X86 and PC archutecture

Is there any saving grace with the x86 PC archutecture besides backwards compatablity? x86 seems massively bloated, even compared to better and well designed CISCs like m68k. Compatabity doesn't even fully exist anymore with the advent of locked GPUs, forcing the creation of video drivers, with companies like Nvidia, forcing users to use reverse engineering to create open drivers. Also with a number of features explicitly added in because of Microsoft's word, x86 seems hopelessly tied to Windows, with Linux having to conform to their standards, like Secure Boot. And within the coming decade, Intel will rip out 16 bit BIOS emulation out of UEFI, leaving real mode absolutely useless, making the purpose of x86 null and void by being a 32/64 bit archutecture only. This is also ignoring the massive personal freedom issues like the Intel ME, SSM, and AMD PSP, which brings us to the massive duopoly x86 is controlled by, unlike ARM (even though ARM has lots of negatives by itself). Even the Raspberry Pi line of computers seems way better than the cancerous gaming reddit crowd, with their multicolor motherboards and led setups.

Attached: 83fcc3f5c2e5d12df6704774d7f92a2c9e8edec44e8fd657ef4b725acfb915a9.gif (512x512, 579.55K)

Other urls found in this thread:

riscv.org/
riscv.org/software-status/
sifive.com/
lowrisc.org/
twitter.com/NSFWRedditVideo

ARM is the best archingteching without government backdoors and is much more advance then x86 PC. That is why I use it for everything.

Some do have backdoors. It's a wash and you have to look at the chip's data sheet. The beaglebone black is pretty great if you want Amiga-era perfomance, but it hasn't gotten an updated hardware version in years, besides the pocket beagle and blue.

I am a computer scientist and I can tell you familur arketecurs with mutil coloured motherboards are a dangeours combo. You are safe using ARM.

Download Telos 2 and install gentoo

S U P P O R T R I S C - V
riscv.org/
riscv.org/software-status/
sifive.com/
lowrisc.org/

Attached: riscv-logo.png (265x90, 15.62K)

Cool idea, but honestly I don't see any use besides microcontrollers. Call me when they have a full workstation processor.

POWER really seems like the non-pozzed future.

What is your number?

Pick one.

If you had anything more than a superficial grasp on the architecture you'd realize how retarded you are.

Let me get this out of the way so you retards don't have to

Attached: 10928847_1542113199407234_74315882650777205_n.jpg (346x250, 11.2K)

I'm just asking what's the point of x86 today if backwards compatability is thrown out the window? Can you give me some examples of x86 being better than platforms such as POWER? You sound massively assblasted.

Attached: 8cde496f6aaf5e5cdf405d416526d413a6b0993a1215f4643595c63d3546d4da.png (500x747, 178.98K)

That's exactly what you're asking

I don't see your argument.

There's a lot of old, used x86 machines that people throw away on a daily basis. So that's something. That said, I'm not particularly interested unless it's a really old x86 (basically anything before PCI bus).

Despite all x86's technological faults it still offers best performance/price on consumer market due to sheer economy of scale (and some other non-tech related factors).

seems like a good point to me. Whatever arguments people used to make about architecture the x86 ended up being the best on cost/performance ratio etc, so even if you say the architecture was shit, it was still being used on the best manufacturing processes and sold for the best price at least once Intel really became dominant

but where are we anyway ? is it all just kinda fracturing off. Sandy Bridge has been around since 2011 now. eh, i'm gonna leave this post unfinished

I wish a Lisp-ready architecture would exist already. It seems like processors these days are designed for the exclusive use of programs that were written in C.

Did you know they are gutting earlier POWER support from the Linux kernel? PPC64LE will soon be the only supported POWER architecture. Meanwhile i686 will be supported for another decade. You can close your eyes and pretend to be in the land of make believe, but when you open them back up reality will be waiting for you.

lisp-niggers mad because they are on the losing side of history

this.
Although forreal tho, both are trash. We need to switch to Rust

pneumatic computers are the future tbh

One day, the sun will shine again, our sons will write Lisp in peace, free from the oppression of Oracle and its Unix minions.

Can you tell me more about Rust? I see it mentioned on /g/ all the time and I'm curious to see how long my attention span really is.

Rust is a shit language that tries to overcome some shortcomings of C and C++ by using a convoluted syntax that is as unintuitive as possible, to make sure the only thing that ever gets done in Rust are rewrites of existing software, and not original programs. See the Rewrite It In Rust meme, it sums up the philosophy quite nicely.

On the other hand, Lisp is an extremely powerful language that doesn't need you to remap your brain, its syntax is as easy as it gets, without falling into esolang territory, and most importantly of all, Lisp has survived 60 years, it has seen languages be born and die (like ALGOL 68, or BASIC).

False, nigger. Lisp is making a big comeback in AI, because it is an amazing language for programming neural networks. Not only is it making a comeback, researchers are thinking of actually implementing large-scale sillicon plants for Lisp CPUs, for extremely fast AI computing.

Attached: 30.jpg (1200x984, 309.63K)

This. Languagelets on suicide watch by 2020, year of the Lisp desktop.

VM's running on a Zen chip are the best available today. Pi is bullshit, so goddamn full of holes. Cool for retro gaming, though, if you want to have a compromised box running in your house, hell why not? Your satellite/cable box(es) are a base station for the NSA, why not add a couple more?

((( )))

lisp isn't bad, you're just a shitty represenative

source?

Art style looks like an artist called "Incase"

x86 is, and always has been trash. It's a prime example of cancer out-marketing far superior and sane systems. It's the burger equivalent of dining out. No, it's eating out of the trash can and fighting angry racoons over scraps thrown out at a meat packing plant.

ARM is just as bad. A few hours researching how to write bootable hardware-level code for ARM exposes its true nature: an uncontrolled trash fire smothering anything useful.

That would be pretty rad for lisp computers to take place, but I would have to relearn everything software wise about computers. Will lisp exist more further in hardware in these computers, or will they still have instructions?

It just now hit me that Apple will be switching to ARM in their desktops by 2020. Nobody will be buying Apple computers until then, and the people that do buy ARM Macs will be using them to develop Linux and BSD software for ARMv8. What type of AIDS could Tim Cook be brewing inside his pozzed asshole for these machines? Will he tank the hardware like he has tanked OSX, Final Cut Pro, Aperture, the Mac Pro, and Mac Mini by shutting out other operating systems? Could he be that retarded? Will the NOT PowerVR GPU engineers, which were poached, run a train on Tim's gaping ass? Will they go after people attempting to reverse engineer their hardware, like PowerVR, so it can be used for something productive? Will these desktops even be anything more than an iphone?

I hope. What's the status on modern MIPS chips?

PowerVR sold the IP to venture capitalists, so it's dead outside of communists countries. I'm afraid she will never come back, Jim.

iMac Pro and Mac ARM Edition are the best things Apple has done in over a decade.

Gay Cook's realized that he can't just do what Steve was doing towards the end by relying on marketers. Somebody's finally let him know that he should go back to performance computing and business again.


Now if they release a touchscreen Macbook, merge the Macbook and the now redundant Macbook Air series, and create some kind of legacy platform they can survive off of like IBM in case the consumer market goes bad for them; then I'd be in great confidence.
New iMacPro should be an answer to the Surface Studio Follow this simple playbook I've outlined and Apple will thrive like it never has before.

Attached: RedApple.jpg (400x456, 45.58K)

PPC to Intel was rushed and so you needed to recompile apps and users had to run legacy PPC apps on Intel with Rosetta.

This time, its been in the works for years. Remember NeXTs multiarch targrting and packaging from OPENSTEP platforms?
Expect to see something similar included in this year's WWDC xCode release...

Attached: Curious_Raccoon.jpg (1920x1280, 404.49K)

Either English is not your first language or you're being retarded on purpose

Attached: f73576be2d62fd3fc170ed9a4b92cb17.png (1280x1659, 1.71M)

Why is compiling shit so frowned upon?

Because proprietary software is written so badly major compiler changes break their builds

Because x86-64 CPUs are still the fastest on the market and are used in the majority of supercomputers?

Because of its unmatched dedication to legacy compatibility while also allowing for modern extensibility?

Because large corporations don't have the same level of autism as the average user of this board and don't fall for appeal to novelty fallacies because their knowledge of computing architectures are completely based on conjecture?

Attached: bearly_understand.jpg (346x346, 24.95K)

Didn't M$ try the exact same "switch to ARM" thing, but screw it up completely by preventing Windows RT from running any desktop software other than Office and locking down the mobo's boot ROM, as a trial balloon for killing non-Windows Store software?

Modern iFags might actually be braindead enough for such a strategy to work this time around, but it still strikes me as too much of a gamble.

It's not like older versions of OSuX or Windows NT haven't done multi-arch the right way in the past, I just really wish somebody would drive a stake through 80x86's heart.

Attached: uC15s.png (1366x768, 136.55K)

Attached: brainlet 2.png (219x250, 30.19K)

Fastest? No, POWER and new SPARC chips are faster. It's just familiarity, like Windows.
The whole part of the OP was that Intel was going to throw away legacy applications nulling the purpose of x86.
Most companies are filled with chinks and pajeets, with kikes at the top. Literally one white man can outperform and create better technology than the fag/chink/pajeet tripe entantra.

Attached: 4b40c54f798142052fee1f18cb23cd855d5831dc90b4d7ae2c0e91e02f9d259c.png (890x656, 688.5K)

Not technically wrong on the latter point, if only because "supercomputer" has for years now meant "scaleout cluster of commodity garbage" rather than an actual scaleup system made with powerful components.

Attached: Processor_families_in_TOP500_supercomputers.svg.png (1024x728, 436.96K)

That is the ugliest fucking graph I've ever seen

This sounds incredibly implausible. Scientists generally cusom build their own apperatus. If you have to write your simulation from scratch, as long as power supports fortran why would power be abstained from on the basis of not having Windows API's / famiiarity with assembly of x86? This seems to me to misunderstand what a super computer is for.

x86 is probably used because the volume of production of x86 allows economies of scale that create awesome horizontal scaling for clustering at very marginal costs.

I hate intel, and would like to see more diversity in the CPU market, but price per watt , price per flop is affordable

?
Nobody runs Windows on performance computers. Perhaps he meant Windows in the application market.

On point.
You can also infer the follow:
Western companies sold RISC Hardware with software platforms that ran on them. Keeping Software in the West but shipping Hardware to China means that unless Microsoft decides to push RISC, no company can put out RISC software for hardware you can't even get because it has no software.

This has changed recently with the wanning of Microsoft in business in the past few years but Linux Subsystem for Windows might allow them to recover in the server market unfortunately.

Attached: 6zjt6aslndo01.jpg (712x960, 101.17K)

Why?

Wow you guys sure proved him wrong

Microsoft switched their business model to servers years ago. They're still one of the top companies on the planet.

Not defending their Kikery. But I cringe when I read this level of conjecture on this board (and holy shit there is a lot of it) It's hard to have a discussion when the people you're talking too was educated by basement dwelling idiots who never go outside

No they don't. They're scientists. They hire middlemen to create the hardware for them. These middlemen have budgets, and they pick the cheapest and familiar processors. These are usually Intel CPUs.

This is difficult for me to explain to you without hurting your feelings, Tim. Not everyone is fucking retarded.

Interesting fact: The only server ISAs other than x86/PPC/SPARC/ESAME to see any serious development over the last decade are from China (ShenWei) & Russia (Elbrus), created primarily by spy agencies over distrust for sabotage from their western counterparts.

Hopefully at some point we'll get RISC-V consumer boards that are 64-bit and as powerful as a Raspberry Pi. Even they're $200-300 lots of people would still buy them. But until then we don't have very many decent options. There's stuff out there like the Talos II workstation that can be run with only free software, and it's extremely capable with dual POWER9 CPUs which are made for servers. The downside is that it's not at all portable, it's a goddamn space heater, and it's incredibly expensive. The main board alone is like $2k. What we really need is mid range RISC machines that are in the form factor of laptops, tablets, and mini PCs. There's currently nothing filling the gap between Raspberry Pi and Talos II workstation that's worth using. Most ARM devices are locked down trash that's generally underpowered and running Android.

I think my biggest gripe with modern CISC architectures from Intel and AMD is that they're slow and draw a lot of power. They're slow in that these manufacturers are cutting corners with stuff like speculative execution and other little hacks and tricks which boost performance but have been shown to be insecure. Then the battery life suffers from all of these hacked on extensions. The battery life is also a huge issue for me. Why the fuck is your average Intel laptop so goddamn power hungry? I can hardly get my 2 year old ThinkPad to last 10 hours on a single charge when doing basic text editing and web browsing. I've tried Windows 10 LTSB, Debian w/XFCE, Xubuntu, OpenBSD w/XFCE, and NetBSD w/XFCE. All of them are about the same, so I know it's not the OS causing the power draw. PowerTOP is telling me it's various Intel chipset components that are causing this. Meanwhile, some shitty $40 ARM SBC can do all of the same tasks and last almost a fucking week on a 96Wh battery like the TrashPad has.

I really want more RISC shit so that I'm not tethered to a wall. What's the point of laptops if I can't get longer battery life than a laptop from 2008 did? What the fucking fuck.

Attached: 1473782132917.jpg (249x237, 31.87K)

This 100%. I'd even be fine with an RPi but there's no portable option unless you build the entire thing yourself. Maybe an ARM chromebook?

Don't be retarded. A workstation is just a powerful desktop computer with lots of expansion options and ports and shit. There's literally no reason why you couldn't make one with an SoC. If you were to bundle the CPU, iGPU, and maybe some small amount of system RAM in there, what's the problem if you can add more RAM, hard disks, graphics cards, etc? It would be totally fine as a workstation. You could use a Raspberry Pi as a workstation if it was all you needed, since there are expansion boards and other accessories that could be used to turn it into a small desktop. Might actually be pretty comfy if you installed a UNIX-like system and compiled CDE for ARM.


So fucking what? Legacy architecture support that obscure and old should absolutely be gutted from the mainline kernel, especially from a monolithic one. But this is free software, so what stops hobbyists from putting out PPC only variations of Linux? Things will only get better once microkernels and very modular hybrid kernels become mainstream as hardware becomes more powerful. Then you can literally pick and choose drivers and modules to load independent of the kernel core and other essential components.

Your argument is literally "oh no, old hardware is losing mainstream OS support and you'll have nothing to run!" Well, take a look at NetBSD ports. This OS runs on toasters. Believe me, there will be communities picking up the slack.


Somebody needs to start 3D printing and selling Ras Pi 3 enclosures in a laptop form factor. They had them for a while but they were trash for kids. We need something that feels like the shell of a business laptop with no logos. It needs to come with a 11+ inch LCD screen, cables to hook into the display, a SATA HDD adapter board (bridged by USB), a keyboard with trackpoint and mouse buttons, and some male to female cables to route the USB ports and MicroUSB port for a battery pack so you can use a wall adapter. Most of these parts already exist. You just need to make the outer casing, and maybe design an adapter board to fit a ThinkPad keyboard. You would then bring your own Pi, battery pack, HDD, and software.

PPC64 support was god awful a decade ago. That's a cute flag.

The Linux kernel is open source. I'm not exactly an experienced kernel dev, but theoretically what stops me from simply forking the Linux kernel and continuing support for PPC? Literally nothing. It might be late on some patches or not get them at all but it would still be more secure than other outdated software, especially for simple home use. Enterprise users are irrelevant in this since none of them are using decade old PPC machines anyways.

Find a flaw, asshurt x86 bitch.

If you're aiming THAT LOW, you know you can just get an FPGA SBC for under a hundred bucks & install whatever ISA you want on it, right?

Soft cores don't come out of the air. The have IP's attached with them and the ones that are open are either Amiga cores or microcontrollers.

Kernel dev isn't gratifying. Good luck finding the motivation.

Not everyone can use a Raspberry Pi, go back to x86 Windows for gaymen, then go back to that Pi. Keep your chest puffed up retard. Maybe somebody will buy into it.

They're slow because they're fast?

Take your autism medication

even if they were, the FPGA would still be slower than a wet week
FPGAs as CPUs is piss in the wind at the moment

sorry for fobbing out

*even if you had an open ISA to play yer games on, ...

More than fast enough for all these "muh core 2 thinkpad" memers

i don t know about computer archetecer what is best archetecer for computer games??

they default to having ssh on accessable by default passwords it doesnt tell you to change

So what're my options for an ARM based PC that I can code on? And no a raspberry isn't strong enough.

Riscv is taking forever.

core 2 is still fast

Beaglebone since VI does not require an X server

By apperatus, I intended software. If you are computing on a super computer, its probably NSF owned, your probably writing some simulation or other in fortran/root etc. You probably give no fucks about x86_64 's domance of consumer computing market , or winodws availability or unavailablity - you essentially want standard fortran + niche tools that probably arn't x86 dependant.

What i am saying is the idea that super computers are x86 because of the avaiability of software for x86 is retarded.

I believe he was saying super computers are built on x86 not for cost / flop but for existance of domain expertise/ familarity. Apparantly, he believes super computers are built by 'pajeets' who can't do anything but x86, and that 'white men' with sparc processers (lol) could do way better (but somehow the are opressed? Reason for the non-Existence of white mans sparc #1 of top 500 super computer is unclear)

This is ridiculous of course. Super Computers are built for performance, and applications are custom built to solve single problems , and don't give a fuck about software avaiability beyond development runtimes (fortran 6).

x86 is used because its cheap, and its cheap because its mass produced. some performance clutsters where being built out of playstations because they were available and cheap, not due to some elaborate oppression of the white man by x86 kikes. This poster is absolutely retarded.

LOL PARENTHESES AND PREFIX NOTATION

You still need software to handle parallel computing, hypervising, and an OS with up to date drivers. Also I said nothing about software. It's just that x86 is a well establish platform, even if it a pile of dog shit.

Applications are NOT custom built. Why waste that time and resource when you can just tweak existing code? Time is money.

porting a microkernel to a different architecture isn't quite as simple as installing a different device driver.

He's sort of orthogonal from the actual benefit of microkernels, in theory, which is that you'd be able to simultaneously run multiple OSs on the same shared kernel (say, MKLinux and OS X on Mach if Mach hadn't been completely bastardized in every actual implementation), much like a hypervisor can today, minus the inflexibility and inefficiency.

What's wrong with that, brainlet?

it rather needs a coup de grace tbh fam

Attached: intel_is_fuck.GIF (516x400, 47.66K)

kek

Attached: _untitled.png (579x601, 433.48K)

I post here a lot.

Attached: 1519857204661.jpg (657x527, 55.83K)

It's easy to port the modules you need rather than everything at once. Microkernels are also more stable and secure, which is the main advantage of them. Not running two operating systems at once. That's secondary. Their modular nature should be used to create a system that's nearly impossible to crash. Modern systems are already pretty stable but not stable enough. I also want the driver isolation that it can offer, again, for security reasons.