Originally Posted by Quantium40;15353597
I never realized just how much of a good thing that AMD acquiring Nexgen was back in the day. Before Nexgen, AMD did not really make any revolutionary CPU designs as compared to their k6, athlon, a64. Without the CMD team from Nexgen, none of these CPU's would have ever come about, and these are really the best of AMD's accomplishments.
It sounds like AMD is going to need new brilliance from somewhere if they are to survive in the CPU market.
The design wasn't some complete revolution. It was better, but it was similar in many ways to the Pentium 3. The biggest revolution of the Athlon days was the Pentium 4. P4 tried many new things. The P4 was a failure as a consumer or server mainstream processor; however, from the ashes of the P4, new Intel chips have risen. With each step past the original core architecture, Intel has added features that were present in the P4 (in a manner of speaking, each generation brings us closer to P4). Intel has overcome the problems of the P4 and has implemented its strengths.
The next revolution in computing is the switch to CMT. All the big chip designers have talked about switching (it has been theoretical since the 70's and has had more serious discussions this century), but only AMD had the stones to give it a try. The initial Bulldozer design was scrapped (45nm) as too early. The second iteration (or perhaps a die shrink of the first) has quite a number of design problems (nothing that invalidates the CMT theory, just good ol' fashion engineering screw-ups).
The larger problem here is that unlike the P4, there is no way to go back (like Intel went back to the P3 design) and gradually add features. The CMT design is all or nothing. If AMD is able to fix its design problems, It is very likely that the future will see Oracle (SPARC), IBM (POWER), Acorn (ARM), MIPS, Intel, Via, etc all switching to CMT design while taking lessons from AMD's mistakes.