The Move Away From x86 To ARM Processors On The Desktop To Start Soon - Survey

[qubit]

It looks like there's a subtle but relentless push to get ARM CPUs into desktop PCs. Morgan Stanley recently surveyed 30 PC makers (names not revealed) and discovered that 40% of them are interested in trying out ARM-based PCs within the next two years. As we reported previously that the Wintel alliance appears to be crumbling, this finding appears to add weight to that assertion. Of course, there's a huge mountain to climb before ARM processors can compete head to head with high performance x86, as explained in our article, not least because Microsoft won't begin supporting ARM until Windows 8 is released late next year and the fact that the vast majority of existing software won't run on ARM. A real catch-22 if ever there was one. Just as crucially, the many high performance enhancements and interface standards that currently go into making a modern x86 chip fly will also have to go into an ARM – and developing that isn't going to be cheap, although it may not take that long, since these are tried and trusted technologies that need to be applied. Still, the interest is there and Morgan Stanley expect that 10% (39 million) PCs, excluding tablets, will have an ARM processor at their heart. If true, it will make for interesting times.

Source: Focus Taiwan

[erocker]

I don't currently see the point of an ARM "desktop" at all. A lot needs to be done before it's even remotely viable.

edit: 5-10 years isn't "soon" in the computer world.

[qubit]

Originally Posted by erocker

I don't currently see the point of an ARM "desktop" at all.

Without all those issues being addressed, I'd agree. Yet, there's interest, so one has to ask why that's so.

Think about it: imagine that the ARM architecture gets the needed makeover and let's put aside that it should handily beat an equivalent x86, you will have lots of manufacturers making it. That equals lots of competition and therefore low prices (cartels aside...) Imagine getting something like a 2700K for half of its current price, or less? That would be fantastic and we would have cheap-ish, high performance processors from many sources to choose from, just like we do with other components such as memory, motherboards, PSUs etc. Ultimately, these manufacturers might be interested in this simply because it will make their boxes cheaper.

EDIT: I don't think we'll see serious competition for 5-10 years though.

[Syborfical]

Id love an alternative to x86

An arm based server would be sweet.

[v12dock]

I will believe it when I walk into walmart and buy one

[JATownes]

Originally Posted by v12dock

i will believe it when i browse Newegg and buy one

fixed!

[jpierce55]

I doubt it is for the US market. It is probably for China and countries where low power consumption is more critical.

[cadaveca]

Sounds like nVidia marketing, to me.

[newtekie1]

Note the wording, "ARM based PC". That could mean Tablets, laptops, or desktops. It doesn't specifically mean desktop. I see ARM making a big push into Tablets. I don't see them even coming close to being even a minor player in the desktop market any time soon.

[Wile E]

Ugh, here we go again.

ARM will not even be able to make any significant impact in the desktop market for years and years to come, period.

[1c3d0g]

Just like Linux...

Oi! Did I say that out loud?

In all seriousness though, everything has their place.

[tigger]

Arm is a uk company, so it might be possible with a bit of research to find out if they are working on a arm x86 alternative, though i guess if they are it will be a secret.

Just got back from work too

[option350z]

Problem being we already had a company try using an IBM RISC in the PC market. Remember the G series of Apple towers anyone? We all know what happened with that. Problem with RISC processors is that in order to do the same complex task as one of our AMD or Intel Processors can do; That RISC processor has to process more lines of simple code, which is not always the quickest route. Along with the hardware aspect, this is more suited to be a GPU architecture if anything in my opinion. Our GPUs operate similar to this. However they are different.

[Moose]

ARM are a UK company? Opinion of them improved hugely lol!

[Lipton]

This depends on how you define "desktop" or "PC" as they are very broad terms, but saying 10% would be ARM-based in the future (though no date?) isn't all too unreasonable to me. Just think of thin clients for example which I expect to increase even more with cloud computing being more and more prominent in production environment. And not to mention tablets (and to some extent smartphones) which are primarily alienating the netbooks for everyday consumption and light production.

Personally I wouldn't mind a HTPC using less than handful of watts during playback. Would it matter to the end-user if it's ARM of x86 based? I doubt it if the front-end is similar (or the same). Would power consumption matter more in the future? I'd say so, people are getting more 'aware' and the less something uses, whilst delivering the same experience, the better.

Dying to see what the first Haswell chips will bring to the table.

[MagpieX]

Originally Posted by option350z

[...] Problem with RISC processors is that in order to do the same complex task as one of our AMD or Intel Processors can do; That RISC processor has to process more lines of simple code, which is not always the quickest route.[...]

Pretty sure that's more an issue of optimization during compilation, only very rarely software is written in assembler language these days.
RISC differs from CISC in that RISC op codes are like building blocks - like LEGO blocks - forming the command and every opcode uses the same small amount of cycles to execute whereas in CISC architecture all codes may use up a certain amount of multiple cycles even empty ones for alignment and sub-operations ignoring. In general CISC architecture operates on load-execute-store-erase basis requiring the processing unit to load a value from and back to memory every time even if that one is needed again for the next command.
RISC architecture is able to store values for later use within registers reducing the amount of (redundant) memory access needed. The actual down side of RISC architecture is - as hinted in my first sentence - the bigger code size because say a command to divide two values will just be the op code for the command and two references to the operands whereas in RISC it's several opcodes with the references included. Depending on the size and optimization it might however be negligible.
Either way modern optimization techniques in hardware and software made RISC architecture's advantages minimal in comparison to CISC and RISC CPUs weren't popular not because the architecture sucks but because in case of Apple the political decision was made because IBM didn't make much progresses with their PowerPC architecture especially in terms of performance per power usage and Intel seemed to deliver.

[robal]

Originally Posted by erocker

I don't currently see the point of an ARM "desktop" at all. A lot needs to be done before it's even remotely viable.

Don't even get me started on that, because I may start trolling badly....

FFS !
Use open source solutions !

OK. ok... calming down... few breaths...

Windows 8 is going to be a major step in good direction.
Developer twats that used proprietary APIs will realise that using OpenGL, OpenCL, OpenAL, and OpenWhatever will allow them to release their products for Linux, Max, x86PC and ARM PC with little porting effort.

I think the ARM architecture is inherently better than x86 for everything, not just smartphones.
Have you seen Ubuntu running on Gumstix Overo ?
A "full-sized" desktop ARM CPU will wipe the floor with x86.

[qubit]

Originally Posted by robal

I think the ARM architecture is inherently better than x86 for everything, not just smartphones.
Have you seen Ubuntu running on Gumstix Overo ?
A "full-sized" desktop ARM CPU will wipe the floor with x86.

Yes! So true. So many so-called "knowledgeable" types have no clue of this. As I've been saying repeatedly, put the same technological "turbocharger" technologies into an ARM CPU and it will indeed wipe the floor with x86, as it's inherently more efficient and faster.

Now, this is pure speculation, but I wouldn't be surprised if Intel have made such ARM prototypes to investigate this and have found it to be true. The only reason they stick to x86 and don't embrace ARM is because of that semi-exclusive licensing deal they made with AMD.

[cheesy999]

Originally Posted by robal

Windows 8 is going to be a major step in good direction.
Developer twats that used proprietary APIs will realise that using OpenGL, OpenCL, OpenAL, and OpenWhatever will allow them to release their products for Linux, Max, x86PC and ARM PC with little porting effort.

I fail to see how Windows 8 would change it too much, since Direct X, Physx etc are mostly built into windows or the drivers, A lot of proprietary API's will remain in just as much use as they always did

A higher uptake of Linux might make them use Open standards, but windows 8 probably won't

[Thefumigator]

don't want to be a dreamer, but If AMD made the first 32 and 64 bit processor maybe they make the first x86 and ARM processor, enabling windows 8 to run both x86 apps and ARM.

I'm saying something crazy?

[AsRock]

Originally Posted by Moose

ARM are a UK company? Opinion of them improved hugely lol!

hehe, then when\if arm makes a impact MS or some body will buy them out as there is always some foreign country buying out UK business. With MS saying they will support ARM i am sure it's opened a fair few eyes of other companys.

Any linux support arm ?..

I think they might reach for nettop\laptop side of things and smaller.

[qubit]

Originally Posted by Thefumigator

don't want to be a dreamer, but If AMD made the first 32 and 64 bit processor maybe they make the first x86 and ARM processor, enabling windows 8 to run both x86 apps and ARM.

I'm saying something crazy?

No, it's not impossible. A transition layer between old and new is what's needed. The ARM is so much smaller than x86 that you could probably build both to be high performance in the one chip. Or perhaps both processors could be put on the one motherboard. The Acorn Risc PC, which used an early ARM processor did just this and it worked really well. I could run RISC OS natively and Windows 95 on a physical 486 plugged into it. Both could run at the same time, too. Very impressive for 1994.

[Wile E]

Originally Posted by qubit

Yes! So true. So many so-called "knowledgeable" types have no clue of this. As I've been saying repeatedly, put the same technological "turbocharger" technologies into an ARM CPU and it will indeed wipe the floor with x86, as it's inherently more efficient and faster.

Now, this is pure speculation, but I wouldn't be surprised if Intel have made such ARM prototypes to investigate this and have found it to be true. The only reason they stick to x86 and don't embrace ARM is because of that semi-exclusive licensing deal they made with AMD.

You put those same technological advances into an ARM cpu, and it loses it's efficiency just as quickly as x86. It's low power because it's reduced in function.

[qubit]

Originally Posted by Wile E

You put those same technological advances into an ARM cpu, and it loses it's efficiency just as quickly as x86. It's low power because it's reduced in function.

No it won't lose anything: remember, I'm not talking about making a low power ARM as in watts consumed. I'm talking about turbocharging data processing performance, like x86 has been - let ARM consume 100 watts or more! The ARM architecture is very streamlined, so will respond very well to these enhancements and I believe it will leave x86 (and x64 extensions) in the dust in every respect. Heck, the x86 instruction set wasn't terribly efficient when it came out, let alone now.

[erocker]

Originally Posted by qubit

The ARM architecture is very streamlined, so will respond very well to these enhancements and I believe it will leave x86 (and x64 extensions) in the dust in every respect.

How so?

Heck, the x86 instruction set wasn't terribly efficient when it came out, let alone now.

In what ways is it inefficient? How would you improve the x86 instruction set to make it more efficient?