When will other archs than x86 be viable for desktop use again?

>>59635787

When they actually get work done.

Ever try to chroot into GNU/Linux on an Android phone? Even if performance increases 10 times it wouldn't suffice.

>The less useful a thing is to the general populace the better that thing is

In a few years. Technically we already have ARM Chromebooks. Whether they will surpass x86 is another story.

>>59635787
Not before the US makes a law that forces ARM CPU manufactures to include backdoors.

Can't let the population run to compromised hardware.

>>59636486
Wait, are you too young to remember POWER Macs?

>>59636521

*uncompromised

ducking autocorrect.

>>59635787
Whenever you get a desktop that supports it.

The problems with the other archs right now, is that you can't get desktop motherboards for them, without spending 10x as much as on a desktop mobo. When people complain all these things are slow, they're comparing either old desktop/dev boards to modern intel rigs, or new-ish phones and embedded boards to modern intel rigs.

When you can get a board with modern interchangeable and standard memory (DDR3 or DDR4), with pci and pcie slots, and probably on-board SATA 3 (6Gb) controllers, THEN you'll find they're already at the point where they're viable desktops.

>>59636525
No, I also have a VIA embedded arm system running right now.

That doesn't mean it's good for the GENERAL POPULACE you fucking illiterate.

PCs, as in x86 computers, are modular. PCIe as a bus, ACPI to configure devices you just plugged in. ARM systems tend to be locked down, you have to hard-code memory addresses into the kernel or else it won't even boot. That's why on phones every phone needs a ROM of its own. You don't need a different ISO for each PC you buy.

I know this is not directly tied to CPU architectures but today we live in a world where x86 machines tend to be the easiest to install free software on. So I hope it remains the standard.

>>59635787
64 bit Z80 cpu when

>>59635787
Only when this "other" architecture will be able to emulate x86 with better performance and full software and peripheral compatibility.
TL;DR Never.

File: aizen.png (468KB, 680x451px)

468KB, 680x451px

>>59636521
Since when were you under the impression that ARM SoCs didn't already have backdoors?

>>59636564
well the Z80 is pretty much an 8080 clone, and the 8086, which current x86 cpus spawned from, it assembly-level compatible
so you could argue, stretching a lot, that we are using pretty much using 64bit Z80's

>>59636540
This is really what I'm waiting for. While I'd prefer something like POWER, I could make do with ARM or MIPS if it was available in that form.

When there's a market for that.

>>59636920
8086 CPUs could run 8008, 8080, and 8085 instructions but running code this way wasn't optimal. It was intended to allow easy upgrades for existing legacy hardware.

On the other hand the Z80 is a true superset of the 8080, it was intended to run everything the 8080 could run natively but it also added stuff on top of that. It had a whole second set of general purpose registers and a few unique instructions of its own. 8080 and 8086 processors do not have those things so Z80 code would cause interesting errors running on those processors.

>>59638549
i don't know all the details, just that they're at least somewhat related
would AMD64 really be much or any different if history had the Z80 in place of the 8086?

Architectures other than x86 will be viable for desktop use when and only when Steam is available for those platforms.

>>59635787
The OpenPOWER foundation has a lot of members. You can buy a POWER8 rack mount system right now and use it as a desktop machine if you really hate yourself. They have good performance at house fire power levels and not everything will compile on it, but it's an option, and it's certainly far less backdoored than x86 or ARM.

>>59638698
I think the alternate bank of registers wouldn't really change much, I don't think it would even make sense to use them for anything nowadays because modern x86 processors basically do this automatically through a technique called register renaming.

Basically on modern x86 processors there may be like 8-16 or so "named registers" such as AH, AX, or EAX but the names do not actually directly map to any particular register. The processor might actually contain as many as 100 (or more) internal registers and during execution it might map the same name to any of the hundred or so registers. Names being strictly assigned to one register means that that the register has to do everything on its own in sequence. Register renaming allows the processor to break sequences of code into self-contained parts that can be executed out of order for even faster execution.

>>59636455
Gnu/Linux works just fine on arm devices like rpi and such.

>>59635787
never

>>59639119
They're not at all good for desktop use though, and they're way too expensive. Had the Talos Secure Workstation campaign panned out correctly, it would have been sorta affordable after a while. It sucks that it failed, I think mentioning the second run a bit too much was a major cause, the Osborne effect and all.

>>59635787
Yes. When we develop CPUs with neural networks that are capable of multi-arch.

ARM version of Windows 10 would be good enough for normies who don't play games.

>>59639243
>Yes. When we develop CPUs with neural networks that are capable of multi-arch.
"What was the Transmeta Crusoe?"

https://en.wikipedia.org/wiki/Transmeta_Crusoe

>>59639222
Talos was doomed to fail just because of the funding reqs and cost per machine. The 500k they got would be plenty enough for a bootstrapping run if they had a market. The problem is, nobody is going to buy a $5,000+ workstation. The unit cost needs to be less than 3k for it to be a viable product and POWER clearly can't be done for that price yet. It needs to catch on big in the server and supercomputing markets first.

>>59639444
That does not use neural networks and can only emulate x86. The idea of CPU with neural networks would be the user (or vendor) being able to install or manually teach the program for any architecture emulation. Maybe we can have this in 50 years.

File: bobsgame.png (276KB, 714x442px)

276KB, 714x442px

>>59635787
SuperH / J2
The open source BSD licensed VHDL code for the J2 core has been proven on Xilinx FPGAs and on ASICs manufactured on TSMC's 180 nm process, and is capable of booting µClinux.[5]

Anyone tried one on poorfags altera or on xilinix?

>>59639706
found it
http://j-core.org/#get_hardware

>>59636695
It depends mostly on the manufacturer.
Qualcomm already has a page in the Vault 7 leaks. I don't know much about others, like NXP, STM, Texas Instruments or Atmel.

Resurrection of SH is interesting. I heard it is popular for vehicles, but consumer products are seldom since Sega consoles.

>>59639119
I had a seminar on POWER systems by IBM and of course it is not 'great' for daily use at home but at enterprise levels POWER systems are much better than x86 systems. They have systems that have up to 8 threads per core which could be great for virtualization

Debian supports the SuperH but only some boards which appear to be for industrial applications or something?

https://wiki.debian.org/SH4

Looks like they also support SH3 kinda but not officially.

>>59635787
Power9, arm64 already available.

>>59639922
Yeah it's just that they have shit clock gating or something. The idle and full load power usage are close to the same. So you really need to keep the workload high to get good energy efficiency. They're perfect for supercomputers or IaaS where you schedule work out in slices days in advanced.

>>59639793
>>NXP
freescale development was so impressive, Their POWER products growth rate was way more impresive than intel or amd shit 5% ipc.
Than nxp bought them...

File: 1407955331589.png (140KB, 500x500px)

140KB, 500x500px

The problem is that other architectures don't receive support
Windows didn't made a POWER version or 3dfx a GPU that was compatible with POWER.
Now we do have windows that is compatible with ARM but still no Nvidia GPU, much less games that are optmized for ARM.
And if other architectures don't receive normie stuff like that then they will never take off

I remember /g/ telling me ARM is only popular because its low energy usage but that developing for it is hell, is this truth?

>>59641703
>Windows didn't made a POWER version or 3dfx a GPU that was compatible with POWER.
Windows NT did have a powerpc version. And a MIPS, and Alpha. They were, in fact, supremely better than NT4 on x86. Mostly because the base hardware for each was completely controlled by one manufacturer, and it was end to end high-end high-quality hardware. And people still didn't buy it, but that was because there was little to no software for it, and microsoft made it clear that they weren't going to continue developing it.

>>59641783
So CPU architectures are like game consoles
That just makes things harder

You can get a softiron overdrive 1000, has an amd opteron 1100 arm cpu in it, runs linux, it could be a fine normie desktop.

>>59635787
>When will other archs than x86 be viable for desktop use again?
They already are, and simultaneously never will be depending on how important backwards compatible is. The reason being x86 has been the stuck around so long is backwards compatibility.

File: 308Ichigo_injures.png (394KB, 710x461px) Image search: [Google]

394KB, 710x461px

>>59636695
I have this little getsugatenso called RISC-V, Aizen

>>59635787
I see closed source in general as the real problem for other archs to emerge for the masses. With open source you just recompile. With closed source you have to emulate if there's no binary for your arch, and that's slow