home | SVG

Richards Bench

These implementations follow the specification quite closely. Compare them with the OO Richards Bench variation.

The Richards (aka Bench) benchmark simulates part of a light-weight kernel of a multi-tasking operating system.
Martin Richards originally designed the benchmark to compare systems implementation languages.

Mario Wolczko provided a systematic study of different implementations in "Benchmarking Java with the Richards benchmark".

implementation seconds
XDS Oberon 2 #580.323
MS C #100.361
Sun Server Java #280.947
Sun Java #210.956
MS C#.NET #200.968
Nice #191.175
VW Smalltalk #171.883
BCPL #268.503
Psyco Python #2335.915
Python #2258.248

Measurement

Timing begins after parsing the command-line and ends before exit - see the source code. For each start value (10,000 to 1,000,000), the test is repeated, until the mean of 5 timings is within a 95% confidence interval.
(Pentium 4 2GHz, MS Windows XP.)

Acknowledgements

Martin Richards kindly granted permission to distribute versions of the benchmark. Mike Jordan gave permission to publish his C language implementation. Peter Hansen provided an initial Python language implementation.

Creative Commons License