CSV file output show wrong number of iterations

Aleksey Shipilev aleksey.shipilev at oracle.com
Wed Jan 22 14:24:00 PST 2014


On 01/23/2014 01:20 AM, Staffan Friberg wrote:
> Looks like the CSV file output only counts the iteration from a single
> fork run, and does not add them up as done in the stdout output.

Good catch, thanks!

In current version, human-readable log has dropped "Count" to mean
"Iterations", and instead means "Samples", which more closely resembles
to how statistics is performed. In most modes number of samples equates
to number of iterations. Some modes, notably SampleTime, sample more
frequently, and you can have millions of samples over the run.

Hence, I changed CSV output to mean "Samples" as well, and to look
exactly like stdout does, here:
 http://hg.openjdk.java.net/code-tools/jmh/rev/d664e79b3c0b

-Aleksey.



More information about the jmh-dev mailing list