Intel CEO Gelsinger dismisses 'pretty insignificant' Arm PC challenge Intel CEO Pat Gelsinger has downplayed the threat of rival chipmakers creating processors based on the Arm architecture for PCs. "Arm and Windows client alternatives, generally they've been relegated to pretty insignificant roles in the PC business," he told analysts during the x86 giant’s Q3 earnings call Thursday. "We take …
He has to say that or the stock crashes and he loses a lot of net worth. ARM is undercutting the PCs the way PCs undercut the mainframes. Forty years ago IBM was saying PCs were no threat to their business. How many people are buying IBM's mainframes now?
On balance, I believe in about half of this statement, which is the part where Gelsinger is saying what he needs to say to calm the shareholders, who are like easily-spooked sheep.
Whether that signals actual weakness on the part of Intel is a different thing altogether. The Arm fanboys desperately want Arm on the desktop and in the data center to happen, and who knows, it might. Intel has a tremendous advantage in technology and business relationships, so I think it's beyond premature to count them out.
He's trying to sound confident, whatever he may actually think. However, history isn't quite as one-sided as you paint it. ARM has always been in phones, because it has always been power-efficient. Intel trying to break into phones is taking X86 to ARM's home front. ARM trying to take the desktop space is the opposite, and there is reason to expect that it will have similar troubles that X86 did when it tried to go mobile. After all, AMD was even less successful at producing mobile-capable chips. While Intel didn't really succeed at getting its chips into mobile phones, they have had a lead over AMD in the low-power low-performance space for quite a while.
That's no guarantee that ARM will continue to fail as it attempts to take the desktop, but keep in mind that this is not the first time it's been tried. The previous attempts have resulted in some prominent failures. We still see a lot of people dismissing Windows on ARM, even though it's been much more successful than I thought it would when it first came out. I think a lot of the attention is here specifically because of Apple's successful switch, and that has been quite impressive, but that's no guarantee that Qualcomm or someone else will be able to successfully copy it or that the market will care enough to adopt it.
I don't even care about Windows support, I would gladly purchase an ARM-based laptop that could match concurrent (say Zen4) x86 price & performance and allow running Linux natively, just for sake of being geek enough to want to tinker with one.
However, having so far not seen one to emerge to market makes one a bit skeptical if that shall appear now. I am not holding my breath.
The best ARM PC-alikeish offering as of currently is Apple M, yet even that is more expensive and slower than Zen4 x86 and does not really well support native Linux installation. Then there's been a bagful of Chromebook-and-similar laptop-like products based mostly on Qualcomm chipsets, yet those have either been too slow or too expensive to be an appealing alternative for x86.
Are there any Linuxes on ARM yet where the OS is actually installed, vs. imaged ('dd' or similar) onto the system disk?
Perhaps this isn't a big issue for most people, but it seems to me that being able to determine your own filesystems, package selections and so on during installation, rather than accepting someone else's choices and defaults in a pre-built image, would be desirable at least for non-hobbyist users.
I'm not sure where Windows on ARM stands in that regard either. Though since most Windows people presumably end up running whatever the PC came with, particularly on desktops and laptops, perhaps they care even less than the hobbyist Linux folks might.
There is Windows on ARM installation media which works basically the same as Windows on X86 installation media. It's a different disk, but as far as I know, it will give you the same result. There are some distros that have made ARM installation disks, but the reason they're not common is that most Linux on ARM images won't run on anything else. The people making the media would have to make a hundred different versions for each board it's going to run on, and they get bored first. There is general-purpose media out there for anything that uses a standard boot system. The list of devices that actually uses that system is... nothing. Okay, it's probably not nothing, but none of the stuff you're ever likely to buy. Meanwhile, Windows isn't designed to run on most ARM devices, and only on Qualcomm or rebadged Qualcomm chips, so they can more easily limit their target to something that will at least use Microsoft's standard.
"Nothing you're ever likely to buy" sounds about right; and I suspect until that list of standard boot system devices is not "nothing", or at least mostly not "nothing", uptake will continue to mostly be by the hobbyists.
Which is fine -- lots of fun and even useful things started out as someone's "let's see what it can do" project. But it's a different audience/market from people (and companies) deploying Linux on Intel.
Personally I'd love for there to be some kind of "ARM reference platform" or something, with whatever bus definitions, boot system, even mundane things like board form factors and port cut-outs, specified to enough of a degree that OS and firmware porters and developers have a blueprint of sorts to follow. 'dd'-ing someone's notion of a Linux or BSD image onto an rpi boot disk is OK, but seems a bit limited.
This is excessive praise for the Pi 5, and by doing that, you're harming its image. The Pi 5 has a faster processor than the Pi 4, and certainly for some uses it is enough processing power, but it is not remotely similar to X86 chips. Even when just considering the processor, the Pi's using four Cortex A76 CPU cores, which is not really the height of speed that ARM has available. For context, the popular chip for use in SBCs that aren't made by Raspberry Pi is the Rockchip 3588, which has four A76 cores at the same clock rate as the Pi does, but also another cluster of four A55 cores. When compared against the low end of Intel's chips, it is slower on nearly every benchmark but more power-efficient. The Pi's CPU is slower than the 3588 and, since it's made on a 16 nm fab, whereas the 3588 is made on an 8 nm one, the Pi doesn't even get those nice power consumption figures that the 3588 does.
This isn't even including the Raspberry Pi's slower storage, which has a large effect on perceived speed when used as a desktop. By pretending that these things are equal, you're making the Raspberry Pi compete in an area where its designers didn't intend it to. The designers didn't make the Pi 5 to be as fast as an X86 desktop, which is why they went with a relatively basic quad-core A76 SoC rather than getting a more modern set of cores on a better fab. Qualcomm's desktop and laptop chips are much more powerful, coming with eight cores and usually using ones that aren't already five years old, though that varies. If you pretend it's something it's not, then you run the risk that users are put off when they find that it doesn't do what you suggested it could and assume that it's much less powerful than we know it to be.
I'd compare it to Netbooks. I liked the concept of a really small and light laptop when they came out, but when people tried to make it do everything a large laptop could do rather than consider that it's low performance was useful to making it small, light, and with acceptable battery life, they found that it fell short. Their subsequent requests effectively killed the model. If you're using a Raspberry Pi as a desktop, then you have to know what it does and what it doesn't do, in which case you'll have no trouble using it as you'd expect. If you try to make it do everything, you'll find some of those things it doesn't do well.
I agree with everything in your post, though I would argue the trouble with netbooks wasn't so much consumer expectations, as the forcing of Windows by manufacturers onto devices that just weren't meant for it, and lest we forget browsers are probably the most memory hungry applications around, meaning lots of memory swapping.
IIRC the original Netbooks had a lite Linux offering and eMMC which whilst slow, was not as a slow as spinning rust dealing with memory swapping. I remember borrowing a Windows 7, hard drive based Netbook for a holiday and my goodness, it was as good as unusable.
Any of the RK3358 boards with 32GiB of RAM I would say are definitely a viable option to replace most low end x86 desktops.
And then there are the very high end ARM workstations, with up to 768 GiB of DDR4 RAM and up to 128 2.6 GHz Arm v8.2 64-bit cores, which make even the latest high end Apple computers look slow.
But that kind of number crunching power is not cheap e.g.
https://www.ipi.wiki/pages/ampere-altra-developer-platform
Youtube video: 128-core MONSTER: the Mac Pro Apple SHOULD have built!
Pay more and you can even get 3GHz cores,
https://amperecomputing.com/briefs/ampere-altra-family-product-brief
This "pretty insignificant" competitive threat held the position as the top supercomputer in the Fujitsu design, and dominates mobile. The majority of this is fabbed at TSMC, who also dominates Intel.
ARM1 was 25,000 transistors, while the 80386 was 290,000 for a lesser-quality machine.
Does this guy need new glasses?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
