07b1146b15
Fix #173 : Commit feb0346 removed latency sum from reporting
...
Add latency sum back to reports.
2017-09-19 08:48:27 +03:00
feb0346fc3
Various cleanups and improvements in benchmark reports.
2017-09-18 23:14:01 +03:00
b57453df60
Report events per second in the CPU benchmark. Ref. #140 .
2017-06-07 21:49:46 +03:00
7f56e219f7
Rename --num-threads to --threads.
...
Accept the old name with a warning for compatibility.
2017-02-01 17:42:39 +03:00
788fcbb30b
Rename --max-requests to --events, --max-time to --time.
...
Print warnings if old names are used, but accept them for compatibility.
2017-02-01 17:20:22 +03:00
04b4f756d9
Refactor stats reporting to make custom reports possible.
2017-01-29 19:32:14 +03:00
49016568ba
Implement custom commands + parallel custom commands.
2017-01-28 15:45:16 +03:00
45e219d038
Change the command line syntax: remove the --test option.
2017-01-25 00:02:59 +03:00
c3fd06203b
Better formatting for help/usage text.
2016-12-30 21:23:29 +03:00
d5b661a019
Implement --histogram option.
...
When --histogram is used, print latency histogram as a part of the
benchmark report. For now, the histogram is actually a text
representation of an internal data structure used by sysbench to
calculate latency percentile.
2016-12-30 20:42:35 +03:00
f9764f1445
Fixes #54 : Reduce overhead from percentile stats
...
Rewrite percentile calculation to scale on modern hardware. Rename the
underlying data structure from sb_percentile to sb_histogram.
2016-12-05 11:18:40 +03:00
3660c3655b
Fix tests for the luajit branch.
2016-10-09 19:33:51 +02:00
ccd9845321
Adjust tests to 1.0.
2016-09-11 16:15:07 +03:00
1cfcacf84f
Basic regression tests for cpu, memory, threads and mutex benchmarks.
2016-09-11 14:37:35 +03:00
