Usage of :¶

[-A] <accelerator> use either GENWQE for the PCIe and CAPI for

CAPI based solution available only on System p Use SW to use software compress/decompression

[-C] <card> set the compression card to use (0, 1, ... ).

RED (or -1) drive work to all available cards.

[-P] Use polling to detect work-request completion/only CAPI. [-t] <test_data.bin> [-l] Enable system load logging

sadc - System activity data collector and gnuplot must be installed

[-M] <threads> Maximum number of threads to be used. [-v] Print status and informational output. [-V] Print program version (https://github.com/ibm-genwqe/genwqe-user) [-h] Print this help message.

Input data is to be placed in /tmp/test_data.bin. If it does not exist, the script will generate random example data. Using random data will cause performance to suffer, since it will not compress nicely. So using something more realistic is certainly a good idea.

Note that the path needs to be setup to find the zlib_mt_perf tool.

E.g. run as follows:¶

Use GenWQE accelerator card 0:

PATH=tools:$PATH tools/zlib_mt_perf.sh -A GENWQE -C0

Use CAPI accelerator card 0:

PATH=tools:$PATH tools/zlib_mt_perf.sh -A CAPI -C0

Use software zlib:

PATH=tools:$PATH tools/zlib_mt_perf.sh -A SW