Is there any reason why people don't use risc architectures on general use computers nowadays?

Is there any reason why people don't use risc architectures on general use computers nowadays?

Also, what does Jow Forums think about RISK-V? Is there aby future on it?

Attached: images (1).jpg (400x400, 13K)

Other urls found in this thread:

riscv.org/membership/
twitter.com/NSFWRedditImage

In the late 1980s and early 1990s the RISC philosophy let you spend all that die space that would have otherwise gone to a decoder on more productive outlets like registers, or omit it entirely for a small, simple and efficient chip that could make up for its shortcomings by clocking incredibly fast, usually double or even triple that of most CISC designs. On top of all of that, it was easier to develop an optimizing compiler for.

But transistors kept shrinking, while decoder hardware didn't get any more complicated. New use cases emerged that benefited heavily from new instructions and co-processors that made simple but fast-clocking RISC designs increasingly worthless, eventually both schools of design just took the best parts from each other and ultimately coalesced into what they are today. With most modern RISCs being out-of-order, accelerated abominations that are about as "reduced" as my mother's fat ass while many CISCs are relatively clean and simple under the hood where it makes sense in an era where people don't really develop software in pure assembler anymore.

>Also, what does Jow Forums think about RISK-V? Is there aby future on it?
Probably not outside of academic and hobbyist circles. It just doesn't offer anything practical.

Attached: KL_Hewlett_Packard_PA8000.jpg (1368x1359, 896K)

>Probably not outside of academic and hobbyist circles. It just doesn't offer anything practical.
One look at their site would suggest otherwise
riscv.org/membership/

A bunch of logos on a web page doesn't really mean anything. It's still little more than a novelty that's 99.9% talk and 0.1% product, the only products of which are mostly incredibly expensive development boards that don't serve any useful purpose as anything but a marketing tool.

nogaems

>Is there any reason why people don't use risc architectures on general use computers nowadays?
Sheer momentum alone.
>Also, what does Jow Forums think about RISK-V? Is there aby future on it?
There seems to be enough interest in it. It may be able to carve out its own niche.

it will slowly happen now that the Chinese are getting their own semiconductor industry

the average user don't really care about this stuff

Attached: 1692f34c5f2f4.jpg (1000x500, 141K)

The issue with RISC-V and anything that isn't x86 is the fact there is no standardized system arch that is somewhat agreed upon
Even with ARM, software still has to be developed with individual devices and computers in mind
The last thing we want to go back to operating systems limited to specific computers and compatibility outside is not guaranteed

most risv chips will probably be designed to support "global" compatibility

making different architectural components seamlessly boost performance in x or y task

I think for Linux specifically, they decided on RV64GC for the baseline of what they'd support. It's basically what you'd expect from a modern CPU.
CPUs with fewer extensions/features than that would probably be intended for very specialised situations and not be using a general purpose OS.

Established RISC platforms had plenty of momentum and users too, but they were just pointless on a desktop by the 2000s. Most of the tasks you used to need a workstation for could be performed just as easily or better on any cheap NT machine that could do even more on top of that, those that didn't were mostly dinosaur legacy applications that were being phased out or ported to saner platforms that didn't carry a horrifying price tag, horrifying performance and a horrifyingly shitty juggernaut corporation trying to vendor-lock you at every turn.

Sun and powerpc also had their own standard.

Modern x86 chips bolt an ugly CISC architecture on top of an underlying RISC one, that's overhead. Theoretically a RISC architecture should still be more performant, all else being equal.

Theoretically, but in practice? Not really. High-end RISC desktops were terrible in the years leading up to AMD64 almost completely invalidating their continued existence.

As did SGI and most other desktop platforms, embedded is always a shit show though.

More like CISC instructions get converted to micro ops to compute on the RISC core
I also second what said
There really isn't going to be massive gains going to a pure RISC core from x86 with a RISC core
If there was gains to be made than even ARM should have been able to wipe the floor with x86 but that hasn't happened yet

this desu

I don't see why that's a problem unless your software is proprietary.

More work for developers to support a million shitty proprietary implementations of the same thing and keep it updated (yes, even if you have the source.) Promotes anti-user design practices. Promotes shitty hardware and products all around. You're a fucking idiot if you think desktops becoming big smartphones is a good thing.

For enormous volumes where your saving is over ARM licenses it makes sense. That is embedded systems like the 3 controllers on a hard disk. For lower volumes perhaps stick to ARM.

Good points. The future is probably in CISC-6. You read it here first.

It's literally all RISC now, just because the ISAs used are CISC does not mean the chips aren't RISC.

When did they redo Coldfire as RISC?

Intel tried to topple x86 themselves and failed, so don't expect a change in desktop ISAs.
RISC-V is nice though, and being open actually does bring a lot of advantages. The interest in it isn't just a meme, but unfortunately it is very vulnerable to "embrace extend extinguish", since it is designed to be easily extensible.

>>Intel tried to topple x86 themselves and failed
At least three times, actually (iAPX 432, i860, IA-64)