Computer "benchmarking" is often very complicated because you need to understand how an operating system schedules tasks, how memory swapping or paging work, how multiple CPU cores and caching work, etc., etc., etc.
You have a good start --- your "standard script." If this script is characteristic of the typical workload (either current or projected), then its performance gives a good clue to actual workload results.
IMPORTANT NOTES;
- If the bus speed for memory is limited, then a faster CPU is slowed down to memory speed.
- If the entire program fits in the current memory (that is, no paging or swapping), then adding memory does not improve speed. - - --see: 132 132 133
- A slow CPU with lots of memory can perform a task faster that a fast CPU with insufficient memory. This is because input/output often takes thousands of times longer than memory access.
- . . .
DO NOT EXPECT A LINEAR CORRELATION.
Instead, use a standard performance measure like Microsoft's Windows Experience Index in its Windows operating systems. According to Microsoft, "the Windows Experience Index measures the capability of your computer's hardware and software configuration and expresses this measurement as a number called a base score".
