Every programming language has two sorts of velocity: velocity of growth, and velocity of execution. Python has all the time favored writing quick versus operating quick. Although Python code is sort of all the time quick sufficient for the duty, typically it isn’t. In these circumstances, it is advisable to discover out the place and why it lags, and do one thing about it.
A well-respected adage of software program growth, and engineering typically, is “Measure, don’t guess.” With software program, it’s straightforward to imagine what’s incorrect, however by no means a good suggestion to take action. Statistics about precise program efficiency are all the time your greatest first software to creating purposes sooner.
The excellent news is, Python affords an entire slew of packages you should use to profile your purposes and be taught the place it’s slowest. These instruments vary from easy one-liners included with the usual library to stylish frameworks for gathering stats from operating purposes. Here I cowl 5 of essentially the most important, all of which run cross-platform and are available both in PyPI or in Python’s customary library.
Time and Timeit
Sometimes all you want is a stopwatch. If all you’re doing is profiling the time between two snippets of code that take seconds or minutes on finish to run, then a stopwatch will greater than suffice.
The Python customary library comes with two features that work as stopwatches. The Time module has the perf_counter perform, which calls on the working system’s high-resolution timer to acquire an arbitrary timestamp. Call time.perf_counter as soon as earlier than an motion, as soon as after, and procure the distinction between the 2. This provides you an unobtrusive, low-overhead—if additionally unsophisticated—approach to time code.
The Timeit module makes an attempt to carry out one thing like precise benchmarking on Python code. The timeit.timeit perform takes a code snippet, runs it many instances (the default is 1 million passes), and obtains the full time required to take action. It’s greatest used to find out how a single operation or perform name performs in a decent loop—as an example, if you wish to decide if an inventory comprehension or a standard record development shall be sooner for one thing executed many instances over. (List comprehensions often win.)
The draw back of Time is that it’s nothing greater than a stopwatch, and the draw back of Timeit is that its foremost use case is microbenchmarks on particular person strains or blocks of code. These modules solely work in the event you’re coping with code in isolation. Neither one…
