up and coming tech, and the way that could threaten the US's political hegemony in general? Not trying to put words in your mouth, just trying to make sure I understand
@takinabradley
@takinabradley 5 місяців тому
@@PatArthurRules I could give a rat's ass about my country's political hegemony. But, I worry that the facts are that this may be perceived as a threat in the near future, and cause some amount of- possibly violent- unrest. I don't worry about China achieving 'tech parity' with the US. I don't perceive it as some kind of threat to " 'muh country ". It will happen eventually, and I personally believe that any tech that improves lives should probably migrate around the globe. I, as a citizen that would prefer if it's country did not start a large potentially armed conflict, just worry it might start instigating as 'tech parity' gets closer. In short, the future seems scary and uncertain, and that's really all I'm getting at.
@PatArthurRules
@PatArthurRules 5 місяців тому
@@takinabradley Totally understand and respect that! It also makes me very nervous that even things like this could develop into world-threatening conflicts
@JustMortHandle
@JustMortHandle 5 місяців тому
I've been waiting for this since 2015 or whatnot. I remember being super excited for it. RISC-V has some so far since the beginning and it looks so damn promising for the future!
@Mystic_WW
@Mystic_WW 5 місяців тому
For a moment I thought it'd be some weird thing. Something like the CPU being super power efficient and offsetting its cost by offering outsourced processing power or something like that.
@rusinsr
@rusinsr 5 місяців тому
I love it when you guys do videos that go in depth on a topic from all angles and teach the viewers about cool tech. Really well done video! :D
@burnin8orable
@burnin8orable 5 місяців тому
It's impressive that Fedora supports RISC-V.
@carlosrfs99
@carlosrfs99 5 місяців тому
Debian and Archlinux supports too, also gentoo.
@autohmae
@autohmae 5 місяців тому
Similar as with ARM, it took something like 10 years to get to this point (ARM took longer, but as mentioned in the video the effort on ARM helped RISC-V).
@guiorgy
@guiorgy 5 місяців тому
3:20 Didn't the A in ARM stand for Acorn, the company that originally invented ARM?
@sarreqteryx
@sarreqteryx 4 місяці тому
@LinusTechTips ARM currently stands for Advanced RISC Machines. it used to stand for Acorn Research Machines
@TMKCTech
@TMKCTech 5 місяців тому
6:30 - You would be looking for Box86. Same as Box64 but... well... for x86 applications instead of x64. You can install box Box86 and 64 at the same time and run them both as the same time.
@user-xr3rb6pn9m
@user-xr3rb6pn9m 5 місяців тому
That's probably what they ended up doing for running Eurotruck Simulator, even if they cut explanation. I tried both Box86 and Box64 on an ARM machine. Very cool when it works but very painful to set it up. Tons of development time is needed to make this thing as transparent to users as Rosetta
@Metal_Maxine
@Metal_Maxine 5 місяців тому
@@user-xr3rb6pn9m On Floatplane, Tanner commented that they were having (unspecified) problems with Box 86 on rv64.
@_DRMR_
@_DRMR_ 5 місяців тому
You forgot to mention that Western Digital has been investing in RISC-V since many years. Nearly every harddrive contains at least one ARM chip (or at least they used to), so imagine the number of royalties that companies like WD have to pay to ARM. If they can mitigate this by developing their own license free processors this could be a serious cost saving for them. In 2019 WD bought (and opensourced) PlatformIO, with the explicit intent to improve development for embedded RISC-V.
@meneldal
@meneldal 5 місяців тому
I'm not sure they'd be saving so much money. The ARM chips in a hard drive are tiny, the licensing comes up to only a few cents for that. At least short-term, investments in risc-v are a big loss, it will take them years before they get to a design that is on par with ARM,
@fab1604
@fab1604 5 місяців тому
This has to be one of the best LTT video of the year, so informative and important, great job by the staff
@woedendstewadpier4922
@woedendstewadpier4922 5 місяців тому
I also like the one in which Linus falls in the pool. But for different reasons.
@ThatNateGuy
@ThatNateGuy 5 місяців тому
I love RISC-V and I can't wait to see where it goes from here. I'm looking forward to seeing handhelds and other consoles developed on it.
@SianaGearz
@SianaGearz 4 місяці тому
There are only two fundamental differences between RISC and CISC. First is that CISC includes instructions that perform 3 very different things at once - read-modify-write cycle on RAM. On RISC, the memory operations are fully separated, so memory instructions only copy CPU's internal register to memory, or memory to register; while arithmetic and other instructions only work on registers and cannot touch RAM. A corollary is that CISC processors are more reliant on memory interfacing, while RISC processors usually have a larger register set and interface to memory less. A cute side effect is that almost all RISC processors can switch their endianness, because they only need to swap out a very small amount of logic for that, since it only affects memory instructions and not all instructions. The second major difference is that CISC instructions are all of different length, and can include prefixes which make things insanely complicated. So parsing the CISC instruction stream takes up a lot of logic. While all RISC instructions are the same length. In the end, modern remaining CISC processors convert most instructions to a RISC-like internal representation, and also manage a larger register set than exposed to software by "register renaming". I would not say that developing directly in machine language for either architecture is easier, after all the large register file is a massive advantage; and in turn code density is not a priori worse on RISC ever since Hitachi SuperH and ARM Thumb, which have 16-bit instructions. It's wrong to say that the instructions of a CISC processor are more powerful or more "robust". The 32-bit ARM instruction set (a deprecated classic) contained conditional tags in each and every instruction, which easily allowed branch-free code, which is just such an insane boon for both security relevant code and DSP algorithms. Lack of a similar mechanism in modern processors has opened up numerous secure data exfiltration side channel vulnerabilities.
@platin2148
@platin2148 5 місяців тому
Imagination Technologies isn‘t a startup it’s probably as old as the iphone. And already was inside an iphone also most of the apple gpu people seem to come from there.
@ChronoXShadow
@ChronoXShadow 5 місяців тому
Dang. Really cool to see RISC-V on a graphics card. I was thinking that the M1 on a GPU would make it an ai and simulating beast. Hopefully, if RISC-V can be utilized well, it can become the competitor Intel was supposed to be in the gpu market.
@autohmae
@autohmae 5 місяців тому
Some of the first big companies who were looking at adopting RISC-V for their products were: Nvidia for their GPU cores and Western Digital for their harddisk controllers.
@pedroramirez2215
@pedroramirez2215 5 місяців тому
This is the content man, most ppl don't even know risk exits and even less are up to date, well done
@arnabbiswasalsodeep
@arnabbiswasalsodeep 5 місяців тому
One project I love is dark risc v, written by a guy in one night on spartan 6 fpga. So one smart madlad & one over the night, dark, programming session later you had an implementation of risc v architecture, which is nuts. Just hardware paths need R&D along with better software compatibility. 1st is easy via big companies like Google as mentioned in vid, 2nd should be fine since software takes less time than that.
@nocturn9x
@nocturn9x 5 місяців тому
You clearly have no idea about what you're talking about if you think software takes less time than designing the architecture. They go hand in hand.
@fridaycaliforniaa236
@fridaycaliforniaa236 5 місяців тому
Given how the Chinese managed to greatly improve their space program in less than 15 years (almost starting from nothing), I'd give them less than 10 years to make this become as commonplace as Android phones or ARM Macs...
@0ptixs
@0ptixs 5 місяців тому
This is the kind of videos I love from ltt. Great info, real world applications, excitement about the future of tech. Love it!
@howardwhite9773
@howardwhite9773 5 місяців тому
Old time Computer Science guy here. Please review the work of Seymour Cray at Control Data Corp in the 1960s. The CDC 6500 Compass assembler had 64 total instructions. Various bulletin board posts of the same time had a variety of "IBM style" mnemonics poking fun at the existing S/360, S/370 complex instructions sets. "Sharpen Light Pencil" and the like...
@metalhead2550
@metalhead2550 5 місяців тому
Imagination isn't a startup they've been around for decades
@FlorinArjocu
@FlorinArjocu 5 місяців тому
Before getting to the M1 moment, Arm was the top edge for many years in phones. So, that was not exactly overnight. I don't think we can have a Risc-V overnight, either. Right now it is not on par with Arm when talking about performance / watt, but of course, improved designs could achieve that. In 5-10 years it could be possible to have that "M1" equivalent moment with Risc-V.
@autohmae
@autohmae 5 місяців тому
It is on-par with ARM when talking about microcontrollers-size, next up is embedded space.
@nocturn9x
@nocturn9x 5 місяців тому
Yeah it'll probably take a decade or so
@vandee28
@vandee28 5 місяців тому
Until there is a party that will offer testing and /very important/ validation this will not move into consumer / business -space. The reason most companies choose ARM is because the validation that comes with the license.
@danielburger2550
@danielburger2550 5 місяців тому
I really liked the cut-aways to Linus explaining the topic in depth. Awesome video!
@SilentShadow-ss5xp
@SilentShadow-ss5xp 5 місяців тому
Imagine tomorrow Apple comes out and is all like "yea we are transitioning our architecture once again to riskv" lol
@h2Lu
@h2Lu 5 місяців тому
lol
@the_undead
@the_undead 5 місяців тому
With how much Apple has invested into arm, I highly doubt that will be the case for at least the next year or two, however, if the next generation Chromebooks are running riskv and have no apparent downsides compared to previous Chromebooks then I think Apple will seriously consider switching to riskv because it'll be cheaper
@No-mq5lw
@No-mq5lw 5 місяців тому
​@@the_undeadThey will more than likely start with microcontrollers for airpods and the like, while keeping the threat of switching to home grown RISC-V processors in their back pocket as a bargaining chip.
@the_undead
@the_undead 5 місяців тому
@@No-mq5lw That sounds a lot more like what Apple would do then just switch to risk v
@lophilip
@lophilip 5 місяців тому
More likely, is Apple making moves to purchase ARM
@silversonic1
@silversonic1 5 місяців тому
I prefer to mispronounce it as "Risk Vee", so I can can call every company working on it a "RISC-V Business" Yes, my dad humor is in full force and I'm not even a dad!
@codytappen
@codytappen 4 місяці тому
I studied computer architecture in RISCV at UCSC, but we used MIPS for our assembly course graduated in 2020
@looptypoop
@looptypoop 5 місяців тому
I literally learnt about this yesterday for my course and having ltt as study material is just awesome
@cyberneticghostofchristmas
@cyberneticghostofchristmas 5 місяців тому
So, when are you water-cooling it? Perhaps sub ambient?
@650kp
@650kp 5 місяців тому
Legacy is why they use old ISAs. Theoretically all ISAs do the same and we use lots of microchips so let’s just go in on and develop RSA
@No-mq5lw
@No-mq5lw 5 місяців тому
ARM doesn't care about legacy. x86 is really the only mainstream isa out there that really cares about legacy.
@IngwiePhoenix
@IngwiePhoenix Місяць тому
Coming back to this after tinkering with a VisionFive2 and being in the process of setting up automated cross-compiles. The JH7110 by StarFive/SiFive is almost kernel mainstream and the amount of T-Head device drivers I have seen is genuely mindblowing. They are even working on an edk2 port, which is effectively an open-source BIOS... well, UEFI-BIOS...thing. BUT! Anyway; things are exciting - and let me tell you, this video? Outdated. x) The Milk-V Oasis that will drop by the end of this year will do some numbers and it'll actually be insane. Since this video, RISC-V Vector extension has also dropped... which you need for AI. So, if your CPU has the V-extension and you get a GPGPU-capable GPU and smash them together, we could see a ~100w AI box. And in the process, even better performance, as MESA and friends improve their drivers. So, this'll be quite fun ^^
@AarPlays
@AarPlays 5 місяців тому
The production on this video really stands out. It was informative and fun, good job
@n00bnetrum
@n00bnetrum 5 місяців тому
This gives me real wine 0.9.8 vibes. If you've ran GNU/Linux and tried to game back then you know what I mean. And if you compare that ancient version of wine with today's proton then you know where that RISC V stuff is headed.
How is THIS a Gaming Mouse?
23:23
Linus Tech Tips
Переглядів 1,6 млн
I Upgraded His Car While He Was On Vacation
17:13
Linus Tech Tips
Переглядів 129 тис.
Анита просто на химии, поэтому такая сильная
00:21
Женя Лизогуб SHORTS
Переглядів 2,8 млн
BMW просто издевается над нами! Силы на исходе…
1:34:41
ИЛЬДАР АВТО-ПОДБОР
Переглядів 3,4 млн
Explaining RISC-V: An x86 & ARM Alternative
14:24
ExplainingComputers
Переглядів 410 тис.
I'm Embarrassed For Her... - PC Build with NCIX Esther
27:39
Linus Tech Tips
Переглядів 3,4 млн
Phone repairs are getting ridiculous - Here's why.
22:24
Mrwhosetheboss
Переглядів 2,3 млн
Something Strange Happens When You Follow Einstein's Math
37:03
Veritasium
Переглядів 7 млн
RISC-V 2024 Update: RISE, AI Accelerators & More
14:03
ExplainingComputers
Переглядів 73 тис.
Don't Waste your Money - $30 vs $10,000 Racing Setup
30:37
Linus Tech Tips
Переглядів 3,5 млн
A $9 Introduction to the RISC-V Future of Computing
26:34
apalrd's adventures
Переглядів 326 тис.
Amazon... more like SCAMazon - Fake SSDs
19:06
Linus Tech Tips
Переглядів 4,3 млн
Mark Rober! I Built You a Computer! - Double NAS Build
33:17
Linus Tech Tips
Переглядів 3,5 млн
Water Cooling a Laptop for $30
35:06
Linus Tech Tips
Переглядів 2 млн
Result of the portable iPhone electrical machine #hacks
1:01
KevKevKiwi
Переглядів 7 млн
The power button can never be pressed!!
0:57
Maker Y
Переглядів 9 млн