The new performance focus is on throughput. Since we have many simpler cores, that possibly take long for a single instruction stream(due to the lack of ILP support), but give higher throughput since many such simple cores are present on chip.
coffee7
Adding on to what @paracon said, it is hard to make processor speed faster because of power limits, so it is hard to reduce latency. Therefore, it is good to focus on getting more things done at a given time (i.e. improve throughput). An example of this is instruction pipelining.
The new performance focus is on throughput. Since we have many simpler cores, that possibly take long for a single instruction stream(due to the lack of ILP support), but give higher throughput since many such simple cores are present on chip.
Adding on to what @paracon said, it is hard to make processor speed faster because of power limits, so it is hard to reduce latency. Therefore, it is good to focus on getting more things done at a given time (i.e. improve throughput). An example of this is instruction pipelining.