Why isn't x86 dead yet?
>>108589942just go convince some of those "com" kids that x86 is a good name for their group and it'll be dead in a year
>>108589942That which is dead cannot die.
Because it's superiorARM is complete goyware that you don't own and RISC-V is vaporware
>>108589942because anything else sucks
>>108589942x86 is dead. x64 replaced it.
>>108590013>CPU encoding>In the big 2026You still use Winstone Benchmark scores too? Put a tripcode on so I can filter you.
>>108590041>x86 is dead>x86 replaced it??
>>108590072Why do you even care for CPU architecture at all if you wish every computation task would be done by a fixed-function hardware? >What? We don't have accelerators for your particular use case? Have you thought about walking outside and touching grass, chud?
>>10858997268K was a better base than all of them combined
>>108589942it just werks
>>108590073x64 isn't x86. You can cope as much as you want with your head in the sand.
>>108590013>4.0Somewhat surprising results. SVT-AV1 got a lot of optimizations for ARM in the last year or so. I suppose x86 is still king in raw performance, although completely dismissing ARM because of that is inane.>>108590072ASICs are great if you don't care about quality. There's merit in speed, but for archiving and non real-time streaming, you really don't need that.
>>108590072HW encoding is only used by marketing to sell next generation of hardware. the quality between SW/HW encoders is not remotely comparable (10-100x br increase).
>>108589942>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!Face it, it's never dying.
>>108590140I dismiss ARM based on the fact that every release needs a baked-in uboot. Kek.
>>108590041x86-64 is x86 for anyone that isn't a nitpicking sperg
>>108590252Repeating the same line of thinking over and over in your head still doesn't make it correct. Sorry!
>>108589972Every x86 has had built in management spyware for decades now.
>>108590324name a single relevant x86 32 bit piece of hardware released within the last decadethis is like being upset at the fact that phone no longer refers primarily to landline phones
>>108589942what's the text from?
It's dead for any battery powered device.
>>108590190x86 died shortly after amd64 took over
>>108590400If you're gonna be a pedant, at least be right. Every new Intel and AMD processor still supports 32-bit x86, and x86_64 is nearly a perfect superset.
>>108590410yes, that's why x86 is dead. amd64 can do the same and more
>>108590435x86_64 is still the x86 family. It's never gonna die.
>>108590364The fuck are you smoking? I already said x86 is dead. Nobody refers to it as that anymore unless you're a Mac cultist.
>>108590454yes, they do, when you refer to the different ISAs currently in use you say x86 and arm and everyone knows what you say when you call it x86, no one is retarded enough to assume you're implying people are still using pentium 4s
>>108590477No, you don't.
>>108590583>no one is retarded enough to assume you're implying people are still using pentium 4sWe'll make an exception for you.
>>108590213Better than a closed-source UEFI by American Megatrends or whatever with gazillion glowie-blobs.
>>108589942Because there's really no good competition for it, you either get to choose between an anemic ARM SBC with celeron performance OR:there is no equivalent to macsmacs have:>soldered SSD/RAM/WLANdoesnt exist on 99% of PCs>when pajeetOS runs out of RAM it swapfile rapes the SSD to death within 3 yearsdoesnt exist on 99% of PCs>when the SSD dies it bricks your entire currybook because the EFI is stored on the soldered SSD to save $0.05 on a dedicated chipdoesnt exist on 99% of PCs>components serial numbered and tied to the motherboard to prevent repair and replacement (including battery)doesnt exist on 99% of PCs>riveted keyboard that requires total destruction of the chassis to replacedoesnt exist on 99% of PCs>flexgate cables that are so brittle they crack from opening your screen past 90 degrees more than 2 dozen timesdoesnt exist on 99% of PCs>screens so fucking shit they have over 90ms response times (essentially 10Hz)doesnt exist on 99% of PCsAnd so on and so on...Basically while x86 offers worse energy efficiency it makes up for it with user freedom, upgradability, customization, incredible value (ie $50 office PC can be turned into gaming PC with a $200 GPU + $50 PSU).
>>108590609So now you get a closed-source u-boot fork by whoever with a gazillion glowie-blobs, except now there's no menu and no way to install other operating systems. So much better, right?
Legacy compatibility.I'm fine with moving to a different paradigm as long as the emulation capabilities of that platform mean x86-64 code runs just as well.
>>108590607>Heh. Got em!
>>108590609>Here's our device tree blob bro! You can pass that to the kernel instead in-place of any ACPI!
>>108589942>Decades of backwards compatibilityWe have a bunch of machines that nobody knows how we got them or how they work. To figure that out we'd have pay big bux to some specialists. They work. X86 ensures compatibility. They work and will continue to work.
>>108589942Because you can put code from 40 years ago and it runs, thats why.If you are a macfag you can't use code from 6 years ago.
>>108590013no x86 cpu is actualy x86, they are all a risc pretending to be x86.in fact we'd have performance improvments if they no longer were designed to pretend.
>why do I completely misunderstand machine architecture, not know shit about anything, and then want to throw it all out even though there was never an issue and it is working fine for people who know what they are doing so that we can remake the same shit all over again with slight differences that will somehow magically better than the thing we threw out? Because my RUST bootcamp buddies said so.>>108590447>>108590252big if true
>>108590757>t. read a retarded reddit comment years ago that must be true!
>>108590140>ASICs are great if you don't care about quality. There's merit in speed, but for archiving and non real-time streaming, you really don't need that.AVX is also ASIC
>>108590757C/RISC brain rot.. it's called microcode and every modern processor does it. we're not in the 90s anymore, stop talking about RISC, it's an academic meme, it does not exist, no risc-v is not risc.
>>108590779i don't even know what you are referring to but you'd know it if you read any cpu architecture book, which you most definitely didn't.
>>1085899428086 was actually extremely elegant. i386 was pretty decent as well. amd64 is where is all went wrong, but before that point it was actually decent.
>>108589972>RISC-V is vaporwaretpbp and fucking /thread
>>108590799https://chipsandcheese.com/p/why-x86-doesnt-need-to-dieAlso, x86's complex instructions mean you need fewer instructions to do the same thing, so code is denser and fits in the i-cache better.
>>108590782AVX is programmable and flexible so it does not qualify as an ASIC. You can use AVX for PS3 emulation, video encoding, 3D rendering, physics simulation, AI workloads, and so on and so on.An H.264 ASIC can be used to decode and encode H.264. That's it.
>>108590819now show what that instruction gets broken down into by the CPUs encoders
>>108589972>RISC-V is vaporwareThe Chinese are pumping big money into RISC-V now in an effort to wean themselves off of US tech dependencies.
>>108590832>one 512-bit memory load (single operation)>one round of AES on each 128-bit chunk of zmm0 (single operation)Still going to insist ARM wins here?https://uops.info/table.html
>>108590873even R-M instructions are a single uop, what's the risc's next cope?
>>108590819>what microOPslmao
>>108591025I'd say to check it yourself, but you have no idea how so tough.
>>108590102the m68k isa was so fucked that apple stopped using them in favor of powerpc, and motorola even got in board of the powerpc train becasuse the 88k was DOA
>>108590102True but Motorola was a bit too greedy with their pricing, even back in the 6809 era, giving competitors an easy road to the consumer market, while Motorola wanted to be the high end. As it turned out, most businesses were more interested in what had the largest install base, and thus the largest software library, than having the best CPU architecture in their machines. There's another timeline where Motorola had lower pricing and OS-9/68K became the default Unix alternative (assuming Microware stopped being greedy corn jews with their pricing).
>>108589942Legacy software. Turns out it's pretty fucking hard to just start from scratch.Until another architecture can emulate x86 straight up better than an actual x86 chip, it isn't going anywhere.
>>108590013and nine series ryzen to encode fullhd in 1:1 time
>>108590364what is landline phone f type line with network identifier card is not enough for you
>>108592424did you haved a stroked
It is? I have all 4 ARM computers and only one x86 one.
>>108590190Risc v is just a meme for trannies to tinker and dilate.
>>108592476>main PC = x86>laptop = x86>tablet = ARM>phone = ARM>raspi = ARM>orange pi = ARMThank you for reminding me that my x86 devices are actually in a minority. And the tablet isn't even one of those Android tablets it's a Windows 11 one, so it's more of a "proper" PC. The raspi is intended to be a PC as it's the 500+ model.
>>108590177>Lust provoking image
>>108592508Macbook Pro, Macbook Air x2, Mac Mini, iPhone, iPad, RPi4, RPi 5 = 8 ARM computersDesktop PC, T480 = 2 x86 computers
>>108590368AI.
>>108590611This Anon knows.
>>108589942Why don't we design an OS that is less retarded with resources instead of requiring everything have 32gb of LPDDR5X at 8200 MT/s with a 40 core(16 e-core and 8 ultra E-core) CPU?
>>108589972>>108590806RISC-V already is the price leader in IoT and millions of chips are being used. You're probably using RISC-V right now
>>108592476>>108592508you aren't counting the embedded (maybe ARM, maybe RISC-V, maybe others) processors, and your (MIPS) routers, and...
>>108590368Snowclone of Charles II's autopsy report.>>108593899God, I wish my router ran on MIPS. It's just ARMv7. No RISC-V anywhere either.
>>108594217god, intel's new logo is so fucking gay
>>108589942but can your meme arch C40448AB260345AE260305C1E802EBF2 ?
>>108594217how long would it take you to skim 42 bind folders
>>108594265wtf is a "bind folder"
>>108590435Repeating this doesn't make it true. x64 is just x86 combined with 64-bit pointers. That's it. It's still x86 under the hood.
>>108590618These glowies can't even figure out how to redact pdf files from copy and paste, I'm pretty sure we're safe.
>>108589942Because it is ubiquitous, cheap, easy to use and there's already tons of software supporting it. Also, all the supposedly superior replacements just fucking suck ass in reality.The x86 is basically a computing equivalent of simple machines.
>>108590804The 286 was an improvement. The 386 was where it went wrong. AMD64 made it even worse. The 8086 and 286 actually had a lot of reasoning behind the design of everything, but people tried changing it into different things like adding SIMD and making it more RISC-like.
