Intel and Qualcomm Clash Over Arm-based PC Return Rates, Qualcomm Notes It's "Within Industry Norm"

[londiste]

That being said, ARM is now also becoming freakishly bloated as happened to X86... Maybe RISC V will make an appearance or maybe some combination of tiles a la intel.

It is not becoming bloated as such. It is getting solutions, optimizations and features that x86 architectures have largely already gone through. All this is a relatively known path although the approaches and goals may be different. Apple is a very specific example of that - they basically widened everything compared to established norms and since they own the entire ecosystem were and are in a place to make that work perfectly.

When talking strictly about core architecture the cores in state of the art ARM SoCs are already on par in terms of size with what AMD or Intel are doing. Snapdragon X Elite are in the same size range as Zen5, fairly direct comparison as they are produced on the same node. Intel's Arrow Lake cores are about the same size if not a bit smaller but they are on 3nm as well. Apple cores are slightly smaller in comparison but they are also on 3nm as well as looks like the high density variation of that.

Make no mistake - RISC V will have the same choices to make. They can stay small, simple and thus cheap as they are and leave a lot of performance on the table. Or they will follow the same path as others and gain performance along with size and complexity. Both are valid ways, just a matter of choice.

 

[phanbuey]

This isn't the first time an ostensibly superior architecture tried to replace x86. Remember Itanium? Granted it was targeted at servers and HPC instead of consumer machines, and performance of the initial models was disappointing, but architecturally it was in many ways better than x86 and even x86-64. And unlike these ARM machines the early models actually had hardware support for x86 without the need for emulation. In the end it didn't matter because x86 was so dominant, well studied, and cheap. Itanic (aka Itanium) by contrast was expensive, poorly supported and late to market to the point that x86 and others had already caught up to it. I expect ARM in PCs will go just as poorly, and this isn't even the first time they've tried it.

Every time they try it they get better though, closer. It's only a matter of time.

 

[psydroid]

this seems like one of those things where they project really hard trying to make it happen.
why would i want a machine without backwards compatibility. thats pretty much the entire selling point of windows.

I think they count on Intel fumbling it in the future, so non-x86 will be all that's left and consumers won't have much choice.

I don't see the point for Windows either and for Linux you can usually get by with much cheaper hardware, whether that's ARM or x86.

 

[Patriot]

I think they count on Intel fumbling it in the future, so non-x86 will be all that's left and consumers won't have much choice.

I don't see the point for Windows either and for Linux you can usually get by with much cheaper hardware, whether that's ARM or x86.

As soon as qualcomm finishes support for linux on these chips ill give one of them a spin.

 

[Darmok N Jalad]

As soon as qualcomm finishes support for linux on these chips ill give one of them a spin.

Curious if that will ever actually happen. If the sales are indeed sluggish, there may not be enough incentive to push it out.

 

[ScaLibBDP]

I think it's a problem that solves itself... Alot of the backwards-compatibility doesn't need to be performant, it just needs to work, and with increasing abstraction this will eventually happen. That being said, ARM is now also becoming freakishly bloated as happened to X86... Maybe RISC V will make an appearance or maybe some combination of tiles a la intel.

Definitely a chip ahead of it's time, however. They didn't spend enough time on the compatibility aspect before trying to roll out a "Premium" product.

>>...Maybe RISC V will make an appearance or maybe some combination...

RISC-V ISA follows a wrong-path of x86 ISA. That is, too many RISC-V extensions released, significant ISA fragmentation, and the most important, RISC-V fails to enter a hot consumer and HPC / Data Center markets.

Do Not pay attention to all these RISC-V Single Board Computers ( SBC ) since all them are Not an end user friendly.

A regular web-surfer will Not buy an RISC-V SBC to stay online for 4, 6, or 8 hours a day.

Despite ARM going nowhere on PC right now Holthaus still seems desperate, why are Intel CEOs so doomed?

I think that article on HpcWire could answer your question:

CPU King Intel Faces Rocky Road to Achieve GPU Dreams

Intel’s Falcon Shores GPU may meet expectations; Gaudi chips are a weak AI entry product; Interim co-CEO is still learning the Xeon product line. These were some of the grim […]

www.hpcwire.com