Intel Continues Prepping The Linux Kernel For X86S

[Gonk]

The 8945HS is kind of a weird product, mostly because of AMD's naming scheme. 8945HS is an 8C/16T part and the 7945HX is an 16C/32T part, both Zen4 and both 780M gpu. Naming would imply that the 8000 part is the higher end, but it is actually a rebranded 7940HS. The 7945HX has 13B transistors which would imply that the NPU takes up most of the budget on the 2 other chips. Gotta love marketing departments.

Don't blame AMD's marketing department (at least not in this particular case). Generation rebadging is a demand imposed by the big OEMs. When AMD/Nvidia/whoever releases a new generation thing, eg. a B700 (upper-mid range of the B generation - upper-mid to mid-high being the usual first in generation part), OEMs immediately demand the B equivalent of every A thing they have been shipping in their desktops - even the $75 A150. They don't care that AMD/Nvidia/... won't have an $75 B part ready for two years, they are convinced that consumers won't buy a desktop with any A gen part now that there is a single B gen part on the market. So they tell AMD/Nvidia/... "look, just give us an A150 that says it is a B150 so we can sell a full line of new and improved desktops". And if AMD/Nvidia/... releases a C gen before they get around to making a low power, $75 B part, the OEMS will demand an A150 rebadged to C150. In a perfect world the part makers would refuse to play this game, but when Dell says "give us a B150 now or we will switch to <parts competitor>", they have to oblige.

[Artim]

Lot's of embedded systems use Dos. However, they will probably not run very new hardware.

Then it may just be time to move on when they eventually run out of x86 processors. Either to a different architecture or a different OS (or at least to something like a rudimentary Linux with a QEMU emulated DOS on top).

[dragon321]

To all those supporting x86S, I am not saying that it is worse than x86_64. I am saying that, both ISAs are inherently bad (I can elaborate that, if people are interested in it), compared to other architectures. So, the only reason to use/implement any of the two, is backwards compatibility. If you are willing to sacrifice that (I am all in for that btw), why go for x86S and not something better?

X86S is not sacrificing compatibility. The only legacy x86 thing that matters nowadays is 32 bit application support and that isn't going anywhere in X86S. Why removing 16 bit and 32 bit operating system support would have any significant impact on x86 users where every widely used operating system was ported to 64 bit years ago? X86S changes nothing to probably something like ~99% of x86 users.

[r_a_trip]

If you are willing to break backwards compatibility, you should go for something other than x86S.

From a technical standpoint you are absolutely right. Something more elegant than X86S would make sense. Except, reality is not elegant. It is messy and it is chock full of x86 and x86_64 software.

Intel tried with IA64 and that turned out to be a disaster. Nobody was in a hurry to refactor their software in IA64 and Itanium's built-in x86 emulator ran like dog shit. AMD saw the writing on the wall for their business if IA64 took off, so they scuttled Intel's chance on a new market by heading them off with x86_64.

Software vendors didn't have to refactor anything, because x86_64 ran their wares unmodified, even if it was 32 bit. The only things that needed to be refactored were Operating Systems. X86S is just removing support in silicon for 16 and 32 bit Operating Systems. Making the complexity reduce by scrapping stuff that isn't needed anymore. It will still run 32 bit and 64 bit user space software as if nothing happened.

Blame the installed base of 32 bit and 64 bit x86 software for keeping x86 around. Nobody is eager to redo millions of man-months worth of software just to move to a more elegant architecture. X86S runs our old legacy crap for a few decades longer? Adopt it and we can eek out a few billion dollars more.

[Phoronos]

We are still stuck with X86 and X86_64 for many many many years because of backward compatibility,
The only other possibility is to create something similar from SCRATCH and AT LEAST AS FAST and which can EMULATE X86 X86_64 for old softwares.

[S.Pam]

Then it may just be time to move on when they eventually run out of x86 processors. Either to a different architecture or a different OS (or at least to something like a rudimentary Linux with a QEMU emulated DOS on top).

I would agree. But you would be surprised at how many huge and expensive systems are based on old code, and replacing those are multi million dollar projects. We are still using OS/2 for some facility maintenance systems at my workplace. We also have 386 and 486 based manufacturing machines that does their job well. Investing in new equipment is not an easy sell.

[drakonas777]

For the absolute majority use cases dropping pre-AMD64 stuff is fine, at least in HPC SKUs. Embedded SoCs may preserve legacy modes for longer.

You may jerk off your geek fetish around RISC as much as you want, but x86 is here to stay for foreseeable future.

[NotMine999]

X86_64 is a shitty architecture with 1 redeeming quality, backwards compatibility. X86S looks like a shitty architecture with 0 redeeming qualities...

Intel Itanium all over again. GEEZ

[SigHunter]

but why is it not called x86s_64 or x64s or whatever? x86s gives the impression that it is 32 bit if you otherwise know nothing about it

[marios]

I might have misread the spec. 32 bit x86 userspace seems to run on x86S. I am not certain what happens with the few changed instructions though. So people pointing out that fact, are probably correct, at least for most applications.

I cannot disagree more with others though. Those calling RISC a "geek fetish", should take a look at aarch64, that is RISC and dominates the low-power devices and also has some strong presence in HPC (i.e. Fungaku). And aarch64 is only one architecture that rivals x86.
Also the fact that most performant CPUs are x86, does not mean that x86 is a good ISA. First of all, good luck dismissing A64FX, Ampere Altra, M*, Neoverse etc as worse than their x86 counterparts of the same era. Secondly, a fast CPU can be created, based on a bad ISA, if you throw a shiitton of money at it. This is what Intel and AMD are doing for a long time. It seems like some others have started throwing shittons o money on other ISAs, and they are quickly catching up (who knows, soon they might mop the floor with x86 performance-wise). Just to be clear I am just stating the fact, I am not supporting ARM or other companies that implement their ISA.