Because AMD was focused on CMT for a while and are finally moving back to SMT which is a solid move (if I am incorrect, correct me
).
Plus the manufacturer that builds AMDs chips (can't remember their name) are more technically limited on the nano scale; so that's why Intel get big chip dies. Their manufacturing process allows them to fit a lot of clever stuff on their compact chips.
AMD doesn't have the resources, nor manufacturing process to build a chip. Sure they may be able to
design one, but unless they can physically build it then it is out of the picture. Basically its
physically difficult to design something that will compete.
For a company that is actually quite small compared to Intel I don't think they are doing too bad
I prefer AMD as a company for definite, but their compute side of things are still with Nehalem (that was released how many years ago, 2010 or something?). So they are kind of lagging behind in pure raw compute performance.
However, they are trying to get us foks and developers to buy into the idea of "HSA", which to be honest, is something we need. Offload that CPU work to the GPU.
Its becoming seriously difficult to design better performing CPUs, were nearing 14nm, soon we will hit that brick wall in terms of pure physics and dynamics.
So in a way its a good move for AMD, don't try and follow Intel and find the ever increasing difficulty, let things pan out for a while and get support for other areas.
HSA is promising