Precise benchmarking is a complex topic. The results on these pages are intended more as an illustration of the difference in performance between platforms and configurations than an in-depth exploration of performance.
- Nightly Benchmarks - graphs from the performance example in each language, run on GitHub hosted runners as part of the nightly builds. They give an indication of consistency over time.
- Hardware Benchmarks - results for version 4.5.70 of the Device Detection C library on seven machines, from a desktop PC built in 2008 to a 128-core server, and version 4.5.34 of the IP Intelligence C library on five of them.
If performance is important to you then we highly recommend running your own profiling.
Running your own tests🔗
The performance examples for Device Detection and IP Intelligence are a simpler version of the hardware tests. In C they hold the whole data file in memory and run two threads over 10,000 records by default, using the Lite data file and the sample evidence that comes with each library. Their figures won't match the hardware benchmarks, but running them on each of your own machines shows how those machines compare with each other.
Performance Options describes the settings that affect speed and memory use. If you'd like help sizing a deployment, contact us.