You should note that Parallel Computing Toolbox can only be used to run multiple simulations of the same model concurrently through "parsim" and "batchsim". So, for performance improvement over multiple simulations, PCT will indeed improve performance.
Now as to why a single simulation cannot be sped up via PCT, for the most part, a single simulation is single-threaded, so increasing the number of cores will not improve the performance of a single simulation. This is why you will only see a simulation using up to 100% of one core, so if you have 10 cores you will see 10% usage in Windows Task Manager.
Multi-threaded simulations are supported but only in extremely limited situations. Some reasons for these limitations are that multi-threading requires the presumption that the model can be broken into components that can run independently. For models that include Simscape components, there is no support for multi-threading different parts of a Simscape network. The lines or signals in a Simscape network represent physical connections, not the flow of data, which means that you cannot easily cut pieces of a network off and solve them independently.
Please refer to the following link under the "S Function Block Limitations" section to determine whether the Simscape blocks satisfy the restrictions of being able to run co-simulations or not.
Also, if the memory utilization is not maximum then increasing RAM will not improve performance.
To improve the performance of a single simulation, refer to the following article which has many helpful tips to achieve the same:
