You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It doesn't support benchmark resource reinitialization between benchmark attempts. In many cases we need to prepare benchmark data before running the benchmark (to measure time precisely for the interesting operation).
Extra dependency to install and build (for Sanitizers build).
We have to support both benchmark/native versions of the same benchmark code. For this, we have to support conditional macro blocks.
We have 2 components that are enough to make benchmarks without libbenchmark:
bsls::Stopwatch
bsls::Stopwatch stopwatch;
stopwatch.reset();
stopwatch.start();
for (int i = 0; i < k_BENCHMARK_ITERATIONS; ++i) {
EVENT_TYPE event(&lsa);
(void)event;
}
stopwatch.stop();
There are few issues with
libbenchmark:We have 2 components that are enough to make benchmarks without
libbenchmark:bsls::Stopwatchbmqtst::Table: https://github.com/bloomberg/blazingmq/blob/main/src/groups/bmq/bmqtst/bmqtst_table.h