I agree with everything you mentioned, and to add onto the bugs, it seems like MIPS are being calculated very inaccurately, and also you lose the ability to select lower clock speeds once you research higher clock speeds and lithographies.
While I was playing, I didn't notice any problems with calculating MIPS, but maybe I just didn't pay enough attention to it. Regarding the frequency, I think this is how it should be, there is no point in releasing processors with severely low frequencies, this will not give a big advantage in consumption and production cost, instead reduce the size of the crystal, so modern processors have a 3GHz+ frequency with low consumption due to small crystals.
The Intel Core i7-13620H which was released 2023 has only a base clock of 2,4 Ghz.
And to my understanding is a reason why we stopped increasing it that much, is that the thermal load would get to high. Reason why we have gone with multi cores and bigger caches.
- Performance scaling with higher clock speeds has diminished over the years. While increasing clock speeds was once a simple way to improve performance, this method has become less effective. Modern processors are more complex, and simply increasing the frequency doesn't provide linear gains in real-world performance.
- Many tasks are bottlenecked by factors other than raw clock speed, like memory access, latency, bandwidth, or parallel processing limitations.
Architecture improvements: Rather than just increasing clock speed, modern CPUs focus on improving architecture. Techniques like out-of-order execution, pipelining, cache optimizations, and branch prediction help processors achieve better efficiency and performance at lower clock speeds
etc