Quote:
|
Got a question on the module design.
How much different is this design then the original? Not in the physical representation, but in shear multi-threaded workloads. Lets say, if it were to be explained in a step by step process would be nice. Traditional I see it as, thread -> cache -> free core; new thread -> cache -> free core; and so on. Is the new process more thread -> cache -> module -> 2 cores? Or does it act more like a watchdog? Where the main thread computes, and the second core does spare instruction sets? How exactly is information processed, step by step, in bulldozer? I'm a newb and I bet it has been discussed many times before. Cheers on input. |







