These are the microbenchmarks for the DBMS. We use Google Benchmark library for all of these.
IMPORTANT: If you add a new microbenchmark, you must add it to benchmarks.py
to have it run automatically in our nightly builds.
All of the multi-threaded benchmarks use one thread by default:
$ ./data_table_benchmark
You can set the number of threads to use with the NOISEPAGE_BENCHMARK_THREADS
environment variable:
$ NOISEPAGE_BENCHMARK_THREADS="9" ./data_table_benchmark