For overall tests (that include network, driver and services) I prefer benchmarks from Windows with nic settings like Jumboframes and interrupt throttling disabled, either tests like a Crystal Disk benchmark over an iSCSI target with sync always/disabled or the AJA testsuite for video editing with 4k RGB over SMB2 see http://napp-it.org/doc/downloads/performance_smb2.pdf
While 1G performance was quite similar, I saw the biggest difference over 10/40G
For my last local tests I use a filebench test suite script (Menu Pools > Benchmark) see http://napp-it.org/doc/downloads/optane_slog_pool_performane.pdf
I may add a Solaris test to the latter to check differences that are more related to Oracle ZFS v37 vs the newest OpenZFS.
While 1G performance was quite similar, I saw the biggest difference over 10/40G
For my last local tests I use a filebench test suite script (Menu Pools > Benchmark) see http://napp-it.org/doc/downloads/optane_slog_pool_performane.pdf
I may add a Solaris test to the latter to check differences that are more related to Oracle ZFS v37 vs the newest OpenZFS.