News Intel Ponders Transition to 64-Bit-Only x86S Architecture

[truerock]

VMware is a better option in my opinion than virtualbox, virtualbox limited with 256/128MB max graphics memory while VMware Workstation supports up to 8GB graphics.

However some aggressive DRMs blocks playing games in VMs if thats your thing.

I remember Virtual PC software, used back in the day to install Windows 7 and Windows XP mode, it was real useful and awesome.

Microsoft murdered that software and stopped supporting it, that's what Microsoft do these days since then. Like Windows Media Encoder and MSN Messenger both awesome software back in the day and all MS did to them is stop supporting and removed download links. Nothing good comes out from MS except Windows XP and Windows 7.

I've supported a lot of VMware servers - but, VMware never seemed to have very good support for things like DOS and Windows 3.1 - in my opinion.

 

[Findecanor]

Oh, that'll never happen. It would break waaaaaaaaaaay too much software. That horse is already out of the barn. Currently, stack addresses are interchangeable with heap addresses. You can't just make them exclusive, now.

I'm thinking specifically of the Safe Stack feature in Code Pointer Integrity, which uses two stacks. One that is in the same address space as heap addresses and one that is safe, kept hidden.
It is the default on Google's new Fuchsia OS, and I think it has been deployed on FreeBSD as well.
But just as a software-based solution, it is dependent on address space randomisation for safety, which is why a hardware-based protection is preferable.

Obviously, software would have to be compiled anew to use it. But that is a solved problem. There are many security projects out there who have successfully disassembled existing software, added their instrumentation and reassembled.
The only insurmountable problem is with programming languages that have its own incompatible stack handling, such as Go.

A Shadow Stack protects only the return address, not function pointers on the stack nor any variables that are used for control-flow decisions. And even with Intel's hardware solution: CET, it suffers from the same problem that programs have to be re-compiled to use it without crashing. For instance, exception handling breaks the normal model, and has to write to the Shadow Stack using special instructions.

 

[ikjadoon]

Addressing modes are something the CPU needs to implement support for, in a performance-critical part of the core. I could believe there might be some slight efficiency improvements by not having to mux paths for 16-bit addressing and 32-bit mode.

Slight efficiency gains: perhaps? Unfortunately, Intel does not mention any efficiency improvement, as I would have expected: only reduced complexity. Reduced complexity can help efficiency, but not necessarily. Intel is not a company to sell itself short.

And if they aren't talking efficiency: margin of error, no gain, or they don't even know yet because the uArch that would launch as X86S hasn't been completed.

Arm, when they released an actual 64-bit-execution-only CPU core, noted sizeable benefits (warning: auto PDF download). But Intel's proposal is not 1% of what Arm targeted: Intel 64 / AMD64 CPUs still need to fully accept + handle all kinds of 32-bit processing.

According to them, the modes they're using aren't even used by modern operating systems. The only buy-in would be from the UEFI firmware developers, and their response would likely be a big "thank you", for making their lives just a little bit easier.

After the bring-up, then the OS isn't needed, IIRC. Intel explains current Intel 64 / 64-bit OSes are incompatible + the OS will need to be reworked:

While running a legacy 64-bit operating system on top of a 64-bit mode-only architecture CPU is not an explicit goal of this effort, the Intel architecture software ecosystem has sufficiently matured with virtualization products so that a virtualization-based software solution could use virtualization hardware (VMX) to deliver a solution to emulate features required to boot legacy operating systems.

//

These modifications can be implemented with straightforward enhancements to the system architecture affecting the operating system only.

Confirmed on page 10 of the PDF. X86S is a new CPU and OS ISA.

You don't strictly need AMD buy-in. It's up to AMD whether they want to follow.

Unless this is a revolutionary change (and it doesn't seem like one), AMD + Microsoft are likely key for X86S to actually launch. Intel doesn't have the same sway in commanding Microsoft as it did in the early 2000s."

It's likely why Intel is asking for ecosystem comments: they know X86S will not be the next version of x86, unless the x86 OS + CPU ecosystem moves forward together.

Even Intel's hybrid cores required Microsoft's buy-in during Windows 11 development.

Not really. The motherboard firmware is what has to deal with the 16-bit -> 32-bit -> 64-bit mode transitions, and that's already CPU-specific.

I meant OSes supporting two ISAs, as Windows 11 today isn't compatible. The OS will need to support any new ISA.

I might be wrong, as I don't know much else beyond what Intel claims.

Maybe this is a trivial update and it'll take no time to implement. Maybe it's such a no-brainer that Microsoft and AMD will commit to within a month's time. Maybe it'll even launch later this year in Meteor Lake and Microsoft will ship an updated Windows 11 build to add compatibility, AMD be damned.

But I'm not counting Intel's chickens yet until we hear more progress, particularly on comments received.

 

[ikjadoon]

Because x86 computers as you and I understand them to be* still boot as if it were an IBM 5150 PC. I believe what Intel wants to do is get away from this for a couple of reasons:

• It simplifies firmware and boot process. It could also simplify the OS code, depending on how much it still has to mess around with setting up the CPU mode, such as flipping a gate open to resolve the A20 Line issue. Simplifying code should in theory lead to less bugs and other issues.
  • Overall, it also eases the development process for x86 based systems and hardware itself. I'm sure AMD would also love to move away from this.
• There's legacy hardware that in order to support it, caused Intel (and possibly AMD) to do a few things as workarounds, at least one of which led to a security problem. Legacy modes either don't have a concept of security (especially the initial mode x86 CPUs boot in) or are limited to what it can do.
  • One issue was when Intel introduced the Advanced Programmable Interrupt Controller (APIC) to replace the original interrupt controller. But they didn't actually replace the original interrupt controller, because backwards compatibility is king in IBM PC compatible systems. The work around to allow usage of the older interrupt controller caused a vulnerability known as the x86 Memory Sinkhole (it was fixed internally after Sandy Bridge CPUs)

* I say "x86 computers as you and I understand them" because there are x86 computers/systems that aren't IBM PC compatible: the PlayStation 4/5. You actually can't boot a bog standard x86-64 build of Linux on at least the PS4, because the OS is expecting legacy hardware/features that the PS4 lacks. However, once you modify Linux to not look for those, it runs most x86 applications as if it were a typical PC (see

View: https://www.youtube.com/watch?v=QMiubC6LdTA
)

That is quite interesting. I don't disagree that X86S brings simplification & reduced complexity: it is wacky how we boot up CPUs.

However, my assumption: this will take a not-insignificant effort by all parties, yet the benefits will be minimal and can be otherwise surmounted with alternative solutions.

Why I think it'll take more effort: X86S needs "ecosystem comments" because it can't / wasn't mutually agreed-upon by AMD, Microsoft, Linux, etc. already (all have official, private channels), etc.

Now, I hope AMD + all modern OSes back X86S, if it's as unequivocally beneficial & feasible as Intel suggests with only backwards compatibility with "legacy 64-bit OSes" as its primary drawback.

It's that I'm more pessimistic Intel, AMD, Microsoft, other OSes etc. will see eye-to-eye on implementation, timelines, compatibility, future revisions, etc. See Pluton, AVX-512, even MMX's branding alone led to a lawsuit between AMD & Intel.

But, I hope it's as trivial as we're imagining: just needs a few checkboxes and away we go to X86S.

 

[ikjadoon]

When Arm actually removed 32-bit execution completely, they found notable benefits (from the PDF above)

Intel's X86S proposal, however, won't come with these benefits as 100% 32-bit application + OS compatibility remains and x86 doesn't have the same drawbacks / opportunities as Arm re:32-bit.

And even then, we didn't see a revolutionary change in the updated cores, though it's fair to say both X86S should be given a few generations to tease out its improvements.

 

[hotaru.hino]

When Arm actually removed 32-bit execution completely, they found notable benefits (from the PDF above)

Intel's X86S proposal, however, won't come with these benefits as 100% 32-bit application + OS compatibility remains and x86 doesn't have the same drawbacks / opportunities as Arm re:32-bit.

When you consider modern x86 requires supporting 5 operating modes, Intel's looking to get rid of 3 of them, and usage of 2 of those modes is virtually zero, I'd say that's still a huge step towards the correct direction.

And even then, we didn't see a revolutionary change in the updated cores, though it's fair to say both X86S should be given a few generations to tease out its improvements.

This is going to be a problem with the consumer facing side of things. Mostly because if something doesn't directly benefit them, or they see there's no feature updates or speed improvements, they won't care.

They don't understand that hardware companies have at least two customers: the end user and the developer.

 

[kerberos_20]

This is going to be a problem with the consumer facing side of things. Mostly because if something doesn't directly benefit them, or they see there's no feature updates or speed improvements, they won't care.

They don't understand that hardware companies have at least two customers: the end user and the developer.

if we talking about x86, it shouldnt affect consumer that much unless he wants to run some old operating system (say DOS for bios update), he will just have to switch to uefi shell which can pretty much replace DOS for these basic tasks
or 32bit operating systems which is gettin phased out anyway, win 10 was last and linux would follow suit later if x86s kicks in
64bit os still can run 32bit apps same way as its doing now

as for developers...most developers doesnt work with cpu registers, and those they do, im not seeing how that would affect them

the only thing that will need to be adressed by OS seems to be just ring level changes
x86 has four ring levels
x86s only two

 

[hotaru.hino]

if we talking about x86, it shouldnt affect consumer that much unless he wants to run some old operating system (say DOS for bios update), he will just have to switch to uefi shell which can pretty much replace DOS for these basic tasks

What I meant is people who don't really understand what's going on are going to go into this kicking and screaming. They expect every major hardware change to make computers 1000x better or something.

For example: Windows 11's TPM requirement. They get FOMO. They want to upgrade to the latest and greatest now. Oops, can't use it on their hardware. Said feature doesn't improve the user experience in anyway. Windows 11 sucks now and TPM is dumb.

 

[Dobesov]

didnt intel already try this with IA64? and how it went, tho back then 32bit was dominant

Nah, Itanium was a totally new architecture and instruction set, the issue wasn 32 vs 64, but that it didnt run ANY mainstream software, also turned out the whole thing was complicated and slow. What they are proposing here is keeping the 64 bit x86 architecture and ditching the 16bit and 32 bit modes. This would cut down on chip complexity, lower heat, lower price. Most regular users would never notice because everything they run is 64bit. 16 and 32 bit modes could easily have a software solution if you needed it, and would likely be a little faster than pure emulation, since we are talking about instructions from the same family. Plus, 16bit and 32bit code is now so old that none of that software would expect to have modern speed anyway. This would essentially kill real DOS compatibility... no more installing FreeDOS on real hardware, but who does that on a new machine anyway? Maybe it will finally force a FreeDOS-64 😉

 

[Dobesov]

Agree with this. Hope it happens. AMD
Should also

I'm surprised they have not already. Why on earth does the XBOX and Playstation have a 16bit and 32bit compatible architecture? Cutting that could have saved money and power and they would already be ahead of the curve when ready to release it on consumers.

 

[user7007]

Sounds good. I think many people misunderstood. They're suggesting removing things nobody or next to nobody is using. As others have said, you can't properly run an old OS on new hardware because they lack hardware support. You'll be running a new Linux kernel or windows 11 etc. And these operating systems don't need or use the legacy features they want to remove. If you need to run an old OS on new hardware it's normally emulated in some way and that means they can emulate the few missing bits.

 

[Eximo]

Also fun fact, you can run 16-bit apps in Long Mode (the 64-bit mode in x86-64), but the only reason why you can't in Windows is because Microsoft didn't include any 16-bit compatibility shims like they did with 32-bit apps (also probably because they figured nobody was using 16-bit apps)

Tell that to all the people I worked with when we switched Windows 7 32 bit to Windows 8.1 64 bit. Now I had done a lot to clean up old applications when we switched from XP to 7 but there were so so many 16 bit apps they didn't tell us about. Poop really hit the fan when be banned Windows XP from joining the network. All the hidden laptops that people kept to run their old stuff stopped working. Can only warn people for so many years before you have to pull the bandaid off.

 

[danny009]

"removes outdated" "legacy reduced"

Intel deciding what to buy for a long time, it is time to get some regulations from tech officials to put an end to this. Remember when back Intel claimed "4 cores all you need" then switched otherwise because tech goes to "other" direction? Big tech simply have no right to tell me what to buy and what to choose, if they want my dollars, they should do things what consumers really want.

Give us options, give consumers options more than 1 or 2. They have billions, anything is possible todays tech, so do it?

Consumers also should stop buying new hardware all the time. Stop jumping into hype wagons in my opinion, my computer nearly 5-6 years old perhaps even more than that, and it works fine and can run many games including triple AAA games assuming keep your expections in check.

Play games in 30-45fps for once, 60+fps doesn't grants you special rights, it does not make one person "better" than others in tech. Sure it is your dollars, but serve to a bigger and friendlier cause. See the big picture.

By buying new Intel/AMD/NVIDIA things almost every year simply because it have more stuff inside results in broken industry for everyone.

No offense included to anyone in this post

 

[hotaru.hino]

Big tech simply have no right to tell me what to buy and what to choose, if they want my dollars, they should do things what consumers really want.

I'm a customer, and I don't care for legacy modes. Ergo, they're listening to what I want.

If you don't like what they're doing, you're free to not buy their products. Nobody's holding a gun to your head and forcing you to buy them.

Give us options, give consumers options more than 1 or 2. They have billions, anything is possible todays tech, so do it?

And are you actually going to buy it? Is enough people going to actually buy it? There are a lot of things people said they'd want if someone would just make it, but when someone actually delivered, it didn't take off for some reason or another.

Consumers also should stop buying new hardware all the time. Stop jumping into hype wagons in my opinion, my computer nearly 5-6 years old perhaps even more than that, and it works fine and can run many games including triple AAA games assuming keep your expections in check.

As far as I can tell, the average consumer doesn't actually buy new stuff all the time. I know people who kept their phones for 5 years. Who still have the PC they bought or built 5 years ago. And in this economy, people are going to hold onto their stuff longer because they have bigger fish to fry.

Play games in 30-45fps for once, 60+fps doesn't grants you special rights, it does not make one person "better" than others in tech. Sure it is your dollars, but serve to a bigger and friendlier cause. See the big picture.

But what if I want 60FPS+? You can't tell me what to do

By buying new Intel/AMD/NVIDIA things almost every year simply because it have more stuff inside results in broken industry for everyone.

I don't know anyone who actually does this. At least, anyone who doesn't wipe their butt with $100 bills.

 

[kerberos_20]

"removes outdated" "legacy reduced"

Intel deciding what to buy for a long time, it is time to get some regulations from tech officials to put an end to this. Remember when back Intel claimed "4 cores all you need" then switched otherwise because tech goes to "other" direction? Big tech simply have no right to tell me what to buy and what to choose, if they want my dollars, they should do things what consumers really want.

Give us options, give consumers options more than 1 or 2. They have billions, anything is possible todays tech, so do it?

you want to keep support for 16bit or 32bit operating system? you better stock up some tech before it hits museum

 

[railtrace]

Is this just Itaninum all over again? Even if it isn't, there's no point if there's no performance improvement.

 

[SomeGuyonTHW]

So as per the technical theory, a direct 64-bit reset state will now eliminate several stages of trampoline code to enter 64-bit operation, since booting CPUs (SIPI) starts in real-address mode today and needs a 64-bit replacement.

Intel also says that in this proposed architecture, it is possible to switch to 5-level paging without leaving a paged mode. So unnecessary legacy modes are now removed, and should lead to faster systems in the future ??

We will see how this works out though. Not very optimistic about this.

Very insightful and nerdy, thank you.

 

[richardvday]

didnt intel already try this with IA64? and how it went, tho back then 32bit was dominant

It might seem like that is the same thing but they didn't just drop some modes or instructions.
It was basically the opposite of Risc. It used very long instructions(Vliw) and the onus for scheduling and other things the cpu does now were shifted to the compiler.
Not at all the same thing.
That's a very basic description google it for a more comprehensive explanation.
This would only lose real mode etc..
No more 8bit, 16 bit or 32 bit modes.
The 8bit and 16bit have been deprecated since the 90s at least.

 

[bit_user]

It was basically the opposite of Risc. It used very long instructions(Vliw) and the onus for scheduling and other things the cpu does now were shifted to the compiler.

Hmmm... I wouldn't quite describe VLIW as the opposite of RISC. It's basically just bundling RISC instructions into long packets that are designed to be consumed synchronously by a superscalar dispatcher.

The opposite of RISC is CISC, where you have lots of specialized instructions that can each involve subtasks like loads, stores, and address computation. x86 is definitely CISC, although I've heard there were some VAX CPUs that were even worse.

No more 8bit, 16 bit or 32 bit modes.
The 8bit and 16bit have been deprecated since the 90s at least.

There never was such a thing as "8bit mode". x86 has 16-bit segment:offset addressing, 16-bit, 32-bit, & 64-bit Protected mode, and 32-bit Real mode. I don't know if there's such a thing as 64-bit Real mode...

As for deprecation, I'm not sure any of those modes were formally deprecated. They were obsolete, at least to the extent that operating systems (beyond DOS) wouldn't use them.

 

[TerryLaze]

I've supported a lot of VMware servers - but, VMware never seemed to have very good support for things like DOS and Windows 3.1 - in my opinion.

You need something like dosbox and PCem that let you choose the VGA and sound card, amongst many other things, for every app or config so that every software finds the hardware it expects, PCem even lets you use any bios you want if you can source the bios dump.

you want to keep support for 16bit or 32bit operating system? you better stock up some tech before it hits museum

By the time the ryzen 7000 and the intel 13th gen end up in museums we will be in the retirement homes, if not pushing up daisies.
That's if these gens will be the last ones which I doubt very very much.

Is this just Itaninum all over again? Even if it isn't, there's no point if there's no performance improvement.

If this will make CPUs cheaper to make than that's worth it to intel, if they get smaller so they can make more CPUs each year than that's also worth it to them, if the CPUs use less power because they have less tech than that's also worth it to them. If all of that makes intel CPUs more attractive to high end smartphones and ultra-portables that hit a performance wall with ARM cpus than that's hugely worth it for them.
You might not care and that's ok, but there definitely is a point to do it.

 

[bit_user]

If all of that makes intel CPUs more attractive to high end smartphones and ultra-portables that hit a performance wall with ARM cpus than that's hugely worth it for them.
You might not care and that's ok, but there definitely is a point to do it.

Oh, Intel wouldn't dare try to push x86 into the cell phone market, again. Even suggesting such a thing is probably enough to seriously damage your credibility, in the halls of Intel HQ.

Certainly, Intel is trying to fend off an assault on the laptop market by Apple, Qualcomm, and ARM (mostly via Mediatek). Then, there's the server market...

 

[TerryLaze]

Oh, Intel wouldn't dare try to push x86 into the cell phone market, again. Even suggesting such a thing is probably enough to seriously damage your credibility, in the halls of Intel HQ.

Certainly, Intel is trying to fend off an assault on the laptop market by Apple, Qualcomm, and ARM (mostly via Mediatek). Then, there's the server market...

Read up on what IFS 2 is all about.
They are working with ARM to give 3rd parties the possibility to make ARM CPUs on intel foundries.
This could all just be to give 3rd parties a cheap-ish low-ish power x86 core for their designs, Single Board Computers have really taken off the last years and this could make them easier and cheaper to make, these cores would also be perfect for their FPGAs that still run ARM cores.

It won't be intel pushing for the phone market and you just saying that shows how out of it you are and how little attention you give to what the other persons are saying.
The high end smartphones and ultra-portables was an IF all of that makes intel CPUs more attractive, not an Intel will do that.

 

[truerock]

You need something like dosbox and PCem that let you choose the VGA and sound card, amongst many other things, for every app or config so that every software finds the hardware it expects, PCem even lets you use any bios you want if you can source the bios dump.

By the time the ryzen 7000 and the intel 13th gen end up in museums we will be in the retirement homes, if not pushing up daisies.
That's if these gens will be the last ones which I doubt very very much.

If this will make CPUs cheaper to make than that's worth it to intel, if they get smaller so they can make more CPUs each year than that's also worth it to them, if the CPUs use less power because they have less tech than that's also worth it to them. If all of that makes intel CPUs more attractive to high end smartphones and ultra-portables that hit a performance wall with ARM cpus than that's hugely worth it for them.
You might not care and that's ok, but there definitely is a point to do it.

I use DOSbox and PCem for specific applications they work well with. My point was, I really, really would like to have PC emulators that were as good as Microsoft Virtual PC 2004 (for 16-bit DOS and Windows 3.1) and Microsoft Virtual PC 2007 (for 32-bit Windows 95, Windows 98 and Windows XP). I just really enjoy tinkering in those environments.