Re: benchmark results comparing versions 15.2 and 16

On Fri, May 5, 2023 at 10:01 PM MARK CALLAGHAN <mdcallag@gmail.com> wrote:

> I have two more runs of the benchmark in progress so we will have 3
> results for each of the test cases to confirm that the small regressions
> are repeatable.
>

They get similar results. Then I tried Linux perf but the hierarchical call
stacks, to be used for Flamegraph, have too many "[unknown]" entries.
I was using: ./configure --prefix=$pfx --enable-debug CFLAGS="-O3
-march=native -mtune=native -flto" LDFLAGS="-flto" > o.cf.$x 2> e.cf.$x
Adding -no-omit-frame-pointer fixes the problem, so I am repeating the
benchmark with that to confirm there are still regressions and then I will
get flamegraphs.

-- 
Mark Callaghan
mdcallag@gmail.com