Benchmark setups


In a discussion about benchmarks, it was noted that /etc/malloc.conf changes can help benchmarks tremendously. Rahul Siddharthan suggested ‘/etc/malloc.conf -> H' and Jeremy Messenger suggested '/etc/malloc.conf -> aj'

Also, Matt Dillon made a number of suggestions on what to check when benchmarking DragonFly vs. FreeBSD (4 or 5)

Matt Dillon quote follows:


"Be sure to compile 5.x without DIAGNOSTIC, INVARIANTS, or WITNESS. Be sure that the kernel is being compiled -O. Doubly be sure that WITNESS support is not on. INVARIANTS and WITNESS *are* turned on in 5.x's GENERIC.

Set maxusers to 0 so it autoconfigs.

You might want to try 5.x's SCHED_ULE instead of SCHED_4BSD. I don't expect it to make a huge difference but ULE is supposed to deal with affinity better so who knows.

Make sure the filesystem partitions for DFly and 5.x, if different, are using the same block and fragment sizes. 'tunefs -p partition' ought to tell you.

Check the IDE setup using 'atacontrol info 0', (1, 2, 3, etc..) under both OS's and 'atacontrol mode 0', (1, 2, 3, etc..) under both OS's. "

Posted by     Categories: Goings-on     1 Comment
1 Comment on Benchmark setups

Closed

  1. Mezz says:

    Why am I suggesting to try do the ‘/etc/make.conf -> aj’?

    In -CURRENT, it sets the malloc -> AJ by default that enable the memory debug. I didn’t check in the manpage of malloc again, but I remember that the ‘J’ can hurt the performance. Let’s see if it will improvement the benchmark.

    BTW: In the RELEASE branch of 5.x, I think it always set the malloc -> aj by default.