I have a process I run regularly on a machine and am starting to spec a dedicated machine for the purpose. If it is possible to choose a processor or memory that are optimised for this, how would I go about analysing the program, perhaps debugging or disassembly, to evaluate the types of instruction it uses most, where the bottle necks are.
The background is I run it presently on a laptop that in two different energy states takes either 37 seconds in eco and 17 seconds in full power. For other processes there isn't such a significant difference in performance so I am assuming this program has some particular requirements of the system that are most impeded by running in eco mode. This makes me think I should think carefully about choosing the right hardware to run this process on in future, starting with analysing the program's behaviour. Following this then I can start searching for hardware known to be good in that characteristic.
I know for instance some engineering software performs best with particular CPUs, graphics cards, number of cores, so am presuming that this software might also have an optimal processor. It might be a particular special instruction that needs to be fast, or a basic instruction run at maximum clock speed. I'm sure the process is single threaded so adding cores won't help.
Clarification:
The application is running in Windows 10, on an Intel CPU. I do not have the source code, it is proprietary software.
