Sujet : Re: Intel
De : blockedofcourse (at) *nospam* foo.invalid (Don Y)
Groupes : sci.electronics.designDate : 05. Aug 2024, 03:56:24
Autres entêtes
Organisation : A noiseless patient Spider
Message-ID : <v8pf0p$fjq7$1@dont-email.me>
References : 1 2 3 4 5 6
User-Agent : Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:102.0) Gecko/20100101 Thunderbird/102.2.2
On 8/4/2024 2:47 PM, Dimiter_Popoff wrote:
In the last 20 years, Intel processors had nothing to do with the x86
architecture, except that they could accept x86 code after reset.
Inside, they are a bunch of RISCs, and there is no EAX register but
some 100s of renaming registers that all could take the role of EAX,
in case of speculative execution even some of them at the same time.
So do they have any instruction set available to the programmer other
than the x86 one or is all this just an emulation of x86 code.
That would be considered "trade secret" as revealing anything of their
internal implementation would give others a leg up in that market.
Is there a programming model to be seen in some public document
describing all those many registers. Last time I (vaguely) checked
all I could see were the x86 registers, just made longer - which
brings with itself all the limitations the x86 architecture has
always had, just going over them faster by faster silicon.
There is no value to their customer base to be able to modify the
microcode that emulates the x86 architecture. Exposing that
programming model would then require *supporting* it (toolchain).
And, force them to preserve it, in addition to the x86 interface.
I suspect there are some details of the internals that might be
made available (under NDA) to third-party tool vendors (debuggers,
ICE, etc.) but that would likely be a very privileged relationship
(and, Intel would likely have contract language that effectively caused
any violation of the NDA to result in forfeiture of the offending
company, just based on size/assets!)