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
3.[com.evolvedbinary.jnibench.common.bytearray](tree/main/src/main/java/com/evolvedbinary/jnibench/common/bytearray), and [com.evolvedbinary.jnibench.common.getputjni](tree/main/src/main/java/com/evolvedbinary/jnibench/common/getputjni) - Benchmarks for [JNI Data Transfer](DataBenchmarks.md).
The above command will run JMH with `ByteArrayToNativeBenchmark` benchmarks `10` times and store result in CSV files in `results` directory. You can also pass `--help` to the script to see additional JMH options that can be used.
55
+
The above command will run JMH with `ByteArrayFromNativeBenchmark` and store the result in CSV files in a subdirectory
56
+
under `results` named via timestamp. You can also pass `--help` to the script to see additional JMH options that can be
57
+
used.
58
+
59
+
The parameter
60
+
61
+
Results can then be plotted using the `jmhplot.py` script. For example, to produce results
62
+
for the `ByteArrayFromNativeBenchmark` example above, you can run:
57
63
58
-
Results can then be plotted using the `process_byte_array_benchmarks_results.py` script. For example, to produce results for the `ByteArrayToNativeBenchmark` benchmarks, you can run:
59
64
```bash
60
-
python3 process_byte_array_benchmarks_results.py -p results/ --param-name "Param: keySize" --chart-title "Performance comparison of passing byte array with {} bytes via JNI"
Command line parameter `p` expects a path to the directory with the JMH result CSV files from running the benchmarks with `jmh-benchmarks-parametrized.sh`.
64
-
The `{}` in the `chart-title` parameter will be replaced by the value from the `param-name` column.
68
+
(Replace the `-f` parameter with the appropriate directory path.)
69
+
70
+
Command line parameter `f` expects a path to the directory with the JMH result CSV files from running the benchmarks
71
+
with `jmhrun.py`.
72
+
Some benchmarks will require an additional parameter `-c` like this:
0 commit comments