Face it. x86 is deprecated. ARM is the future.

>>57183741
fuck you apple shill

Sent from my iPhone

As soon as arm gets native x86 code execution it will be the future.

>>57183779
As soon as someone comes out with decent ARM desktop environment x86 sales will drop off and become legacy equipment only used in niche use cases.

>>57183779
Are you implying arm is good?

>>57183741
x86 is shit but ARM's not an alternative. Now MIPS, POWER, maybe RISC-V, those are the superior alternatives.

>>57183821
Which is never going to happen.

File: 1445167755166.png (120KB, 346x250px) Image search: [Google]

120KB, 346x250px

>>57183821
Since when are DEs architecture specific?

>>57183741
As soon as it is as powerful clock for clock as x86 in general computing and not just Geekbench.

>>57183949
They aren't but no one has popularized it yet

>>57184040
I don't know if it's out yet or not but iirc AMD's releasing socketed ARM Opterons

>>57183949
They might as well well be but normies wouldn't know, and it's not like they'll be able to run them on WangBlows

face it arm is deprecated risc-v is the future

File: XeonPhiGenerations.jpg (311KB, 1089x612px) Image search: [Google]

311KB, 1089x612px

>>57183741
Tell that to knights landing doing over 3 TFLOPS of FP64 with only a TDP of ~250 watts

http://www.anandtech.com/show/9436/quick-note-intel-knights-landing-xeon-phi-omnipath-100-isc-2015

>>57184105
>>>/v/

>>57184128
>HPC coprocessor
>/v/
You can't possibly be this retarded

>>57184128
0/10 bait

>>57183741
Well, ARM is a disaster. It's a humanitarian nightmare. But it has fallen from any standpoint. I mean, what do you need, a signed document? Take a look at ARM. It is so sad when you see what's happened. And a lot of this is because of Apple.

File: serveimage.jpg (46KB, 500x327px) Image search: [Google]

46KB, 500x327px

>>57184105

>>57184272
What the fuck does this dead meme have to do with the most energy efficient x86 processor to date?

File: 1465927879669.jpg (74KB, 640x640px) Image search: [Google]

74KB, 640x640px

>>57183741
>early 1990s
>Face it. x86 is deprecated. Alpha is the future.
>early 2000s
>Face it. x86 is deprecated. Itanium is the future.
>early 2010s
>Face it. x86 is deprecated. ARM is the future.
x86 has been "deprecated" like 5 times already, hell it was deprecated off the bat, but as usual, nobody rightfully gives a shit because there's more to computing than jerking off to 20 year old benchmarks and living in the past, RISC on the home desktop is and always will be shit

>>57183846
>MIPS
Absolute trash on the desktop since the R12K.
>POWER
Absolute trash for integer-heavy, single-threaded "typical" workloads. You only like it because it has the IBM logo on it.
>RISC-V
Vaporware.

>>57184420
it was deprecated before it was ever fabbed. its a shit design.

>>57183779
>>57183821
With Chromebooks getting android app support and more manufacturers starting to release arm Chromebooks, arm might have a future on the desktop one day.

>>57184482
>its a shit design.
Then why does it shit all over every single CPU architecture out there?

see >>57184105

File: Intel-Overtake.jpg (125KB, 1137x557px) Image search: [Google]

125KB, 1137x557px

>>57183741
Good luck faggot.

x86/x64 has slain every other processor type outside of mobile for like 40 years.

>>57184482
But in the end it works fine that there's no reason to give a shit unless you're a low-level systems programmer who still lives in the '90s.

x86's biggest hiccups were documented and worked around decades ago, and the modern chips are so fast it doesn't matter anyway.

yarchive.net/comp/linux/x86.html
Here's a related rant from Linus Thorvalds discussing this with mailing list denizens shilling the Itanium and PPC chips back when they were actually viable, it's a fun read.

>>57184516
Marketshare really isn't an argument here, x86 came out on top because it was all that left worth a shit after the Itanium train, dotcom boom, success of Linux and general dumbfuck mismanagement ravaged the high end in the 2000s, it didn't make it there on pure performance and technical merit alone.

Nonetheless, still bitching about this in 2016 is fucking dumb, it's fast enough and it's very compatible, that's what counts in the end.

>>57184533
Same with supercomputing, a totally irrelevant point when discussing desktop computing, and another market x86 inherited, rather than earned.

>>57184516
billions of shekels for R&D, smaller form factors, more hardware options. I'm sure some of those niche arcitectures would fare OK if asus made enthusiast tier mobos for them

The only reason I support the deprecation of x86 in favor of ARM is the thought that I could natively run GBA and DS executables on my desktop.

>>57184610
Just like how you can run XBOX games on your x86 desktop? Oh wait.

>>57184583
None of these are really true, you didn't necessarily need a huge R&D machine to design a fast chip, there were plenty of weirdo architectures that worked just fine in small systems, and "hardware options" was really more of a problem of driver availability than anything else on those systems that all used standard expansion buses anyway.
>I'm sure some of those niche arcitectures would fare OK if asus made enthusiast tier mobos for them
Yeah, and go ask the remains of DEC how that worked out for the Alpha, they even ran Windows too.

>>57184640
>small systems
i meant smaller nm construction

>>57184622
shit

>>57184654
Oh, sure, I'll give you that then. I think even Intel's own darling Itanium was always a process behind the x86 mainstream.

>>57183821
You know that code can be ported and that there are linux distros for arm, right?

>>57183821
They average tech illiterate doesn't even know what ARM is, so it's pretty doubtful to assume they're telling the guy at BestBuy that "there just isn't a good enough desktop environment yet!"

All it would be to them is a cheap piece of shit that doesn't run Windows nor the software they want with it, like drivers for their obscure 2007-vintage digital cameras and ancient versions of Office they see no reason to upgrade

>>57184671
You do know loonix is hitting 1% marketshare on desktops and that no professional x86 software like x86 adobe products can run on it let alone be ported to ARM right?

>>57183741
do ARM CPUs/SoCs even have a standard way of booting an OS? will they even have good, open drivers for linux?
AFAIU, ARM is a mess for OS devs.

>>57184105
Oh look it's this fucking moron again.

3 teraflops at 250W is not impressive.

>>57184713
>3 teraflops at 250W is not impressive.
Name a single angry birds processor with the same performance-per-watt

protip: you can't

>>57184734
>angry birds processor
Oh it's you

Just ignore the shitposter, everyone.

>>57184707
They maintained ports on 68k, PPC, MIPS, and SPARC just fine. They didn't even start on x86 for fuck sake, ARM's not any different in this matter.

It's not happening not because it's impossible, just because there's no demand for it from people who actually pay money for their software.

>>57184734
This fucking moron and that fucking moron are one in the same? Who the fuck knew?

Anyway, no, there aren't any high power ARM chips in existence, but that of course has nothing to do with what I said. 250W power consumption and 3 teraflops is NOTHING. Especially sicne it's theoretical performance, nothing actually tested. Compare it to, say, a W9100 and you get virtually the same performance with the added benefit of it actually being a display adapter and not just a compute card. Pretty sad your fancy Atom favela compute card is barely better at compute than an old W9100.

>>57184707
>free code is portable
>unfree code is not
Are you retarded?
Also not every profession uses photoshop, for scientific work linux software is mostly there.

>>57184742
Just accept that ARM is, always has been, and always will be a huge steaming turd mountain for desktop and enterprise. It has 0 chance of ever toppling x86. Only reason we use it on phones is because of how cheap it is else we would be seeing core-m processors on the latest fruit phone or samshit hand grenade.

>>57183741

FPGAs are the future.

>>57184785
>Just accept that ARM is, always has been, and always will be a huge steaming turd mountain for desktop and enterprise
I never said otherwise. I'm just sick of your shitty meme. We get it, ARM is prevalent in smartphones and Angry Birds was once a popular smartphone game. Now just call it ARM and actually explain why it's not good for desktops or go eat a dick. You're not funny and you never were.

>>57183741
nah x86 is pretty good.

its only weak point is floating point, and generally floating point heavy work loads are for GPUs.

ARM is just good because its power efficient.

>>57184533
that thing is old, itanium is even discontinued i believe..

>>57184845
Pretty sure Itanium's still a thing because of some contract HPE has with Intel

>>57184774
>275 watt TDP
>"With 2.62 TFLOPs of peak double-precision floating-point performance"

http://www.amd.com/en-us/products/graphics/workstation/firepro-3d/9100

>this is better than knights landing that does OVER 3 TFLOPS of FP64 with a TDP of ~250 watts
Are you literally retarded?

>>57184845
It's not yet discontinued, there's still a new release coming next year. It is, however, supposed to be the last one.

>>57184105
from what I've heard, that's not x86, though. at least not normal x86...

>>57184813
>its only weak point is floating point, and generally floating point heavy work loads are for GPUs.
see >>57184105

retard

>ARM is just good because its power efficient.
It's actually shit, ARM processors require more electricity to do the same thing as a modern x86 one needs. Only reason we use them in phones is because they are cheap.

Ever wondered why the zenfone 2 had better battery life than most phones during its time?

>>57184863
>from what I've heard, that's not x86, though. at least not normal x86...
Actually it is. You can run a bunch of x86 VMs using the atom cores. Only problem is integer performance will be shit. Though programs that use AVX-512 instructions will have ridiculous amounts of performance.

>>57184852
It's 2 years old and offers near identical performance, that has actually been tested, compared to your unreleased and untested thing two years later.

Also Tesla compute cards absolutely rape it. They're doing like 5 teraflops.

Knight's Landing is DOA.

File: AcornArchimedes-Wiki.jpg (452KB, 1596x1294px) Image search: [Google]

452KB, 1596x1294px

>ARM has no place on the desktop

Implications.

But anyway, I don't see why /g/ cares so much about this. Are you all Intel shareholders?

Surely taking a minor hit to performance for the massive gain in power efficiency is a good thing?

And it seems two-faced for /g/ to complain about software bloat and bad programming yet also supporting an industry that facilitates (and enforces) it through hardware expansion.

If you went to an office building and said you could cut their power bill to a fifth of what it is now, with no perceptible loss in performance (remember, most workstations now, as always, are essentially glorified typewriters and spreadsheets) then they'll switch, as long as they can run their very minor-draw software.

Home computers as we know them are on the way out. People do not sit down at a desk to use a computer now. They do not to online banking or shopping or chatting staring at a big screen clacking away at a keyboard.

Sorry /g/, but you're not the people the industry is fighting over. It's a majority of normies, workers, a subset of gamers (the smallest of the three by a wide margin), then at the bottom it's you guys buying a new processor every year because you need your anime in 4K.

doesnt amd have an arm coming?

will this begin to make arm hardware more diverse and mainstram?

File: 1473823685123.jpg (50KB, 406x364px) Image search: [Google]

50KB, 406x364px

>>57185018
AMD used to have the Geode until shareholders forced the company to sell it off right before smartphones exploded.

>>57185034
Geode is x86.

>>57185034
puma is superior to geode in every way

too bad amd is kill that too :^(

>>57184886
>It's actually shit,
>Only reason we use them in phones is because they are cheap.

Then it's not shit, is it? Production cost is EXTREMELY relevant to the merit of a technology. If production cost didn't matter, you'd be living in a cave and eating beetles and shit, and so would I.

Besides, you're wrong. ARM and x86 go blow for blow in terms of power efficiency:

http://www.extremetech.com/extreme/188396-the-final-isa-showdown-is-arm-x86-or-mips-intrinsically-more-power-efficient/2

>>57185018
perhaps

I think AMD is only dabbling in ARM in case it takes off in servers so they wont be left behind

File: 1471258730402.jpg (17KB, 232x468px) Image search: [Google]

17KB, 232x468px

>>57184105
>over 3 TFLOPS
>TDP of ~250
Meh

>FP64
Oh shit!

>>57184975
>It's 2 years old and offers near identical performance
It doesn't. How does 2.5 TFLOPS = 3+ TFLOPS?

>Also Tesla compute cards absolutely rape it. They're doing like 5 teraflops.
True but those Teslas are really lacking in software so knights landing wins by default.

>Knight's Landing is DOA.
>"The Cori supercomputer at NERSC, for example, is being upgraded with more than 9,300 of the new Knights Landing processors in 2016, while the upcoming Trinity system will be equipped with a similar amount at Los Alamos National Lab. At the same time, about 500 of the new processors will be installed in the second iteration of Stampede at TACC. Meanwhile, at Argonne National Lab, an 8-petaflop system named Theta will be deployed this year and will house what is likely to be a thousand or two Knights Landing parts."


https://www.top500.org/news/the-knights-landing-effect/

lol

>>57185098
>It doesn't.
It does. 2.62 real teraflops beats 3 teraflops calculated theoretically. Stop trying to push that + like you're actually the marketer moron who came up with that little Phi slide.
>True but those Teslas are really lacking in software so knights landing wins by default.
Nigga are you retarded? CUDA and OpenCL are FAR more used in HPC than x86.
>blah blah blah marketing bullshit about a bunch of national labs
Call me when it has wide market adoption. You know what else has a few labs and companies backing it? Itanium.

>>57184985
There's fuckall powersavings to be had, my 6600K desktop with an Nvidia graphics card web browses at 26W, it would be much less with a dual core i3 with no dedicated graphics, such computers dominate the corporate world.

>>57185235
>Nigga are you retarded? CUDA and OpenCL are FAR more used in HPC than x86.
Only because we never had ultra high performance x86 processors to compete with GPUs in FP32/64. Now we do so CUDA and OpenCL will start to become deprecated.


>Call me when it has wide market adoption. You know what else has a few labs and companies backing it? Itanium.

>"Intel intends to sell something on the order of 100,000 of the Knights Landing units this year, a claim the company is unlikely to make unless the current pipeline reflected that reality."

https://www.top500.org/news/the-knights-landing-effect/

We'll see soon enough.

I faced it years ago, you're late anon

>>57185382
Nice backpedaling, lad.

>>57185391
Nice, now you can rice/co figure and post in desktop threads with an inferior cpu architecture. That's very /g/ of you desu.

>>57185413
Whatever you say ARM(tm) shill.

>>57183741
ARM only dominates in one field and will CONTINUE to be the dominant mobile CPU Arch and x86 will continue to be the desktop Arch for everything, the TRUE future is OpenRisc V

>>57185417
hope your minecraft girlfriend from wood blocks project works out for you m8

>>57185439
Not once have I said anything pro or anti-ARM in this thread. You, however, legitimately act like an Intel shill. I know you aren't. since no one on /g/ is going to buy a Phi and you do it for free.

Microsoft can't release anything to ARM and their developers don't give a shit about ARM either as they made it to the top with intel.

>>57185472
They actually did an honest attempt with windows mobile but in the end microshaft can't make a mobile OS worth a shit.

>>57185439
It's ok to be wrong and lose an argument m8, relax

>>57184985
>b-but desktops used it almost 30 years ago
yeah in low-end school fleet shitboxes where it was only good for a few years while its competition was equally braindead, nobody was choosing ARM for anything serious outside of the UK where it coasted on Acorn's dominance anyway

>And it seems two-faced for /g/ to complain about software bloat and bad programming yet also supporting an industry that facilitates (and enforces) it through hardware expansion.
are you seriously trying to call modern ARM lean? even their microcontroller designs have more instructions than many classical "CISC" chips anymore, the terms "RISC" and "CISC" both are really nothing more than legacy labels nowadays, stop jerking off to decades-old marketing bullshit and open your fucking eyes, this architecture shilling nonsense is fucking retarded

>If you went to an office building and said you could cut their power bill to a fifth of what it is now, with no perceptible loss in performance (remember, most workstations now, as always, are essentially glorified typewriters and spreadsheets) then they'll switch, as long as they can run their very minor-draw software.
they already do that with SFF systems and zero clients hooked up to VMs running on x86 boxes in the datacenter, not shitty low-end ARM desktops with no software

>>57185531
>not shitty low-end ARM desktops with no software
And yet this same idiot tries to push that Phi garbage as 'capable of running a million VMs for light office work' or some shit.

>>57185531
>not shitty low end ARM desktops with no software
minecraft and stardock aren't professional applications kid, literally every software besides photoshop (which works fine with qemu on ARM) works on Aarch64 as of now.

>>57185608
>which works fine with qemu on ARM
Has that shit actually advanced? I remember trying to boot Windows 95 on a Pocket PC with XScale ARMv5 and it took 5 minutes, a decade later, it still took 5 minutes on a Galaxy S3.

>>57185579
can they even be used that way??? sounds really nonsensical actually, most big iron-class servers used for those kinds of jobs have no problem sustaining hundreds of users at a time anymore

>>57185608
>it's another facebook tard projecting his banal use case on everyone else episode
not everyone runs the same pool of 4-5 big name programs and web apps for everything else
the lack of windows alone makes it shit for business, deal with it

>>57185627
should be fine with devices that got cortex a72, it's at the same performance as server core 2 duo's, though virtual disk images are known to be slow as fuck

>>57185641
That's the thing though, it seems nobody is bothering to optimize QEMU for x86 emulation, so it runs shit no matter what.
There's Microsoft Virtual PC 2004 that ran on PowerPC Macs and that worked kinda OK, it was able to do Windows XP with a G4 1.5GHz, felt like using a Pentium II class machine.

>>57185637
Nigga stop pretending, you know you pushed that.

>>57185704
pushed what?

>>57183741
Ooh an Arm thread? I have so many questions for people who might know, if you wouldn't mind humouring me for a moment....

1) In the early 90s, my dad was very technically minded so we always had a family computer at a time when many didn't. We had an Acorn Archimedes, and later an A3020, and we hung on to that until about 98-99 before he finally caved and went windows. He was insistent that the Acorn machines (powered by Arm processors and running RiscOS) were technically superior to the admittedly faster Intel machines. The analogy he gave me at the time was this - Arm chips improved by increasing efficiency, tweaking and restructuring for gains in performance while keeping cool and low power. Intel chips however just crudely ramped up the power/speed and bolted larger heatsinks and fans on to stop it from melting under the strain. Intel dominated the market because of this speed, but he always said it was a shame people were so short-sighted, because with proper funding the Arm chips were the better way forward. Was he right? Was the analogy accurate? Were Acorn machines actually the shit?

2) Later when I was at uni, I was told by a lecturer that 98% of chips in the world were Arm, basically everything that wasn't a desktop computer or laptop. Was/Is this true? If so, why isn't Arm (a British company I believe) one of the biggest companies in the world, totally dominating intel and amd by now?

3) if all this is true and Arm chips are so amazing, why aren't desktop PCs running a stack of them instead of a single intel chip? Wouldn't it run much better this way? What's to stop a company resurrecting the Acorn philosophy? Are Arm chips the future? Was my dad right all along?

Sorry this is a pretty big ask, but I've wondered this stuff for years now, since my dad passed I haven't been able to ask him what he might think now, so I would love to know your opinions /g/

>>57184105
GP100 does over 5TFLOPs, 3 TFLOPs is unimpressive

>>57185637
>>it's another facebook tard projecting his banal use case on everyone else episode
>not everyone runs the same pool of 4-5 big name programs and web apps for everything else
>the lack of windows alone makes it shit for business, deal with it
i am amused by how triggered you are, and saying that windows is needed for anything other than photoshop shows that you don't know what you're talking about and what kind of applications are used in a business, openoffice is already very popular due to being free for example. It would help you more if you researched things before arguing so you don't end up looking stupid like now.

>>57185745
#2 is a straight up lie, especially at the time the statement was made. ARM Holdings is a rather large company, but all they have is IP.

>>57185745
>Was/Is this true?

Plenty of other architechtures. POWER and MIPS for excample, is the the PS2 and PS3

>>57185802
This was 10 years ago to be fair, but I think he was referencing all the tiny chips - those found in electronic toys, tracking collars, all shit like that.

>>57185842
Yea they're probably not ARM. They're going to be micro controllers, if not straight up standard ICs or even discrete electronics.

>>57185745
>Intel chips however just crudely ramped up the power/speed and bolted larger heatsinks and fans on to stop it from melting under the strain.
What did he mean by "crudely"? Intel, like any other vendor at the time, was adding more advanced features to their chips which inevitably increased complexity, die size, and heat output, while ARM pretty much seemed to stick to the keep-it-simple-clock-it-fast philosophy many early RISC designs tended to follow, if anything ARM was not "technologically superior" because of this simplicity, and it didn't keep that trend going for very long anyway, which can be pretty easily evidenced by the fact that it never saw any use outside of handhelds and acorn stragglers.

>Later when I was at uni, I was told by a lecturer that 98% of chips in the world were Arm, basically everything that wasn't a desktop computer or laptop. Was/Is this true
That's demonstrably false, the majority of "chips" in the world are microcontrollers, which is a very diverse market.

>if all this is true and Arm chips are so amazing, why aren't desktop PCs running a stack of them instead of a single intel chip? Wouldn't it run much better this way?
Not really, other than obvious incompatibilities and the fact that architecture doesn't matter for jack shit in 90% of desktop tasks where it's all abstracted away anyway, you'd just have a hot, shitty stack of shitty CPUs that won't bring anything to the table, not everything is embarrassingly parallel.

>>57185758
>i am amused by how triggered you are
not really
>and saying that windows is needed for anything other than photoshop shows that you don't know what you're talking about
the fact that you don't even recognize that adobe makes and sells things other than photoshop seems to demonstrate the opposite
>and what kind of applications are used in a business
and the fact that you also think that every single "business" use for a computer is banal office droning demonstrates the same as well
>openoffice
has been basically dead for years and the fact that you would bring it up over the vastly more popular LibreOffice pretty much shows once again that you have no fucking idea what you're on about and are simply projecting your retardation on others as I expected
>is already very popular due to being free for example
Windows and MS Office are "very popular" too yet you're sitting here telling me everyone will leave it in an instant the second some piece of shit ARM nettop shows up in reasonable numbers at your local bestbuy, so that doesn't really mean shit anyway using your own logic does it?
>It would help you more if you researched things before arguing so you don't end up looking stupid like now.
...he said as he provided literally nothing of substance

take your contrarian delusions back to >>>/1995/ where they belong

>>57184469
Wasnt the thread about deprecating the arch as a whole?
POWER8 is fucking amazing for parallel workloads.

>>57184469
The ISA isn't responsible for that, it's that SGI kept rehashing the R10k for 10 years instead of actually doing anything

>>57185608
>>57185758
>>57185917
Software on ARM still has a long way to come before it will be widely used in the business world. This is because most businesses don't just run word processing stuff and web browsers. The overwhelming majority of businesses run at least one piece of software on a contract basis for things like inventory, payroll, client information, machine operation, etc. etc. ALL of that software is written for x86 and you're going to have a hard time convincing all these small software companies to rewrite their entire software stack for ARM compatibility and reap no real benefit, as most businesses don't have a compelling reason to replace their hardware with ARM machines.

Pfft

This shit is deader than a fucking fan that doesn't spin.

>>57183741

>tfw PowerPC will never return to desktops

THANK YOU BASED APPLE

>>57183821
uh, you do realize most linux packages will work on arm right? I used to use my old HD2 as a mobile desktop back when it was still a modern device, shit was great.

File: 01b[1].png (39KB, 640x400px) Image search: [Google]

39KB, 640x400px

>>57184501
>android

>>57183779
please tell me you're being ironic

>>57185382
>OpenCL will start to become deprecated.
top kek

>>57184622
The only reason Xbox VMs aren't a thing is because the Xbox disc encryption and the nVidia GPU spoofing are a bitch to get around. x86 has nothing to do with that issue.

>>57184985
>Surely taking a minor hit to performance for the massive gain in power efficiency is a good thing?
x86 is more efficient than arm

>>57189258

>Xbox disc encryption
lol

File: 1476676277626.jpg (53KB, 1002x1199px) Image search: [Google]

53KB, 1002x1199px

>>57185444
Underrated and checked.

>>57189224
What do you mean x86 won't take over?

3+TFLOPS
+
T
F
L
O
P
S

File: pentium4.jpg (133KB, 700x870px) Image search: [Google]

133KB, 700x870px

>>57184568
>and the modern chips are so fast it doesn't matter anyway.

Modern processors are so fast meme.

Seriously where you living under a rock or are you posting from 1995?

http://www.gotw.ca/publications/concurrency-ddj.htm

We are running at the technical limitations of atoms themselves. Normal (not liquid nitrogen cooled) consumer processors hit the 4GHz barrier and are not getting faster.

Multi core is the new rage however working with multiple processors is hard and often gets you bizarre behaviors that you need to know.

The days of making a bloat-ware crap application that eats insane amounts of power for no reason and simply hoping that the processor gets faster are over.

Some non parallel tasks are at doomed and impossible to solve with more cores.

We are living at the end of the silicon age the 90s are over.
The days where revolutionary steps where made From Doom 2 to Quake within years are over.

And you gentoman want us to use a less powerful processor? In some insane delusion that we don't need it?

Changes are needed and they are not going to be pretty. You will need to code closer to the metal (the comeback of assembler?) and the processor architecture will need to be changed.

No virtual machines will save you and you will need to rewrite legacy applications to get more out of them. It will be a different time some will stay with the old x86 and never run their computers faster to have all their windows 7 programs and windows XP programs running others will embrace brave new territories with new low level programing languages on different architectures and difficult programing to get a little more out of the ethereally stuck 4GHz processors.

File: face-with-tears-of-joy.png (58KB, 256x256px) Image search: [Google]

58KB, 256x256px

>>57184610
999% Kook
999/10 retardation score.

You realize poor fag that you can run GBA and DS games on emulators under x86?

For the love of fuck GBA is such weak system that my old win XP computer can run it not 100% it can actually increase the clock speed on the emulation adn run the games faster to skip the BS on the games on GBA.

How kook can you be? How bad shit insane can you be to abandon x86 for nothing?
>You get a erection from having the same architecture of your CPU like GBA or running the games natively?

>>57184711
Get out normie, facebook is this way
>>>facebook.com

Seriously /g/ needs to purge the facebookers and the mods need to ban n00bs for posts like this.

There are linux distros for the raspberry Pi you normal fag fuck. You don't even know what that is normie.

File: 1458657153422.jpg (426KB, 740x740px) Image search: [Google]

426KB, 740x740px

>>57189824
>There are linux distros for the raspberry Pi you normal fag fuck. You don't even know what that is normie.
You mean like debian mobile?

>>57189824
you fucking retard even know what you are talking about? ARM is a fucking MESS, as much as x86, but at least x86 is well documented, unlike all those proprietary ARM SoCs out there

>>57189525
This is just the beginning. Future iterations of xeon phi coprocessors are going to fucking rape GPUs in FP64 performance while maintaining x86 software compatibility.

File: [muffled fan screaming noises].jpg (428KB, 1934x2030px) Image search: [Google]

428KB, 1934x2030px

>>57188790
RIP PPC
How viable is PPC these days?
Can you get a old PowerPC mac, put a custom bootloader and lunix on it without much trouble?

>>57189910
Freescale still does some powerpc shit and there's IBM POWER8

>>57185382
OpenCL can do CPUs too nigga, along with DSPs and FPGAs

post i responded to
>ZOMG you gays can we even like totaly port linux to ARM its like OMG so different from like x86 like man
I give a example why its retarded since it was made
>>57189863
>You mean like debian mobile?
They exist what is your point?

>>57189953
OpenCL is dogshit compared to just plain old C++ code (that x86 cores in knights landing can understand execute).

>>57184004
/thread

>>57189984
Nobody would waste their time trying to shoehorn plain C++ in an HPC envirement without openmp or openaac

>>57189905
No they aren't.

>>57189910
PPC is still alive and going strong in various embedded applications like HBAs, NICs, etc. They even have rad hard versions for aerospace applications.

>>57185382
>Only because we never had ultra high performance x86 processors to compete with GPUs in FP32/64.
Xeon Phi isn't something new you know:
>The product falls into the Many Integrated Core (MIC) architecture and prototype products codenamed Knights Ferry were announced and released to developers in 2010. The Knights Corner product was announced in 2011 and uses a 22 nm process. A second generation product codenamed Knights Landing using a 14 nm process was announced in June 2013.
>OpenCL initial release: August 2009
Xeon Phi had four years to kill OpenCL already, and GPGPU in general, but it's not h
appening, and for good reasons.

>>57189910
>custom bootloader
????
It has OpenFirmware.

>>57185745
2) Later when I was at uni, I was told by a lecturer that 98% of chips in the world were Arm, basically everything that wasn't a desktop computer or laptop. Was/Is this true?
Severs, Mainframes and HPC are 3 markets where arm isn't relevant for example.
>3) if all this is true and Arm chips are so amazing
Arm has lower performance per watt than x86, meaning that you need more electricity to do the same work.
>why aren't desktop PCs running a stack of them instead of a single intel chip? Wouldn't it run much better this way? What's to stop a company resurrecting the Acorn philosophy? Are Arm chips the future? Was my dad right all along?
Well, first off programming correct software is really hard, every nontrivial codebase has bugs. You can have 20 people looking over everything(and some companys actually do this) and you will still have bugs. Space Agencys invest a lot into making sure their code has as little bugs as possible because well, rockets are pretty darn expensive and having a rocket crash because of a bug sucks. Even then, this has already happened.
Now, this is for single-threaded(i. e. non-parallel) code. Parallel Code is much harder to write, and some problems simply cannot be parallelised. Suffice to say that the majority of code isn't parallel, and thus doesn't profit from more than 1 core. This is why single-thread performance matters, and x86 is simply the fastest in that regard.
Now, what if your code is parallel for some reason or another? then well, put it on GPUs with hundreds of cores or something similiar(Tesla, Xeon Phi etc.) or at least a 4+ core CPU and enjoy your performance. And these are far more efficient than a bunch of ARMs would be, see my above comment about performance per watt.

>>57189875
I'm not aware of how well or badly arm is documented, but calling x86 well documented seems wrong to me, Intel doesn't even document how long Instructions take[1], much less things like how cache eviction etc. work.
[1] although third-party benchmark-based documentation exists.

>>57189910
MPC5xx is popular in car PCMs

bosch/seimens/delphi all use it

>>57188946
>sarcastic
anon pls.

>>57186594
Actually, most of these proprietary software items are going to be ported to SaaS platforms (example: the Salesforce platform) within the next 5-10 years. And even the Microsoft Office suite has a workable web version. Besides, if you have any proprietary apps you can always virtualize them and use them on any platform using RemoteApp or XenApp style technologies.

ARM isn't going to become popular because people port their apps to ARM. It will be when all software becomes platform independent, and you just need a web browser and a decent desktop environment to access web apps. Then it all comes down to cost, and ARM wins.

>>57191099
SaaS is fucking cancer.

>>57189910
>custom bootloader
for what purpose
>linux
enjoy doing the same shit you can do on an x86 system but slower, at least install OpenBSD if you're going to be a hipster

>>57189692
What the fuck even is this rambling wall of bullshit? What does stagnant advancement and the difficulty of working with multiprocessing systems have to do with existing chips being easily capable enough for 95% of modern day use cases? There's a reason we aren't in upheaval about it, because there's just no fucking drive for it anymore.

>The days of making a bloat-ware crap application that eats insane amounts of power for no reason and simply hoping that the processor gets faster are over.
Jesus christ, fuck off with this "baww everything is bloated" bullshit. You can run modern operating systems and software on a fucking Pentium 4 from 12 years ago with usable performance, fucking Windows 95 wouldn't even run properly on a $2,000 386SX system made four years prior.

>Seriously where you living under a rock or are you posting from 1995?
What? Are you trying to imply that people actually saw the gimped, slow hardware of 1995 as "good enough forever"?

>You will need to code closer to the metal (the comeback of assembler?)
t. someone who's never even attempted to write anything meaningful in assembly
There's a reason even systems programmers barely write in pure assembly anymore, it's wasteful, time consuming and 90% of the time you're not going to write anything better than a good compiler. And once again, diminishing returns.

>and the processor architecture will need to be changed.
And why? So we can "fix" supposed problems with an architecture developers and compiler writers figured out decades ago? Throwing everything out and starting anew isn't going to do shit, just like it didn't in the '90s you hilariously accuse me of living in while you desire to make all the same mistakes.

>>57191099
>ARM isn't going to become popular because people port their apps to ARM. It will be when all software becomes platform independent, and you just need a web browser and a decent desktop environment to access web apps. Then it all comes down to cost, and ARM wins.
when it comes down to cost you can pick from the slew of sub-$200 x86 options available or just remove desktops entirely and replace them with zero clients hooked up to Windows VMs in the datacenter

nobody wants to waste time migrating over to a new platform, negotiating new contracts, preparing new documentation, re-training retards, hiring more support to deal with the inevitable fuckups people will manage to accomplish, all because some nostalgic teenager on the internet jerking off to specfp benchmark results and marketing drivel from the '80s told them "RISC is the future(TM)"

>>57192967
>when it comes down to cost you can pick from the slew of sub-$200 x86 options available or just remove desktops entirely and replace them with $30 disposable zero clients running ARM hooked up to Windows VMs in the datacenter

FTFY

>>57189692
>The days of making a bloat-ware crap application that eats insane amounts of power for no reason and simply hoping that the processor gets faster are over


Haha, not even remotely close to being true. Just look at all the cumguzzling web devs building desktop applications in JS.

>>57189875
>unlike all those proprietary ARM SoCs out there

Anyone could build an ARM CPU provided they put the money down to license the ISA. Meanwhile, only two companies can design x86 CPUs - one of which is circling the drain. Which seems better?

ARM will win and here is why!

Like x86 before the combination of huge volumes will enable pushing process and software innovation in a virtuous cycle.

Already you see Intel slip as 14nm was late and they are in their third architecture for it as the tick tock cadence has slowed. 10 and 7 nm are delayed as they can't afford to invest nor build the big factories.

TSMC and Samsung have their huge cellphone volumes to develop the technology and payoff the factories.

Already you see them both ramping 10nm while x86 and intel slip.

Just like PowerPC SPARC and the many other low volume architectures were conquered by x86. The ARMy will crush x86 by 7nm. Volume and cadence can't be beat. It is all about economies of scale and intel can't compete anymore.

Intel and x86 destiny is sealed and it is dire.

>>57193182
You can have 300 trillion angry birds processors in a warehouse ready to ship but very few people will buy them because it can't run X86 software, uses more electricity, and has horrible IPC.

>>57191099
Worst possible future imaginable.

Inefficient lagi shit
needs Internet connection for everything
mass surveillance 24/7 in your SaaS
Getting exploited by the SaaS rental service
Updates will fuck you over hard(youtube changes its UI and you can not go back to the old)
no going back to previous versions

Oh and I assume that these magical Saas will run on magic and unicorn farts in the cloud? They are programs on a server that needs a processor and RAM and you are telneting/connecting to this computer for it to rent its programs to you.

Complete scam.
>>57191460
404% agree

File: image.png (328KB, 500x483px) Image search: [Google]

328KB, 500x483px

>>57183769

>>57193021
>and the end user is still doing all of their work on x86 and Windows
how does this advance your claim again? ARM is already popular in embedded applications

>>57193182
>Like x86 before the combination of huge volumes will enable pushing process and software innovation in a virtuous cycle.
good thing x86 won out by developer support rather than volume, most of the volume was in 8-bits and other cheapie home systems until clones really started ramping up long after the PC had already cemented itself in the business world
>blah blah blah muh manufacturing process
end users don't give a shit
>Just like PowerPC SPARC and the many other low volume architectures were conquered by x86.
those "classical" RISC architectures weren't "conquered" by x86 because they were low volume, they were "conquered" because by the 2000s they were slow, overpriced, unsupported garbage that even their own vendors had left for dead by then

File: PROCEXP64_2016-10-22_22-24-28.png (6KB, 461x51px) Image search: [Google]

6KB, 461x51px

>>57192494
>Jesus christ, fuck off with this "baww everything is bloated" bullshit. You can run modern operating systems and software on a fucking Pentium 4 from 12 years ago with usable performance
that's bullshit.I agree with the rest of your points.

>>57192494
Wow dude you are really mad and rambling.
>stagnant advancement and the difficulty of working with multiprocessing systems hav
You need to write good code not make everything in java and wait 2 years for the processor to get faster for your single threaded code.

>existing chips being easily capable enough for 95% of modern day use cases
What cases? Give me examples?
So your plan is to write code like you did in the old days?
So basically you can now use a web browser, a music player and torrent something.

Write code more inefficient and now you can only web brows and no longer use the music player. See the problem?

>What? Are you trying to imply that people actually saw the gimped, slow hardware of 1995 as "good enough forever"?

I'm saying that in the 90 we did see massive revolutions in power and technology. DOOM 2 was the perfection of the 2,5D FPS and in a couple of years it jumped to full 3D and models for every enemy in quake this was a revolution.

However everything after pentium 4 stopped going single processor, code was easy to write if all you did not have multi threaded problems, the same code ran the same way only faster.

>And why? So we can "fix" supposed problems with an archite
Can i get this into your head? The free ride is over. Every other gain will cost something. It will be hard to get more preference out of the computer in the past the processor got faster and did the same things only faster. Now its over and tricks like multi core are trying to push the advancement further.

The other option is to try to change the architecture itself if a better one is found (I'm not saying it was found) and another solution is writing everything closer to the machine maybe with other languages new low level languages.
The solution can not be

>LOL write everything in python
>LOL write everything in flash

We will hit this brier in the future, if a logic gate is the size of one atom (maximal hypothetical limit) you can not miniaturize

>>57193515
>What cases? Give me examples?
shitposting on websites
watching vidoes on youtube
office work
whatever

>>57193515
do you honestly believe performance matters for the majority of application? as long as it runs fluidly(it will) it's enough. Most computer tasks don't become more processing intensive for some reason like you seem to believe.
yes, for the majority of code >LOL write everything in python
is perfectly fine. I'd rather not waste resources for no reason at all, but if devs chose to nothing is stopping them.

>>57193115
>Haha, not even remotely close to being true. Just look at all the cumguzzling web devs building desktop applications in JS.

Are you thinking about any specific project?

And everything can be marketed and sold to idiots (look iPhone, google crap) this will not make the crap magically good.

I'm not saying that there are no insane JS desktop devs I'm saying is that what they are doing is extremely bad.

In the end they end up with a lagi slow application that after starting up takes 99% of system recourses and the normies are happy.
Simply because its made by google or is WEB or SaaS lets make a comparison watching videos on youtube VS on your HDD.
Download the same movie HDD and make a comperision:
YT
>takes time to load
>can not move forward in time line if it did not load
>player can crap up and give you the snow for no reason whatsoever
>video quality is downgraded sometimes
>Internet connection is the slowest part of the computer
>eats more system resources
>packets are pushed in a crappy packet switched network

HDD
>Instant play
>Can move forward and back always, never needs time to load
> the player practically never collapses while on YT its once a day.
>eats less resources
>is faster
>On a far more faster connection then the Internet
>Data is sent directly on dedicated way from the HDD to computer

>>57193678
>ARM makes normies happy
And that right there is why it is the superior platform. Deal with it, nerd.

>>57193583
And we end up with a scenario where you can only run one application at a time since its all so bloated.

Write your OS in python and everything else in it and see how you can only have one web page opened. So you need to close it to open one music player.

I'm not saying that this will never happen, I know it might if the normies are retarded enough and the devs greedy/lazy/insane enough this might be a future we will have.

Get ready for the normies to post
>Look at my BS supercomputer it can open 2 web pages at the same time!


What a shit time to live.

ARM is good for embedded and maybe thin clients, phones, etc. It's just too feeble for anything else.

>>57193514
That's practically cherrypicking, Steam is absolutely awful. But while it may be inexcusable, it's not apocalyptic.
>>57193515
This whole post is fucking unreadable. Why the do you keep rambling about shit code as if migrating architectures and throwing out decades of compiler advancements and general developer knowledge in the process is going to suddenly make shit developers not shit? It's a fact of life, deal with it.

>What cases? Give me examples?
Video, music and other media
Email, messaging and other communication
Documents, spreadsheets and other general office suite tasks
Image editing
Fucking everything else but games and large dataset computation, the resources these chips offer are immense for most problems we use computers to solve.

>I'm saying that in the 90 we did see massive revolutions in power and technology. DOOM 2 was the perfection of the 2,5D FPS and in a couple of years it jumped to full 3D and models for every enemy in quake this was a revolution.
Okay? And? It was the fucking '90s, and the '90s are over. We weren't nearing the limits of silicon itself, and the drive to advance was much more obvious because computers were fucking shit even at the time. That's not the case anymore, and declining PC sales are a testament to it.

>Can i get this into your head? The free ride is over. Every other gain will cost something.
And switching the architecture is again going to change this how? Moore's Law was a trend, not an entitlement. Even established alternatives in HPC like POWER and SPARC bring very little to the table anymore.

>and another solution is writing everything closer to the machine maybe with other languages new low level languages.
Yeah, fuck maintainable code, write everything like a shitty video game from the '80s!

Is ARM as compromised as the x86 platform is, security wise?

>>57183741
They said the same thing when PowerPC came out... and before that, for 68000.

>>57183741
more like desktop and server are deprecated. mobile is the future

>>57193908
You're a retard.

>>57193785
>this level of bullshit exaggeration
yet there are people on this board still dailying 10 year old mid-range C2D boxes, really makes you think huh

software "bloat" in general has been pretty stagnant outside of webdev which is still a relatively new thing that still has obvious teething troubles from the sheer amount of retard developers in it for the fad and the dominant languages being shitty because they were never really expected to be doing what they're doing now

the fact that you think it will ever get to this point shows you're either poor as fuck and still using a toaster or you're really just fucking dumb and don't actually know what you're talking about beyond spouting /g/ memes as facts

>>57183846
>x86 is shit.
what did he mean by this.

File: 1476831923350.png (87KB, 254x271px) Image search: [Google]

87KB, 254x271px

>All these replies

>>57184713
3 teraflops double precision is very impressive. You can also do 6 teraflops with single precision. Intel has basically created a CPU that is on the same performance level as most GPUs. If NVIDIA hadn't responded with their P100 they would actually be behind intel from a performance perspective.

>>57185984
>POWER8 is fucking amazing for parallel workloads.
That's cool for big databases and HPC, but really meaningless for a home/office box, hence why I specified "single-threaded" performance which, yes, is relatively shit.
>>57186006
Which means they didn't improve the ISA with it either, they've got a lot of work to do.

And on top of that, the much more HPC-oriented x86 of today is very different from the mostly low-end desktop-oriented x86 the R10K summarily obliterated in 1995.

>>57194087
POWER8 is the last bastion of freedom for the desktop platform from the evils of Intel ME/AMD AMT, performance be damned.

>>57184985
>minor hit to performance
>ARM

Fucking kek.

>>57193997
It might be impressive if it drew less power, were actually confirmed to hit 3 teraflops, and could be actual display adapters that function in a workstation. Meanwhile the P100 just absolutely rapes it. 1.7 teraflops more than the Phi at the same power draw.
>responded
>to a product that hasn't yet launched

>>57193785
>the majority of code
the majority of code is application code. low-level code like operating systems, compilers and interpreters of course need to be good, and they are. Developers write tons of mediocre javascript? We have very good javascript engines nowaday, which significantly lessens the issue. And if your site overdoes it then your site will feel laggy, which will make people simply stop using your site.
>>57193839
>That's practically cherrypicking, Steam is absolutely awful. But while it may be inexcusable, it's not apocalyptic.
You completely missed my point. The small part of OS memory consumption pictured is already 220 MB. And that's Windows 7, not W10 which also does text to speech(for telemetry).

>>57193883
not as much, but it really depends on your board
https://genode.org/documentation/articles/trustzone

>>57184812
It's not good for desktop because most SoCs are designed exclusively for mobile phones. Very often the only port they support is USB which is the primary reason why the Raspberry Pis run Ethernet over USB. Even though adding more I/O capabilities would literally cost only pennies more and make the SoC viable for desktops. You will basically have to swallow the whole cost of creating or ordering a new SoC. Which is going to be very expensive because you will likely not sell 250k units of your obscure desktop system which doesn't support widely used software whereas selling 250k chink phones with android is not difficult at all. Even the latest Raspberry Pi has hardware that is several years behind current chink phone tech.


>>57184985
It's not just a minor performance hit and it's not a massive gain in power efficiency. If you find phone level performance acceptable then just use your phone. The cost of switching will be greater than the gain. Especially since we are soon reaching the limitations of silicon. If we use a different material than silicon then it doesn't matter which ISA is more efficient. The manufacturer with the new material wins regardless of ISA.

>>57185641
Even though it's already a year ago that A72 cores were used in mobile SoCs the networking gear that is supposedly using it is still not available. The A72 is okayish. An eight core SoC is probably good enough for low power web servers.

>>57194129
>performance be damned.
power has great performance, altough not for the price i guess.

>>57194408
Why don't you amortize the cost of the SoC by also moving 250,000 chinkphones with the same SoC.

File: 1475053434588s.jpg (11KB, 250x230px) Image search: [Google]

11KB, 250x230px

>>57183779
It will be the standard when Apple makes a ARM-based Macbook Air.

File: Linley-node-table-x-800.png (72KB, 500x200px) Image search: [Google]

72KB, 500x200px

>>57193182
7nm, 10nm, 14nm is just marketing. Literally every manufacturer measures their shit differently.

>>57194454
remember when PowerPC was the standard because iBooks were PowerPC-based?

>>57194622
*because all Apple computers were PowerPC-based

>>57183741
>le x86 and ARM are the only archs there are meme
I want to believe in POWER.

>>57185018
Yea and it shares the same socket. So there will be ARM desktops

>>57185382
LOL WHAT
Opencl uses CPU and GPU faggot

>>57194622

back then apple had no relevance, now they are trendsetters so switching their arch would actually be relevant

>>57184482
C++ has a shit design yet people still use it. Adding good features always works. Even with an old programming language or ISA.

>>57183821
you can run any DE you want on ARM, even Windows.

>>57196321
nope, Apple is no IBM and will not open their ARM designs/hardware platform, which would be crucial like Linus Torvalds said
>what matters is all the infrastructure around the instruction set, and x86 has all that infrastructure... at a lot of different levels. It's open in a way that no other architecture is... Being compatible just wasn't as big of a deal for the ARM ecosystem as it has been traditionally for the x86 ecosystem... I've been personally pretty disappointed with ARM as a hardware platform, not as an instruction set, though I've had my issues there, too. As a hardware platform, it is still not very pleasant to deal with.

>>57194454
>>57196321
Their relevance is in the feature set, not the hardware implementation. Besides, even Apple doesn't give a fuck about Macs anymore, that platform is practically dead outside of schools.

>>57196946
There's as many flavors of ARM as there are loonix

>>57194618
Your table shows a far smaller exaggeration than 14 actually being 7 would be.

>>57197365
This doesn't fucking help at all.