X86 preferred by God of Linux

>>>/cuckshed/

I can see why he would say that. If I remember correctly they had to add a fuckton of code to the Linux kernel for all the different ARM based devices they were trying to support. Still doesn't change how ARM processors now days don't have some of the more retarded features that are being included in all new x86 processors such as Intel's ME and AMD's PSP.

Saged because OP should have actually put effort into his post and posted more of Linus' reasoning instead of just retarded consumerist "X BTFO."

ARM's a nightmare if you're trying to get it to do anything out of the ordinary in a compatible way. Virtualization on ARM was such a disaster for so long, and it's such a critical feature these days for complex embedded development, that most companies went back to x86.

protip: nobody gives a rats ass about "muh botnet ME!" in the real world, in the real world, x86 is still the choice architecture for software engineers (because CISC) for normalfags who need more horsepower to play PC games that their dinky little smartphone doesn't have, and finally for supercomputing, where Intel Xeons are currently the most powerful processors in the world

Also how does the "cucklicense" harm you, the user. You're not going to be using developing the software, what do you care if the developers choose to make in less restrictive?

Those Xeon Phi processors look so nice, I'd use them even if they said "BotNet Inside".

i'd rather have a SPARC or DEC Alpha.

The future isn't x86, or ARM for that matter, its RISC-V and GPUs, with PowerPC for the heavy CPU compute applications

...

You wouldn't if Linus was a userland software guy. But a kernel guy has to know his CPUs.

muh baytrail

He may know how it works in an abstract way, but I doubt he knows the exact electronics going on and how scalable or efficient they are.

...

He also used C and made a Unix clone in the 1990s. Unix was obsolete tech in 1970. It was just impressive for the PDP-11, like 8-bit demos are impressive because of the limitations of the hardware.

pick 1

Most supercomputers run it.

...

Most in this case means 99.4%. The other 0.6% run AIX: top500.org/statistics/list/

Show me all the Atom-based phones that get 48 hours in standby.

Logic circuit level? Yeah, he probably doesn't know that....but nobody outside of Intel or AMD can say that either, because that massively implementation dependent.

Are you implying he's wrong? Did you seriously waste a post trying to shoehorn the fact you knew AIX exists? Because last I checked it should be pretty obvious for anyone with IQ above 80 99 percent means "Most" without having to consult a Top500 list.

Thats wrong though, Cherry Trails have over 2M cache

Show me an ARM-based phone that gets 48 hours in standby and has performance anywhere near matching an Atom

Direct in the face.

I think he meant AIX is a unix too. For all the retards saying

2011 Galaxy Nexus. I'll be back in 3 days to post the battery graph.

I believe you though about the standby

How would I be saying he's wrong if I said 99.4% run Linux and he said most run Linux? I mentioned the other 0.6% so someone reading doesn't have to click the link to know it isn't Windows but other ancient OS.

It's got more real cores than any Atom I've ever owned

Cursory searching reveals the following:
cnx-software.com/2015/01/21/antutu-benchmark-rockchip-rk3288-arm-vs-intel-atom-z3735f/
Rockchip RK3288 (aka ARM Cortex A-17) beats Atom Z3735F by 18%.

HOWEVER

cpubenchmark.net/cpu.php?cpu=Intel Atom Z3735F @ 1.33GHz
cpubenchmark.net/cpu.php?cpu=Intel Atom x7-Z8700 @ 1.60GHz
The fastest Atom is actually probably the cherry trail Z8700 which is about twice as fast as the bay trail Z3735F.

browser.primatelabs.com/geekbench3/592358
cnx-software.com/2016/09/28/rockchip-rk3399-benchmarks-appear-on-geekbench/
And the fastest ARM is probably similar to the Rockchip RK3399 (aka ARM Cortex A-72) which is only about 15% faster than the RK3288. Probably a really shitty benchmark though.

Is this conclusive? Nah. Shitty benchmarks are shitty. But it does suggest it to be quite plausible that the fastest Atoms beat the fastest ARM chips at the moment. Also cherry trail Atom has 2MB cache.

hah

It is still actively maintained by IBM fuck face and many large corporates still run on it faggot.

Even though IBM is eventually going to phase it out does not mean it is ancient now, you know fuck all about it.

RISCV is the future. Deal with it. Dot jay-peg.

>hothardware.com/news/antutu-mobile-benchmark-cited-by-analysts-fundementally-broken-heavily-favors-intel-architecture
Did you know Jewtel still use their ICC trick? Now you do.

...

That's 2 years before the benchmark I linked, was allegedly fixed, and intel loses out in the RK3288 vs Z3735F comparison. But it wouldn't surprise me if they're still cheating and invalidating stuff. That's one of the reasons why those quick searches are not conclusive.

POWER8 architecture gets introduced into the consumer market when

Never, because people want gaymen and they trust Intel cause their logo is cool and they have cool commercials.

www2.eecs.berkeley.edu/Pubs/TechRpts/2016/EECS-2016-130.pdf

Shitposting aside, RISC-V is an extremely promising project. According to this paper, the authors' academic RISC-V implementation offers lower instruction counts than equivalent ARM/x86-64 programs, takes fewer instruction bytes, and has a lower overall code density of 3 bytes per instruction versus 3.7 for x86-64, all achieved for an experimental implementation without significant investment from the hardware industry or the compiler optimization side.

Not even Intel could tell you that stuff without having a bunch of different engineers conduct studies and put together a report for a committee review.

We've abstracted out all the low-level stuff. Nobody knows how the entire CPU works anymore, except in abstract terms. As long as each part works correctly, it all fits together.

What Linus is talking about is the instruction set and support. Intel has done a lot of revolutionary work, and they continue to make improvements, so long as they do not have a monopoly. That's not to say they do everything right. x86_64 wasn't even initially designed by them, but by AMD! But there's no overlooking how they offer a complete instruction set for every type of need, and they support all of them with gradual improvements.

Most of the speed and memory benefits in upgrading your processor isn't just "it's faster", it's also much more efficient at making certain calculations. Most of that comes through new instructions.

Linus' job requires adjusting low-level kernel functionality to work efficiently with instruction sets. Of course Intel's is going to be more complete and easier to work with; they've been throwing their money and best talents at it for decades.

He's not saying ARM isn't better from other standpoints, like openness (where it isn't perfect, either), just that it's a complete package that is easy to work with. You guys have done an amazing job of putting words in his mouth.

There is a company making a Power8 workstation machine
raptorengineering.com/TALOS/prerelease.php

Nvidia is adopting RISC-V for their products, dropping ARM in the process.

NEVER EVER

Look at this fag being all reasonable and shit.

Delivering.

Guess that explains why he seems to hate PPC so much in this mailing list argument:
yarchive.net/comp/linux/x86.html
Found it pretty fun to read, actually.