Analysis speed

The speed of analysis of a codebase is influenced by many factors, of which the most significant are probably the structure of the codebase being analysed and the performance of the system the analysis is being run on. For a large codebase it is not unknown for a complete analysis run to take several hours – and needless to say, this can potentially pose problems or even act as a disincentive to run the analysis at all.

Analysis tools such as PC-lint and CppCheck which operate directly on source files (unlike the .NET assembly analysis tool FxCop, for example) must read and process source files - and therefore the structure of your project can be very significant in terms of analysis time.

In a C or C++ project paying close attention to the include dependencies will very likely reduce the analysis time. Similarly, a reasonably specified modern system (e.g. an i5 or i7 processor with fast disks and plenty of memory) will crunch through your codebase in a far shorter time than an ageing Pentium 4 machine.

As most analysis tools are entirely single threaded, adding more CPU cores will not reduce the analysis time unless you run multiple analysis tasks simultaneously - which is of course the approach Visual Lint primarily takes. Depending on the structure of the codebase and the performance of the machine this can in some cases reduce analysis times substantially.

Analysis runs can be accelerated still further by utilising an IncrediBuild grid to augment the power of the system upon which Visual Lint is installed. Please see Accelerating analysis using IncrediBuild for details.



Copyright © 2004 - 2015 by Riverblade Ltd. All rights reserved.