12-second simulation of my neural network on my HP laptop with 1.7GHz processor had taken ~28min to execute. Thinking that a “faster” computer would perform better, I have tried buying another HP laptop with purported 5GHz processor. However, improvement was slight, still requiring 24min. Is there a setting on my VSC platform to speed task execution?
These things are unfortunately not straightforward any more, since processor cores no longer have a single processor speed. The 1.7GHz was probably the “base speed”, which can go up to something like 5GHz if needed, so the actual clock speed difference might not have been as high as you thought. Also, the raw speed of the processor is not the only thing that goes into the performance, things like cache size, memory speed etc. also play a role. These days, speed gains are mostly not coming from higher clock speeds, but from more processors that can do more work in parallel (with GPUs taking things to the extreme).
In your simulations, are you using C++ standalone mode? If yes, and if the network is not too small, it will probably benefit from using multi-threading to spread the execution over multiple processor cores : Computational methods and efficiency — Brian 2 2.10.1 documentation
Alternatively, if you are doing things like parameter exploration, or have to run multiple runs of the same stochastic network, you can also run full simulations in parallel across multiple processes: Computational methods and efficiency — Brian 2 2.10.1 documentation
There might also be ways to reformulate the model itself to get better performance. Using the profiling mode might give you pointers to where the time is spent and where it might be possible to optimize the model: Running a simulation — Brian 2 2.10.1 documentation