Hacker News Reader

Comment by albeva | Hacker News Reader

Competition yes, but I'd prefer from AMD rather than a different CPU architecture.

Putting out two fires simultaneously means fewer resources are spent on each individual one.

I could even fantasize about a scenario where Intel: "if you launch an ARM laptop we'll raise our prices" Microsoft: "fine, we will switch to AMD"

Isn't AMD emulating x86 over RISC anyway?

as is Intel, I don't think there has been a true x86 CPU in a while

I did hear this from my Prof in our university too, where can I find information about this?

https://en.wikipedia.org/wiki/Microcode

I think you were thinking of microarchitecture, microcode is related but probably not what parent meant.

Information about Intel internal architectures is highly confidential, but there are remotely related technologies that are generic and public, for example https://en.m.wikipedia.org/wiki/Register_renaming

Nah, Tomasulo and all that jazz (microarchitecture) is about improving performance once you've got a defined instruction set.

Microcode (which is one of those highly confidential things that both Intel and AMD hold close and dear) is about doing an on-the-fly CISC to RISC transformation because you realized that the legacy x86 ISA is an absolute pain to handle (but you aren't willing to give it up).

I believe you are confusing microcode (which is mainly used for complex and slow instructions and does not need to be RISC) with micro-ops which is the Intel lingo for the internal RISC operations.

Register renaming is a key part of high performance ISA emulation. The x64 has 16 gp registers but the internal RISC normally has 80. Of course this is also a key part of OOO and alike as you mentioned.

Reply on news.ycombinator.com