Is there a particular reason why r=1 is used for rest of the cores (i.e. is it known to be the optimal amount of resource dedicated to a core)?
Also, r is defined as the "resources dedicated to each processing core" but I'm curious what these "resources" really include. Can we think of a resource as some simplified ratio of fetch/decode unit, ALUs, and execution context or does it consist more than that?
kayvonf
Or transistors might be a better metric. R=1 for the "small cores" was chosen just to keep the analysis simple. One could easily conceive of a processor with multiple types of programmable core.
Is there a particular reason why r=1 is used for rest of the cores (i.e. is it known to be the optimal amount of resource dedicated to a core)?
Also, r is defined as the "resources dedicated to each processing core" but I'm curious what these "resources" really include. Can we think of a resource as some simplified ratio of fetch/decode unit, ALUs, and execution context or does it consist more than that?
Or transistors might be a better metric. R=1 for the "small cores" was chosen just to keep the analysis simple. One could easily conceive of a processor with multiple types of programmable core.