Added caveat with all the increasing engineering to get the higher clocks there really isn’t significant performance improvements. Spending 2x to get 4% improvement is rather shitty. ROI unless the workloads are so valuable that the 4% time savings works out to be many many times more valuable than the cost of the silicon. However when that much money is on the line you don’t rely on a single processor you use servers.
Architectural improvements year over year can see the same or better improvements across a product line for the same clocks.
Latest Answers