I just want a hybrid x86-64 CPU with some ARM64 cores on one chip. Maybe throw some RISC-V cores on it too, and then we can run everything.

Quoting: LoudTechiehttps://www.worksonwoa.com/games/

Microsoft is trying to run it's own protonDB alternative for their switch to ARM and :shock: anti-cheat is the biggest hurdle.
They've emulators instead of Wine(windows)/Waydroid(android)/Darling(mac) and they've already ported Microsoft office, but the story is pretty similar.
I'm curious do their emulators also emulate the secret instructions Intel seems to be including specially for them?
I hope their push to ARM is long and painful, but in the end succeeds.
:grin:

I was a Windows Kernel dev for Intel. Secret instructions don't exist. However, there are Intel instructions that took years for Microsoft to finally accept the patches. Also, you can write the patch, do all the testing, once you hand the patch over to the Microsoft Engineer it's up to them with what happens. If there is no immediate benefit, Microsoft wouldn't add support until years later when the instructions became beneficial. Sometimes Microsoft would ask for features in upcoming processors. When those features finally get implemented and then we would start writing support into the Windows kernel so we could be ready with patches when the chips get to the tape out stage. If the feature only gave a minor improvement Microsoft would just drop the support. Like, "Oh it only added 30 minutes of batter life? That's too bad. Thanks for trying." To be fair, somethings add so much complexity it's not worth carrying around the debt in perpetuity. Same thing happens to Inbox Drivers.

Quoting: Linux_RocksI just want a hybrid x86-64 CPU with some ARM64 cores on one chip. Maybe throw some RISC-V cores on it too, and then we can run everything.

External Media: You need to be logged in to view this.

It's an interesting thought experiment, if done correctly in theory it's possible, though it might have to be spread across multiple chips. Kind of like this:
https://psdevwiki.com/ps3/images/c/cd/PS3_-_COK-001_-_labels.jpg

Quoting: BlackBloodRum

Quoting: Linux_RocksI just want a hybrid x86-64 CPU with some ARM64 cores on one chip. Maybe throw some RISC-V cores on it too, and then we can run everything.

External Media: You need to be logged in to view this.

It's an interesting thought experiment, if done correctly in theory it's possible, though it might have to be spread across multiple chips. Kind of like this:
https://psdevwiki.com/ps3/images/c/cd/PS3_-_COK-001_-_labels.jpg

just a thought experiment:
shouldn't it be enough to have a marker included in the code which tells the CPU which instruction set you want to run and then the correct translation unit is activated that processes them? isn't it: current CPUs don't implement the actual instructions anymore but convert it to micro-code / µ-ops which then are actually executed... so it's all risc CPUs nowadays... having multiple cores/CPUs of which only a few are used sounds like a huge waste of money / space / energy (they use power also in standby) / etc.

Quoting: Highball

Quoting: LoudTechiehttps://www.worksonwoa.com/games/

Microsoft is trying to run it's own protonDB alternative for their switch to ARM and :shock: anti-cheat is the biggest hurdle.
They've emulators instead of Wine(windows)/Waydroid(android)/Darling(mac) and they've already ported Microsoft office, but the story is pretty similar.
I'm curious do their emulators also emulate the secret instructions Intel seems to be including specially for them?
I hope their push to ARM is long and painful, but in the end succeeds.
:grin:

I was a Windows Kernel dev for Intel. Secret instructions don't exist. However, there are Intel instructions that took years for Microsoft to finally accept the patches. Also, you can write the patch, do all the testing, once you hand the patch over to the Microsoft Engineer it's up to them with what happens. If there is no immediate benefit, Microsoft wouldn't add support until years later when the instructions became beneficial. Sometimes Microsoft would ask for features in upcoming processors. When those features finally get implemented and then we would start writing support into the Windows kernel so we could be ready with patches when the chips get to the tape out stage. If the feature only gave a minor improvement Microsoft would just drop the support. Like, "Oh it only added 30 minutes of batter life? That's too bad. Thanks for trying." To be fair, somethings add so much complexity it's not worth carrying around the debt in perpetuity. Same thing happens to Inbox Drivers.

I was referring to the undocumented instructions found in the Windows kernel.
Thanks for the assurance that those instructions don't seem to be kept secret with the goal to keep back potential Windows competitors and/or emulators in check and that each and every of these instructions is intended to in the future be available to non-Windows users.
Although "adding specific features on Microsoft's request you don't tell anybody else about until much later" is still introducing secret instructions specifically for Microsoft, but at least it doesn't seem to be intentional.

Quoting: LoudTechieI was referring to the undocumented instructions found in the Windows kernel.
Thanks for the assurance that those instructions don't seem to be kept secret with the goal to keep back potential Windows competitors and/or emulators in check and that each and every of these instructions is intended to in the future be available to non-Windows users.
Although "adding specific features on Microsoft's request you don't tell anybody else about until much later" is still introducing secret instructions specifically for Microsoft, but at least it doesn't seem to be intentional.

Intel has agreements with other partners as well. I know it seems like Intel and MSFT could collude together but that would be unfair business practices and Intel would be looking at a huge lawsuit. All upcoming feature availability get disclosed so all Intel partners get equal access.

Quoting: Highball

Quoting: LoudTechieI was referring to the undocumented instructions found in the Windows kernel.
Thanks for the assurance that those instructions don't seem to be kept secret with the goal to keep back potential Windows competitors and/or emulators in check and that each and every of these instructions is intended to in the future be available to non-Windows users.
Although "adding specific features on Microsoft's request you don't tell anybody else about until much later" is still introducing secret instructions specifically for Microsoft, but at least it doesn't seem to be intentional.

Intel has agreements with other partners as well. I know it seems like Intel and MSFT could collude together but that would be unfair business practices and Intel would be looking at a huge lawsuit. All upcoming feature availability get disclosed so all Intel partners get equal access.

Ah, thnx.
That really eases some of my fears.
Microsoft is the only one I knew had access to this kind of stuff(this is obviously under NDA's and such, so I get my info from reverse engineering efforts and leaks).