How long until we can have ARM based computers replacing x86_64

>>61916178
Well in once sense they already did for a lot of people, they're called smartphones.

People who are willing to accept a computer that sits in one place on a desk and can't be carried around in the pocket are generally making that tradeoff because they want something more powerful than ARM, which is pretty much always going to mean x86. Not that it necessarily has to be huge, NUCs and shit exist. But there's really no compelling case to make an ARM desktop.

>>61916178
No backwards compatibility. Besides, you shouldn't be rooting for ARM anyway, you should be rooting for RISC-V or J2-J4 to be production ready.

Because I want to be able to upgrade

>>61916233
Linux has ARM ports.

Windows software that isn't ported to ARM would just be deprecated.

File: 1498535039182.jpg (183KB, 1000x705px) Image search: [Google]

183KB, 1000x705px

Considering that these SoCs are 10x as powerful as a desktop from not that long ago, the problem lies in shitty software and bloat.

>>61916314
You mean 10x less powerful.

>>61916326
No 10x more powerful

>>61916326
No. Desktops used to have less resources than they do now.

>>61916339
>>61916344
Are you comparing the newest most powerful ARM CPU to a pentium 4?

>>61916233
>>61916289
Well software ecosystem isn't that big of a problem, it's user centric, users can be trained. It will take maximum 1 year for a user to shift from windows to Linux ecosystem. I believe it's a non issue

>>61916265
Oh I should have made this clear, I am not talking about those soldered on SoCs, I am talking of ARM boards with PCIe slots, removable CPUs, and rAM sticks, that kind of.

>>61916420
although it would be easy to shift users, they don't want change

a lot of users expect their devices to "just work" (that also explains apple's popularity)
they don't want to get into the technical details or even relearn the way they work

>>61916178
>There are ARM CPUs that can work for a decade on a single AA battery, on the lower end, but higher end ones are very power efficient too.
But no where near as power efficient as what you said.
>The performance is not too bad either, it's enough for basic browsing, watching 1080p videos, typing out documents.
Those things require very little in the way of performance, particularly decoding video which is usually done on die much like graphics cards do.
Try using software decoding if you hate yourself.

ARM excels in its niche, it doesn't do performance well.

File: Raspberry_Pi_Zero_ver_1.3_1_of_3_1024x1024.jpg (101KB, 1024x1024px) Image search: [Google]

101KB, 1024x1024px

>>61916178
If I could get my hands on an ARM notebook that is well-built and has high quality display I'd get it instantly.

ARM is a smaller, much more simple architecture than x86 and its ISA is just great and I love it.

File: rog-ryzen-gaming-laptop.jpg (103KB, 1667x989px) Image search: [Google]

103KB, 1667x989px

When they match IPC and performance-per-watt. So, never.

pic related is rumoured to be able to run all 8 cores at 3 GHz under load.

>>61916233
/thread
ARM only belongs where power consumption isn't a concern

>>61916900
>anus
No, thanks.

>>61916931
That's not the point. More laptop manufacturers are going to adopt the R7 1700 in their laptop as well. The dam chip is a fucking energy efficient monster.

>>61916900
>When they match IPC and performance-per-watt.
If that was the criteria, it would have been an ARM world, except that its not. ARM is ridiculously efficient at low power and clock speeds, but you need something more powerful than your average ARM chip to even browse the internet these days

>>61917011
>If that was the criteria, it would have been an ARM world, except that its not.
ARM has neither performance-per-watt nor better IPC. Why do you think nobody ever took it seriously on laptops and servers?

>ARM is ridiculously efficient at low power and clock speeds, but you need something more powerful than your average ARM chip to even browse the internet these days
No it isn't. An x86 chip at the same frequencies will outperform it and use less electricity.

The reason ARM is used at all is because it's cheap as dirt and is included as a part of a SoC with a GPU, LTE modem, ect. You can literally buy snapdragon 820 chips for like $20. But then you're core-m7 is like $400.

>>61916855
>ARM is a smaller, much more simple architecture than x86

ARM is NOT a single architecture. There's ARMv6, v7 and v8, further subdivided into M, A and R profiles, there are significant differences within each profile (e.g. Cortex-M0/M1 are Von Neumann while Cortex-M3 and higher are Harvard) and binary code cross-compatibility is extremely limited unlike x86.

>>61917061
>ARM is NOT a single architecture.
I didn't imply otherwise. I was making a general statement about the IPs produced by ARM and their comparison to x86.

>>61917411
The "fattest" ARM variants are already comparable to x86_32 complexity-wise. They have less peripheral baggage (like line A20 bullshit), but that is both a blessing and a curse - you can't take an OS made for an older ARM core, add a few drivers and run it on a newer ARM core without a full rebuild.

>>61916178
>ARM
>No BIOS
>Manufacturers Push Closed Source Bootloaders

KEK.

>>61916900
Why the fuck did they put Ryzen 7 in a gaming laptop.

>>61917529
>The "fattest" ARM variants are already comparable to x86_32 complexity-wise.
Which ones do you mean?
Take armv8 a64 profile for example. The IS is built from the ground up. The ability to conditionally execute the instructions has been removed to make everything cleaner.

They increased the number of general purpose registers to 31 but that's it.

>>61917722
Because they can.

>>61917722
Because it's that energy efficient at least at base frequencies (maybe undervolted too).

>>61916178
>How long until we can have ARM based computers replacing x86_64 machines?

Bet on Apple to move MacBooks over to their custom ARM SOCs within a decade.

>>61916178
Most ARM SoCs will only boot a kernel modified by the vendor. Just look how bad the situation is when it comes to Android.

On top of that, most of these kernels will have binary blobs in them, without which these machines won't be even able to boot.

>>61917862
Yeah and macbooks will still have 5% market share. What's your point?

>>61916178
>How long until we can have ARM based computers replacing x86_64 machines?
Never

>>61917890
If anyone is going to push ARM in to laptops, it's going to be Apple. They have an industry leading SOC design team and an ecosystem that will port x86-64 applications over to ARM.

That will signal the beginning of the end for Intel, AMD already has a Keller designed ARM arch in the back but it got shelved to get Zen out of the door.

>>61917908
>ARM in to laptops
It already happened.
https://www.pine64.org/?page_id=3707

>>61916178
ARM computers are already replacing home theater media computers and low budget PCs. But x86 will always remain king when you need lots of power for things like video rendering and 3D games.

>>61916178
Same reason we don't use linux.
(let's just convert every piece of software on the planet to run on it, should be quick and easy)

>>61917689
I mean, if you're do butthurt about it, YOU could make one

File: 1461835577658.jpg (217KB, 644x1049px) Image search: [Google]

217KB, 644x1049px

>>61917908
The problem is it will still have 5% market share and therefore poor software support.

Also mac os software is so abhorrently bad not even apple themselves want to use it. Porting that to x86 will only make things worse.

>>61918036
>The problem is it will still have 5% market share and therefore poor software support.

Didn't stop developers from porting PowerPC applications over to x86, not an argument.

>>61917917
Really, then why are x86 NUCs becoming so popular?

Too many x86 asm only programs for video processing.

>>61918057
I dont know anyone with a NUC but I know tons of people with ARM based devices like Roku, Chromecast, Smartphones, etc.

>>61918053
I guess but it's nowhere near the end for AMD or intel. 5% will use macbooks that mostly play angry birds and 95% will use x86 computers to do work.

x86 will be fine especially with things like >>61916900 happening.

>>61918057
Overpriced.

>>61918098
Apple Ax SOCs have outstanding perf per watt

>>61918057
They're not, there is a reason Intel is cancelling their chips in that field