@UtterDisbelief,
@tom111 is correct in that Intel CPU's are a hybrid of CISC and RISC architectures.
Intel switched to this hybrid approach way back in the mid 90's with the P6 generation of Pentium Pro CPU's. At that time I was working for a company that received some of the first production runs of the Pentium Pro's in the UK for use in industrial control systems and we where given a presentation by Intel UK that explained that this hybrid approach was the future of their CPU's.
Since then all Intel CPU'S use a CISC instruction set externally but internally they use multiple parallel decoders to simultaneously decode the CISC instructions to RISC micro-ops which are then executed on RISC based cores.
However the decoding of CISC to RISC inside the CPU is very complex, its basically a task specific, dedicated computer inside the CPU whose micro-code can be upgraded via ME Updates. This of course adds quite a bit of overhead to the instruction pipe-line so Intel introduced features such as speculative and out-of-order execution of certain instructions to help limit the overhead .... and we all know hoe that worked out for them
.
Because the x86 instructions are basically defined in software (ME firmware) Intel is able to quickly and easily introduce new CPU features (and fix any issues) without having to physically change the CPU design, this is the key advantage of the hybrid CISC - RISC design and why we see more and more CPU features being added with each generation of Intel CPU.
CPU's are generally classified by their instruction set and Intel X86 is a whopper, it is probably the most complex mass produced CPU instruction set on the planet as such they are classified as CISC CPU's despite them having RISC based architecture at their very core (pun intended
).
Cheers
Jay