11 years and it has yet to produce a phone, tablet, laptop, or desktop PC. What went wrong?
>>108770766>no alu flagsinto the trash it goes
>>108770766Either the entire computer is open architecture or it's pointless.
>>108770766It's in all kinds of shit that will hit in the next two years. I can't be specific, unfortunately. There's somewhat of a revolt against ARM that's taking place, and while tons of stuff has ARM cores in it now, RISC-V is sneaking in almost unnoticed, and my uninformed reading of the tea leaves indicate that this is a long term bet, and eventually the ARM cores go away, except for top-of-the-line stuff where the ARM IP licensing cost x $multiplier is simply passed along to the customers that want/need it.
>>108770766I searched "RISC-V" on youtube and the autocomplete showed moon runes so I think it has a future>funky beat and beeps>welcome to another video.. from explain-ing computers dot com
>>108770827but isn't that only true for embedded stuff?
>>108770766It evolved pretty quickly actually. I watched it go from unable to run linux to running a basic kernel to playing games with box64 in like 2 years flat. There's already laptops for it, and a GPU which is likely a scam since it's ran by an Indian called Zeus by Bolt Graphics. It promises to change the GPU game and offer 5 times the performance of the 5090 for a fraction of the price and power. I'm calling bullshit.
>>108770766ARM is lobbying is what's happening.
>>108770857
>>108770766arm didnt produce anything of value in it's first 30 years, cut RISC-V some slack
>>108771026I'm finally just starting to be able to play my x86 PC games on my phone thanks to Proton11 Arm64, and even now it's still a little buggy.Meanwhile people are already playing x86 games on Risc V with Box64.
>>108770886It just has to generate whatever frames it wants fast enough, anyone can larp into graphics now.
Jim Keller is doing RISC-V at Tenstorrent. The storm that wipes out the pathetic little thing you call computing.
>>108771340Shut up bitch
>>108770776What do those do?
>>108770766Nothing. Normie nigger devices is an end goal. Currently it is displacing all other microcontrollers and entering the server space; for non enterprise use the chinese are dabbling with making emulation handhelds with them and there are RVA23 compliant developer SBCs and a few laptop boards if you want to dabble with making devices work with them. We are maybe 5 years from a lot of low end consumer electronics using RISC-V, as they are currently setting the foundations of interoperability.
>>108770887>Arm is telling people to buy their productsSo why can’t Risc-V do basic marketing? Are they stupid or something?
>>108770766isn't it even harder to understand than the standard architectures
>>108770766>11 years and it has yet to produce a phone, tablet, laptop, or desktop PC. What went wrong?Well you, for example. RISC-V tablet has been a thing for years.https://pine64.org/devices/pinetab-v/
Reminder that ARM is literally trannyware and should be avoided at all costs!
>>108773417the thing about marketers is you have to pay them they don't work for free
>>108773317
>>108773441>they have no moneySo they are stupid then
>>108770766>banjo font
>>108770766>What went wrong?It was way too orthodox and only propped up by old professors that would fail any students who failed to bleat RISC is good.Reality: RISC-V is a replacement for the 8051 while the real work is done on CISC.
>>108770766>What went wrong?It's a processor for embedded applications.
>>108770766>no standardized boot sequencenothing will change until this does because arm is already a thing.it's what you get when academics design magical shit, instead of corporations who actually need to generate a return from a marketable product. they just overlook essential shit. >>108770827risc-v is used in subsystem shit because you can just implement it. but that's a long way from cpus and such.
>>108770766>early 2000s>why did Linux fail?This is you
>>108773449>copilot.microsoft.com.jpg
>>108770827auto industry and defense are interested in it because they are getting sick of the licensing fees and USG has kind of put out a bit of a mandate to own whatever they pay people to make and not have cancerous things in the supply chain which has made riscV a lot more appetizing. lock sneed fartin did irreparable damage to closed source
>>108773449Nice one copilot, didn't even say what ALU even stands for. Useless pos AI
>1979: x86 is a dead end, the 68000 is the future!>1983: x86 is a dead end, MIPS is the future!>1985: x86 is a dead end, ARM is the future!>1989: x86 is a dead end, SPARC is the future!>1993: x86 is a dead end, PowerPC is the future!>1995: x86 is a dead end, Alpha is the future!>2003: x86 is a dead end, Itanium is the future!>2020: x86 is a dead end, AArch64 is the future!>2028: x86 is a dead end, RISC-V is the future!>>108770776fpbp
>>108775534Idiot, x86 was dead before 1978.
>>108775668kekKeep coping. x86 is never going away. It's the best instruction set currently on the market and everything trying to replace it can't stop stepping on rakes.
>>108775715Anon, please. IBM didn't pick x86 for being technically superior. They picked the x86 for political reasons.The whole industry and IT field still coping after 30 or something years using the most closed and retarded architecture ever created.>It's the best instruction set currently on the market and everything trying to replace it can't stop stepping on rakes.LMAO.
>>108775768>The whole industry and IT field still coping after 30 or something years using the most closed and retarded architecture ever created.But enough about Aarch64.>LMAOsee pic rel.
>>108775794Oh, and for all you lurkers.If you don't understand what's in this picture, you're not qualified to discuss this topic.
>>108775768>political reasonseconomic reasons anon.
>>108775794>irrelevant instruction added by shintel to becase implementing AES without side channels vulnerabilities cause a terrible performance degradationI never said ARM was better, though, but as a strongly based RISC instruction set the AArch64 versions is more coherent. In the other side those kind of instruction of the x86-64 are individually IP protected (SIMD, AVX, ect), and also I must say that the number of instructions is irrelevant for modern architectures.
>>108775794>ohhhhh nooo, no no, look what "my super convoluted and non ortogonal CISC ISA"™ with years and years of retro compatibility and technical debt can do in super single instructionIf you can tell the difference between CISC and RISC you are in the wrong place.
>>108775935>>irrelevant instruction added by shintel to becase implementing AES without side channels vulnerabilities cause a terrible performance degradation>irrelevant...>also I must say that the number of instructions is irrelevant for modern architectures.I see, you're hopelessly retarded.
>>108775975>risclet copingSure, keep running 14 times more instructions than you need to. I'm sure you'll catch up, someday. Or more likely not since you're doing 14 times the work, kek.
>>108775985Anon, on any modern superscalar processor the instruction are executed in parallel.
>>108775993>Or more likely not since you're doing 14 times the workThat is not how it works.
>>108771352Is that a desktop or a laptop?
>>108775794Can you explain what this means like I'm 5?
>>108776002CPUs can't execute unlimited instructions in parallel, and they can't execute ones that depend on each other at the same time. Of course, you'd know this if your brain wasn't defective.>>108776031>nuh uhKeep coping, riscicle.>>108776093Computer programs are written as lists of instructions that tell a processor what to do. Processors read the instructions from memory and then do what they say as quick as they can, but different processors use different instruction languages.>AArch64This is the 64-bit ARM instruction set. It's used in phones and shit laptops. Each instruction is limited in what it can do, so it takes a lot of instructions to do anything. AArch64 instructions are only 4 bytes, so there's also a limit to the number of possible instructions.>x86_64 aka IntelThis is what most laptops, desktops, servers, and game consoles use. One x86 instruction can do the work of many AArch64 instructions, letting it run programs much faster. x86 instructions can also be up to 15 bytes long, so Intel can keep adding new instructions to make certain programs faster.>that pictureOn the left is a program with one x86 instruction, that loads keys from memory and then uses it to do a round of encryption on 64 bytes. On the right is an AArch64 program that does the same thing, except it takes way more instructions to do the same thing and it runs slower as a result.>b-b-but no it doesn'tShow me a CPU that can run all that in one cycle. You can't, because they don't exist.>>108775975Here's your clean and orthogonal instruction sets bro. This is just the price of simplicity, I'm afraid.
>>108775868yeah but I'm just here for fun doe
>>108775794arm being shit doesn't make x86 not shit
>>108776383So everything is shit, and words lose all meaning. How insightful.
>>108776401you know that's not what i said, inteljeet
>>108776421What instruction set isn't shit then
>>108776356>picAgain. You are posting an artificial example.
>>108776537>using a number is le artificial!!1!11What will the next RISC cope be?
>>108776356>This is what most laptops, desktops, servers, and game consoles use. One x86 instruction can do the work of many AArch64 instructions, letting it run programs much faster. x86 instructions can also be up to 15 bytes long, so Intel can keep adding new instructions to make certain programs faster.If really think that x86-64 is faster because of the complex instruction set you are genuine retard. Ok, just ignore shadow registers, micro-ops translation and trace caching, instruction cache, etc.
>>1087764316502
>>108776605>registers: one accumulator and 2 addressTruly the pinnacle of computing.>>108776596>you're ELI5 doesn't have enough detailNo shit retard.That said, a more expressive instruction set means more complex programs fit in the I-cache, each cycle can do more work without resorting to hacks like macro-op fusing like AArch64 does to make up for their shit instruction set, and register renaming + shadow regs makes up for only having 16 architectural regs.
>>108776705>Truly the pinnacle of computing.yes
>>108775993The Arm arch reference manual is really only half the size because it still has AArch32, and no one gives a shit about A32 anymore. Sadly it will probably be many more years since they drop it (they have to stop updating V8 first at a minimum).
Its used in nvidia gpus
>>108776082You can get Blackhole or Wormhole cards for PCIe. CPU is still x86.
>>108770766It didn't fail, it's about to get extremely relevant very soon, China is dumping a lot of capital into it they need independence from burgers, Europeans will follow later, 2030's will be the decade of RISC-V and the decade of the fall of the burget technological order.It will be glorious.Long live China.Also you can already buy RISC-V SBCs and make you a computer if you want, consumer shit always comes later.
>>108770766It's largely a money thing. Few companies are even bothering to look at it, much less invest in R&D to improve on it. The Chinks are working on it, yeah, but they're basically the only ones. I want it to be cool, because more options are good for everyone, but I don't think I've seen a single board that don't boot off an SD card rather than NVMe or something normal.
>>108771352>Jim Keller, the pussy killer
>>108775534>1985: x86 is a dead end, ARM is the future!they were right since 1985
>>108776356>it takes way more instructions to do the same thing and it runs slower as a result.im just retard cs grad student and even i know this is wrongnot all instructions take the same time
>>108770827Unless its compatible to ARM I don't see it getting adopted in a meaningful scale
