When you start to investigate a performance problem you need a plan how to do it. Here I'll outline a few tips how to deal with it.
Investigation
Target
Firstly you will need to find a performance bottleneck. This shouldn't be too hard. Make sure you focus your attention on something which is used frequently since there's no point in optimizing something which is not used at all.
Current Performance
Secondly you need to determine the current performance of the system in terms of some unit. In most cases this would be time(e.g. for CPU-bound bottlenecks).
Target Improvement
After you've chosen your target and measured the current performance(let's assume it is the upper bound) we need to set ourselves a target. There are several ways of doing this:
- Take a look at KDE/Windows/etc: If they do it faster then there's no reason we have to be slower
- Ask Google: You can find quite a lot information and benchmarks on many subjects so make
WORK IN PROGRESS