|
Authored by: Anonymous on Wednesday, May 09 2012 @ 10:57 PM EDT |
To be pedantic, even the machine code is interpreted -- it's just data for the
CPU's execution cycle. On a CISC architecture like x86, that goes double, as
the user-visible instructions either invoke lower-level subroutines
(microcoding, which was used in 8086 through 80386) or are translated on-the-fly
into sequences of lower-level instructions that then run on a RISC-like
execution unit (micro-ops, used by Intel at least since the Pentium, and maybe
on the 486 as well).
Note that because the user-visible x86 architecture is in effect itself a
virtual machine, Intel (and the competitors who make x86-compatible CPUs) have
been able to use multiple very different internal architectures over the years,
greatly increasing not only clock speeds but also average instructions executed
per clock cycle.
[ Reply to This | Parent | # ]
|
|
|
|
|