RISC-V is here! Framework 13 news!
HTML-код
- Опубликовано: 11 июл 2024
- Framework just announced a RISC-V mainboard! This is HUGE news but what does it mean? Let's talk about it!
frame.work/blog/introducing-a...
Want more exclusive content? I post videos over on floatplane! You can also get access to a special chat for floatplane members where I hang out and answer questions: www.floatplane.com/channel/DM...
👕dmstech.store/
🐦I'm on twitter! / dms3tv
🎙If you want to join the discussion checkout our discord community:
--- discordapp.com/invite/hgYsgA7
0:00 intro
0:22 What is framework
1:09 Floatplane plug
1:23 RISC-V mainboard specs and details
2:22 Operating system support
2:36 RISC-V benefits and drawbacks
5:33 Open source Framework 13 chassis CAD
5:54 Factory seconds (B-stock launch)
7:25 More content coming and outro 
Наука
I never thought I will watch a RISC-V video from DMS
A surprise to be sure, but a welcome one.
Nice to hear that RISC-V is coming.
I love just how much our interests overlap...
yes I have noticed that myself.
This laptop keeps getting better and better. I hope Framework will eventually launch an ARM motherboard too.
I would absolutely love that
Specifically an ARM chip that can compete with the likes of the Apple M-series, like the Snapdragon X Elite. I'm not interested in a phone processor. 😊 But an open source RISC-V CPU that was capable of the same performance would be even better!
@@mdfricklet's wait for benchmarks to see if the first gen can actually compete with apple! Frankly I doubt it but I have high hopes for 2-3 years from now when I plan to replace my xps 15
I think we will see an ARM version, especially since you will be able to have the choice of Windows or Linux. Imagine the power efficiency!
Could you imagine a Snapdragon X Elite Framework laptop? Running an aarch64 Linux distro on it?
You're a fedora person too? I would love to see a Linux hifi video. Things like what apps/services you can run, how to get bit perfect playback, best settings for external dacs, etc.
Some of the Linux hifi issues I face are things like the option in Fedora for "analog stereo" Digital, and ProAudio and I've never found a good explanation of what those are and how they change the digital signal going to my dac. Also hopefully a video like that would light a fire under the devs at various companies to natively support Linux. The Tidal flatpak won't work on my Framework 13 but
does on desktop and things like the Qobuz downloader only works on Windows/mac so I have to download whole albums track by track.
Also a note on what you said at 4:40, that's speculative execution which is a massive vulnerability in x86 CPUs and Apple silicon. It may give more performance, but IMO I'd rather just have a really good baseline performance if possible.
Well audio in linux is currently in big switch to pipewire, Fedora is one of the first distros that already using pipewire. About pro audio vs multichannel. As far as i know this obsolete because with pipewire. I use MOTU 624 with fedora and for channel mapping i use virtual sink configuration and its low latency like jack. Also you can manage pipewire over wireplumber, there is already simple GUI. And its realy cool, its like simple virtual console for mixing. And its not just for audio :) But yes currently its more for linux power users.
Speculative execution itself is not the problem, I mean, there's some simple forms of speculative execution that are fine, the problem is when things get complex and security is pulled aside just to reach performance in that complex situation, that's why every modern CPU have that vulnerability
Open Source laptop running on Open Source processor with Open Source OS, that's the kind of utopian future i would like.
didn't expect a RISC-V video from DMS and learning he's a linux user at the same time
I was a tech-nerd long before I was an audio nerd haha.
I do want to do more tech-focused stuff on the channel too.
@@DMS3TV make a video on selfhosting audio on a local server, like an old pc running linux.
that would be cool
@@AIC_onyt actually been posting a mini series about that on floatplane, but I'll probably make a nicer-shot version at some point
@@DMS3TV cool :D
@@AIC_onyt old pc ? Nah ... for just audio services like volumio etc. rpi or power efficient pc are better choice. Just add hat with toslink a you are good to go :) same for navidrome. But if you make hypervisor from that old pc and than install services in vms or containers. But not everyone is interested in self-hosted suff.
I'd love to see a Framework with a Snapdragon X Elite
I would be thrilled
Although the AMD version is already killer, especially on linux.
why, it is an overpriced joke
why inferior ISA?
Now this is exciting news! I really didn't expect a video like this from you, and I didn't know you're a Framework and Linux user. Three really nice surprises!
i kept staring at the thumbnail for a few seconds 😂😂
Linux users does exist. I’m one of them 🛸😄
Risc V and DMS that's Exciting ❤
I like that Framework is branching out beyond just Intel and AMD. I was wondering if it would be ARM or RISC-V that got there first, but considering their open nature I'm not surprised it was RISC-V. Since it's open source and people can mess around with the architecture a bit, I think it would be cool if someone did a laptop where it had both an ARM CPU for everyday low power tasks like web browsing and watching RUclips, but also had a RISC-V CPU for more compute heavy tasks like video editing, gaming, or 3D modeling. Basically an ARM based laptop for good battery efficiency, but with a RISC-V co-processor for when you needed a little extra processing power.
Sick! I'm very excited to see this!
I was hoping Framework would launch an ARM version, but this is even better! Long live RISC-V!
This is really exciting!
DMS, sorry to be off-topic but you've got to know this! I just tried Alternative A2DP Driver and it works on my Sony WH-1000-XM4 in LDAC, 24 bit, 96kHz, stereo, encode quality high! If you've not heard of Alternative A2DP Driver, that's a replacement Bluetooth driver for Windows 10 and 11 which unlocks LDAC, AAC and SBC Bluetooth codecs, thus enabling your higher end devices to be used on a Windows computer with their full audio quality. Before this, I was only able to drive my Sony WH-1000-XM4 on Windows in the only Windows supported mode - 16 bit, 44.1 kHz. More people need to know about this. You've got to make a video! :) The software comes with full free trial and then it's a one-off purchase of $5.99 or $8.99 but please don't think I have any connection to the developer. I just tried this now and saw you posted a new video so I wanted to share this with you and the audio community. If you have any LDAC device, try it out for free!
Wait a minute … how does it sound. Can we change the headband? Does it respond well to EQ. Where do you stick the tubes. 😂
'...a fully customizable computer that you can make last for a long time' has been around for a while, it's called a desktop pc :P, but yeah, in terms of a laptop I guess the closest thing we have atm are some old thinkpads, this does look sweet.
Cool to see more content outside of headphones,
I hope RISC-V becomes a competitor for x86 and Arm.
Great video!❤
Finally, a real reason to throw FreeBSD on my framework 13 lol
our interests are the same, love it
Being a dev kit, I would guess once it becomes a full on consumer product prices should drop, and that is the main point in risc v if you ask me (price to performance)
laptop of Theseus - I like that. planning to by the RISC-V version when it comes out
Framework and RISC V in one video, let's goooooo!
I'm excited about a fully open source computer tbh
This is amazing!!
Not as amazing as you
@@WyattOShea thank you
@@WyattOShea hope you have an amazing day
@@karomba123 No u :)
And EXPENSIVE. 😢
The Umami Shirt!
What CPU architecture will be the first implemented in graphene? 6502, z80, risc-v?
With Cooler Master making a case for the framework i think it would be cool having some small portable RISK-V machines. Not sure what if it would be better then a laptop running it but still neat.
It's kind of a waste of a good screen for such a slow chip, I love to see partners making boards though.
you could put that RiscV board in a Cooler Master chassis that Framework worked with, hopefully
Risc-v was designed for the intention of making it easy for CPU designers to develop niche chips without the burden of not having a software ecosystem. It's versatile enough to be used as a general computing platform though it's not going to measure up. It's really expensive for a chip to be manufactured at the most efficient fabs and a niche chip design is not going to get produced in scale to compete in performance.
No wants RISC to punch at its weight class. We want RISC for no hardware backdoors.
what makes it expensive to produce? fabs don't care what to apply to the silicon
@@D9ID9I There's limited production capacity for smaller transistor "node" sizes and the costs to manufacture chips accelerate the more transistors there are.
@@DaveAxiom this has nothing to do with risc-v specifically. It can have as many transistors as any other chip and order volume depends on the wish of big players to bring it to the table.
@@D9ID9I More transistors means more computational power. Products need demand too before they are built.
For me, handheld gaming PCs are what I want to be a bit more efficient. The Steam Deck is definitely getting there, but the 7840u in a handheld is such a powerhouse. And you're able to get them for fairly cheap considering the full price of a framework laptop, just with not all the open source and repairability. I'm hoping there is a merging of the repairability of framework and the more compactness of these gaming handhelds. And I know valve is doing that, but I also mean for some of the more powerful devices, like the apu that DMS has in his framework laptop and the ability to swap out RAM, like with what we are seeing with CAMM 2 modules from Computex this year.
I wish there was more products like this. I don't own a Framework laptop. but my phone is Fairphone. It is build with exactly the same goals and principles. In case of the phone not so much upgraded, but repairability. Fairphone does not release hardware upgrades very often, but the spare parts are available and they are very cheap. And software support is very long term. Which is amaxing in the age of planned deprecation.
I see that you've seen material about branch prediction in the CPU pipeline, but I think you didn't quite get it even though you were close. CPU is a very complicated scheme and in base they are just transistors with 0s and 1s, so people make blocks that can transform input of 0s into 1s etc. These blocks are combined into the processing units and usual CPU core has many of them. Average x86 cpu has multiple ALU - Arithmetical Logical Unit that can work with if/elses and basic math with integers, FPU - Floating Point Units that work with fractional numbers.
These are the units that were used in the first CPUs alone, but as computation got more complex and bigger, wider, deeper, humanity faced a need to work with larger amount of data from memory and storage, so ALUs and FPUs were taken as a base, so called back-end and new set of blocks was created - front-end.
Front-end moves data from RAM, loads it into CPU data registers that CPU use to calculate whatever is in the register that invloves a lot of microoperations nobody wants to work with because it is all binary and uses hexadecimal system for addresses, like where to pull data from, what to do with that, what to calculate, where to move the result etc and doing it step by step with all the microoperations from the scratch waster too much time, so front-end got a special pipeline and simplified instructions that consisted of smaller operations.
Pipeline consisted of instructions that were converted into operations that do all the stuff with data and using ALUs and FPUs. So here is RISC and CISC come into play since CISC had Complete Instruction Set that included whole lot of them and RISC and Reduced Instruction Set, so on, less instructions, but they performed same operations of load-store-execute.
But pipeline, as it's name suggests, is long and consists of both data and instructions, so it may be inefficient if we do the same microoperation with different data, so there came microoperations cache for faster access to them, like if we need to move data from register A to B, from C to D and from H to F, we will just use microoperations load and store from cache freeing pipeline from them instead of sending: 1) load from; 2) reg A; 3) store to; 4) reg B taking 4 places of pipeline instead of just 2 with just the data needed.
And here is where Branch Prediction I started with comes in. Branch prediction lives in the front-end of the CPU and is responsible for predicting what microoperation will be needed to prevent clogging of the pipeline resulting in faster execution of needed instructions. It works based on the algorhithms Intel, AMD and arm are not willing to share, but in the example if we try to calculate pi number on the cpu we do a lot of the same operations, but get more and more data, so instead of doing loading, storing, loading storing after every new digit counted, branch predictor may intervene in the execution and let that fill in all registers and move digits in the batch.
And frame generation on GPUs works differently based on neural networks, but that's too long to describe. Just the fact - they are different. Ah and also, all CPUs, GPUs, NPUs, TPUs and all kind of Us and even memory controllers use ALUs and FPUs one way or another as far as I know.
A lot of this video makes it clear the author doesn't know what they're talking about
I think ARM is where it's at, if you examine the offerings from the big cloud providers. The cost vs x86 is much lower, and it has full Linux support.
What i'd like to see is a DAW that supports riscV
Ever thought about making a high quality audio expansion card for framework laptops?
It's occurred to me a few times yeah
Presumably would be running open source firmware? In which case I'm in.
So, now you HAVE to co-design your own dac -card with 3.5 and 4..4 for the framework port.
Altho, i think you already are.
funny that I basically already own this device. I've had the visionfive2 which has the JH7110 for quite a while now.
Will you make a video about Music/Audio on linux, could be interesting for people who have not the best hardware and want to move on from Windows?
Will snapdragon x elite and OLED display come to framework?
Did you fully switch from your Macbook to the Framework laptop?
I'm still using the Mac at my office for work but that's it.
No way. Same interests down to even the OS.
We need a SD X for Risc-V
Nice line. The laptop of thesius. 😂
Apple used to be a big proponent of RISC architecture but then they abandoned it for some reason. Now they jumped on the ARM bandwagon. Anyway, it is interesting for it to come back to spotlight again. It also brings memories of SGI workstations and Amiga computers. 🙂
Power vs performance was the reason Apple dropped PowerPC to go for x86. I remembered the keynote where Steve Jobs in person gives this as a reason.
ARM is RISC
Because PowerPC got stucked in performance, Apple didn't abandoned RISC at all like you can see but the only serious platform that works well in that moment was x86
PowerPC and Arm are both equally RISC. IBM was making very fast PowerPC chips but they were also power-hungry. Fine for desktops but not laptops. Motorola was making low-power PowerPC (and still are), but their main customers didn't need them to be any faster. Apple bought a startup (PASemi) that was making very fast but low power PowerPCs, but that was in 2008 after they'd already made the switch to Intel. They set the PASemi people to work designing Arm cores and chips instead, but they could just as well have stuck with PowerPC if they'd wanted to. Intel has also been pretty much sucking with high power consumption in the Pentium 4, but a semi-rogue team in Israel went back to the Pentium III and improved it and got a big power/performance breakthrough with the Pentium M / Centrino / Core series, which Apple latched on to.
Oh hell ya, DMS likes Linux.
I am so exited for the RISC-V main board, it lacks some repair ability because the soldered RAM and storage, but I am so here for it. We just need to get a little bit more development for a faster RISC-V CPU, and ideally LPDDR5CAMM2, and we're good.
Hello fellow RISC-V and umami enthusiast...
I'd love to watch n actual showdown between the AMD Framework and the Meteor Lake Framework.
hopefully it has a UEFI bios, getting really tired of uboot and serial tty’s
recently developed interest in memory-safe architecture, is it RISC-V memory-safe?
so you like
1. audio
2. phones
3. computers
4. guns
just like me frfr
I have had a FrameWork 16 since May 1st. Running Linux Mint.
Interesting why you didn't turn on to let see how it performs?
They just announced it (which is why I made the video) the boards haven't shipped to buyers yet.
More framework content plz 😅
I have a Gen-1 Framework 13. It's a cool machine and cool idea. It sounds like RISC-V is for devs and tinkerers, not for you main system or general computing. How can we improve the laptop speakers/sound on the Framework?
I use a custom EQ on the speaker-output only. There's a resonance in the mids (I think between 500-600hz but I forget where exactly) and the upper treble needs a boost.
It's probably a better idea to get the Milk-V Mars or Starfive VisionFive2 if you want to test the JH7110. This CPU is incredibly slow and mostly useful as a development board to try out RISC-V. It's not really something that makes sense in a laptop. As a comparison, the speed is around the speed of a RaspberryPI 3.
But overall it's still cool that the Framework Laptop as a platform allows building such a thing.
i want a good dac module for this laptop
We need an apple silicon/snapdragon oryon moment for RISC V...
I'm given to understand CPU instruction decode complexity contributes more to the RISC advantage over CISC than ease of branch prediction. The primeagen did a good video on this 'X86 Needs To Die' with Casey Muratori.
RISC-V is less encumbered with intellectual property BS than x86 or arm but it can still have some proprietary stuff tacked on to a given SOC. For more detail checkout Andrew Katz's talk: 'Free and Open Source Licensing and RISC-V'
OH MY FUCKING GOD
Review the AR-5000!!!
Wow, didn't think u would make a laptop video, that too on framework, and u are a linux boy as well, and that too Fedora
While I LOVE the concept of Framework launching a JH7110 mainboard, I've got a StarFive 2 (which uses the same CPU). The JH7110 benches out somewhere between a BCM2837 and a BCM2711. It works for VERY light web browsing and word processing. It might be "adequate" for RUclips, but really, you're boiling down to users having a negative user experience. I'd much rather see them go after SpacemiT's K1 (from the DC-ROMA Laptop II) or something equivalent or better. I love the RISC-V architecture, and really want it to get a fair shake. Which means I don't want the whole ISA to be seen as inferior when the problem is with an older, underpowered, development CPU.
The cores in the K1 aren't any faster, it's just got 8 of them instead of 4. The main advantage of it is it implements the latest RVA22+Vector ISA specification. And it's a brand new chip, which would increase the challenges of implementing a board to go into a Framework computer, surely delaying the launch. I'm sure once this one has the bugs shaken out there WILL be much faster RISC-V mainboards, maybe one with the K1, but maybe skip straight to SG2380 which leapfrogs over Pi 4, Pi5, Rock 5, Orange Pi 5 to low end i7 speed, and with 16 cores too. That *could* hit by around this time next year. Meanwhile, this board will hopefully mean the software is fully ready to go at the same time.
I have been very interested in buying a RISC-V SBC devboard for some time now, but have yet to see one with 16Gb of ram. If this is forward looking tech they better have the bare minimum specs for today. 2, 4, 8Gb main memory is 10+ years ago specs. of course that's just my opinion.
The HiFive Unmatched has had 16 GB as standard since around May 2021. It has essentially the same CPU cores as this mainboard, but different SoC. The Lichee Pi 4A has also had a 16 GB option since this time last year. Also, you must have missed the SG2042 SoC on the Milk-V Pioneer which has 64 2.0 GHz cores (same cores as the Lichee Pi 4A, similar to the Pi 4's A72), 64 MB L3 cache, 128 GB RAM, 32X PCIe Gen 4 -- that's been shipping since October, I think.
DMS! Thanks for posting this video
I'd love to see a RISC-V processor that can match up to Apple Silicon. Used to love Thinkpads for the repairability, but fanless is really hard to give up. Definitely want to get a Framework 13 at some point, if only to support open hardware initiatives.
Sidnote: great to see you branching out on your content DMS! This is good stuff.
edit: RISC-V is an ARM-alternative.
RISC-V and ARM are competing instruction sets; you choose one or the other.
@@charleswilkin2521 Ooh, thank you for the correction. Edited the comment!
How much faster is Risc-V than the Risc of Arm?
The question doesn't make sense. Both RISC-V and Arm are instruction sets, not CPUs. A RISC-V CPU with a similar design to an Arm CPU, made on the same chipmaking process, will have basically the same performance. The fastest RISC-V core you can currently license, the SiFive P870, is similar to Arm's Cortex-X3. It was announced in October and will work its way through into chips and boards you can buy in 2-3 years. The CPU cores used here were announced in October 2018, and are similar to Arm's Cortex A55 announced in May 2017, and shipping on e.g. the Odroid C4 SBC in May 2020. Arm is, obviously, a few years ahead of RISC-V (they've been around since 1985, after all) but the gap is narrowing quickly.
fedora for life
even though im switching to nixos soon
Why the change?
If this catches on, I am buying one instead of my arm macbook
You'll want to wait a couple of generations as this one is about the same specs as one of the last PPC G4s or Pentium IIIs circa 2004, though with 4 cores and 4x the RAM. By this time next year I'd expect a new board with early Core i7 speed, and 16 cores. That's still far from matching your M1, M2 etc, but it will be usable for what we consider normal tasks today. 2026 or 2027 will see at least M1 performance arrive in RISC-V. All this stuff is in the pipeline, from experienced and competent people (e.g. the chief designer of the M1, no longer at Apple), with serious financing, but it takes time.
Someone need to integrate these RiscV cores into a CPU and there is none. We have some SoC and micro controllers with it, but no desktop nor server class around to make a product or two. Also, as long as there is no Windows for it, there is no desktop anyway as Linux is not a thing in that space.
i prefer Risc-V on ARM because its kinda the same but Risc-v is open source
Let me know when it can run Linux Mint. 8o)
People talk about the initial price, what they don't talk about is the hidden fees with others. You can fix this when something breaks. By the way DIY pc building costs more than dell and others. BUt you keep upgrading the cpu or whatever over years, and can fix things, and the parts are better. You shouldn't even need to explain this. Take a part a dell or even a macbook (you can) the compare. I'll put in a nvme drive. This is 100% my next laptop (probably the 16 inch normal version though but framework I mean).
It will help to get more influencers talking about this computer. Tell your friends.
I love my passively cooled, completely silent Macbook Air M2, and I'm very much looking forward to something with similar performance and power efficiency coming to the PC market. I don't necessarily love all of the decisions that Apple makes and would prefer to support a company that is more pro-consumer, but the performance and efficiency deficit between the Apple M Series and any x86-64 chip from the likes of Intel or AMD is simply too great to overlook right now. If Framework were to release a main board with a RISC based CPU that is on par with the M2 in terms of performance and efficiency, is passively cooled, and has good software support from the Linux developers, then I am sold.
Yeah, this is the futture.
How did this happen before an arm main board? Oh yeah, cause you don't have to license anything.
I like the move to Risc-V, however I wished they had created a stronger special Chip for Framework. The mentioned Chip is... not the strongest one. I think it's a bit weaker than the Raspberry Pi 4 CPU.
Yeah, though in this case it's mainly meant as a dev-board
The U74 is similar to Arm A55. Both lie somewhere between the A53 (Pi 3) and A72 (Pi 4), or something around the last PPC G4s or Pentium IIIs. A55 boards have been popular in recent years as while they are a little slower than A72 they use a lot less energy / run cooler. Expect to be able to buy a similar RISC-V mainboard with around Arm A78 / early Core i7 performance and 16 cores this time next year. Look up "SG2380". Hopefully this board existing will mean they've sorted out how to make RISC-V work in the FrameWork ecosystem, all the software is ready to go etc by the time the next gen board hits.
DMS, are you ok?
You can't put a laptop on your ears.
Also, that room...
Is it done? How much is hidden by the camera framing. #Studio tour when
Conceptually this is cool, but the JH7110 is too slow for any sane person to use as a main board. They are 1/5 the speed of an Intel Core Duo from 2008 in single core and 1/3 the speed in multicore. Another metric is 1/10 the speed of the similarly priced Arm-based RK3588's.
Potato
This is really cool. Although RISC-V isn't quite ready yet (even Arm is just barely ready in terms of software). I'd love an X Elite Framework
Nani
Dear presenter, RISC-V core IP is free not the ones from companies like Starfive, T-head etc. So its not free and not fully open sourced. Closed sourced CPU like the ones used in Framework Risc-V board JH7110 will have patent control as they still use HDMI IP from others, GPU IP and VPU IP. And It is just another Architecture like x86_64 so there won't be a huge benefit in heavy loads. As you can already see it from Qualcomm X Elite chips reviews and benchmark. I am Arm and Risc-V Board support maintainer so just want to correct your presentation.
Risc-v was designed for niche computing but if the base specification is covered in the CPU design companies you listed, that would mean that there are substitute products and a developer wouldn't be locked into one vendor.
..wait, that's it? I want more. Is it good? Software availability? C
So, very much like Linux, extremely cool and I love what they're doing, but something that's gonna need to cook for a while longer before I can comfortably jump in. No amount of "everything works out of the box" from people that have the free time to tinker can convince me otherwise.
I'll make the jump eventually, just not now.
Would be awesome to see a RISC-V coprocessor in the Framework 16 expansion bay.
Importantly the JH7110 appears in several manufacturers' products. It is fast and inexpensive enough that it would make sense to get a load of them as a compile farm as the VisionFive2 reference board is available from many vendors. JH7110 PCIe host support was recently accepted to Linux -next for 6.11 release. I don't think it makes sense as a daily driver for graphical UI OS but you could if you wanted to, it would just be like that 233MHz HP Pavilion Desktop from history that some of us might have had and got done quite a lot of interesting computing with, or maybe a little faster than that. Even better, the JH7110 part itself is available in single quantities and you could design your own product based on it, which is sort of impossible for any of the other CPU architectures that Framework laptops have featured until now. For those people that are fixated and raw performance numbers compared to the newer Apple silicon, you're missing the point completely.
While the CPU core's architecture is indeed very similar to the original Pentium (MMX actually) in that 233 MHz Pavilion, do bear in mind that it's got 1) 4 cores not 1, and 2) it's running at 1500 MHz not 233 MHz. It's more accurate to compare against a very late Pentium III or maybe the lowest end of Core 2 Duo (e.g. original Macbook Air). It's also got a reasonably modern GPU, so don't forget to add a Rage 128 or RIVA TNT or something to that P3. Pentiums didn't have GPUs and used the CPU for drawing. RISC-V with around early Core i7 performance is coming late this year, and quite likely on a 2nd gen mainboard for the Framework by this time next year.
Gentoo: am i a joke to you?
i own the framework. the laptop feels flimsy and the screen is terrible quality
the trackpad also is barely functional imo, even with the driver bundle
dont get me wrong, i still love it, it just has some issues
So, cool...but also nothing about RISC-V makes it inherently better than x86, far as general compute capabilities go.
Been following the development of RISC-V / ARM vs x86 for a few years now and don't see it taking over the consumer market any time in the foreseeable future.
RISC-V being open source and less complex than x86, while being highly extendible, lends itself well to any narrowly scoped use case, but you'd need to do something like Apples Rosetta compat layer before even considering making this a mass market thing.
You're not wrong, but this is what is already happening with the Snapdragon X Elite. That solves one half of the equation, but the community needed a chip to be excited about before the effort would be put into ensuring all-the-software is also compiled for ARM.
I have a Framework, but also run nix-darwin on my MBP where I'm using mostly what I would think of as "Linux software" that has been built for aarch64-darwin, but also aarch-linux now thanks to the excitement the M-series has brought to a community that has no monetary incentive to spread their after-hours support even more thin. A lot of that is also thanks to lots of modern replacements to common utilities being built in Go and Rust which make the extra builds pretty trivial, too.
@@mdfrick I agree. It’s a chicken and egg problem and that’s *slowly* being solved.
But this will run head first into the same thing everything does nowadays. MSFTs extreme back compat legacy and the literal tons of code written even to this day that uses platform specific “hacks”, ex video games.
I’m not denying that the problem won’t be solved over time, but will it ever be solved fast enough to arguably make x86 obsolete? Idk about that.
Intel and AMD aren’t just sitting on their butts nowadays. It would take quite a storm of events to dethrone x86 conclusively
@@aj0413_ I think we agree, though I'm optimistic that at least for my use case (pretty basic - mostly browser and terminal, not video editing, gaming, etc) it'll be in a pretty good, usable spot in 12-18 months. (ARM, not RISC-V)
I've been pretty happy with the progress AMD has been making and think they and Intel will continue to do so. So bury them? No, of course not. But real competition is a good thing! It's also possible Intel/AMD run the desktop/server market and ARM/RISC-V take more of the consumer portable markets thanks to good-enough performance and stellar battery life.
This video seems off.
how so?
Seems ok to me.
Frame generation and branch prediction are nothing alike, I'm not sure why that comparison was made. And no, what's limiting processors is not CISC vs RISC, the decode portion of chips are fairly small, it's stuff like instruction scheduling and the like. Don't forget x86 is "RISC" internally and has been for decades. There's also the issue of RISC requiring more instructions to perform the same work, so CISC is essentially compressed machinecode that is decompressed-on-the-fly to get both small code size and fast performance.
The MacBook for Haters has arrived !!!!!
As a dev I'm not sure it's good for FE work
too expensive so it's a no-no
It's a dev kit right now so not meant to be accessible for the mass market yet
@@FunctionallyLiteratePerson i mean this brand in particular is expensive for what they offer
The cost makes a lot more sense when you factor in repairs and upgrades. Much cheaper than buying a new laptop every 5 years
Overpriced bad spec laptops, i love it!
RISC-V isn't going to useful and competitive economically or in performant for general purpose use in your laptop for at least a decade, probably 15 years, if it ever happens.
Of course, given the zero licensing costs, its uses for embedded devices are vast, and save just enough money to make it very much more competitive for specialized uses, as he says. But even that won't likely happen for 10 years, mainly because of momentum.
So please, stop with the hype. It's embarrassing.
Make that 3-5 years, not 10-15.