Measuring startup times of programs is highly tricky. Startup times are greatly influenced by caching. You can test this yourself by starting a program twice in a row. The second time will be *much* faster. Though exactly how much faster itself depends on a host of factors.
The problem is that caching affects different programs differently. It typically has far greater benefits for programs written in compiled languages as compared to those written in interpreted languages.
And then there are a whole host of other variables that affect startup speed, especially for programs as complex as calibre. Things like antivirus, disk read/write speeds, CPU load, kernel scheduler strategies.
The most insidious of these is the load time difference between native and cross platform applications. Native applications (i.e. applications written using the libraries provided by the OS as opposed to a cross-platform toolkit) start much faster, because the libraries they use are already loaded into RAM.
For example, on my two year old desktop, where the GUI libraries and python are pre-loaded (as they are used by the desktop environment) calibre starts up in under a second with 450 books.
|