[Tarantool-patches] [PATCH v1 luajit 29/41] perf: adjust scimark-fft in LuaJIT-benches

[Sergey Kaplun]

Hi, Sergey!
Thanks for the review.
Please, consider my answers below.

On 17.11.25, Sergey Bronnikov wrote:
> Hi, Sergey,
> 
> thanks for the patch! Please see my comments.
> 
> Sergey
> 
> On 10/24/25 13:50, Sergey Kaplun wrote:
> > This patch adjusts the aforementioned test to use the benchmark
> > framework introduced before. The default arguments are adjusted
> > according to the <PARAM_x86.txt> file. The arguments to the script still
> > can be provided in the command line run.
> >
> > Checks are omitted since they were not present in the original suite,
> > plus the precise result value depends on the input parameter.
> > ---
> >   perf/LuaJIT-benches/scimark-fft.lua | 19 ++++++++++++++++++-
> >   1 file changed, 18 insertions(+), 1 deletion(-)
> >
> > diff --git a/perf/LuaJIT-benches/scimark-fft.lua b/perf/LuaJIT-benches/scimark-fft.lua
> > index c05bb69a..96535774 100644
> > --- a/perf/LuaJIT-benches/scimark-fft.lua
> > +++ b/perf/LuaJIT-benches/scimark-fft.lua
> > @@ -1 +1,18 @@
> > -require("scimark_lib").FFT(1024)(tonumber(arg and arg[1]) or 50000)
> > +local bench = require("bench").new(arg)
> > +
> > +local cycles = tonumber(arg and arg[1]) or 50000
> Why 50000?

It is the default for x86 arch. I've taken the values from PARAMS_x86,
since this is the most important architecture for the Tarantool, see the
commit message.

Also, it is the same as before the patch.

> > +local benchmark
> > +benchmark = {
> > +  name = "scimark_fft",
> > +  -- XXX: The description of tests for the function is too
> > +  -- inconvenient.
> > +  skip_check = true,
> > +  payload = function()
> > +    local flops = require("scimark_lib").FFT(1024)(cycles)
> Why 1024?

It is the same as before the patch. This constant is the same as in the
original SciMark v2 benchmark.

> > +    benchmark.items = flops
> > +  end,
> > +}
> > +
> > +bench:add(benchmark)
> > +
> > +bench:run_and_report()

-- 
Best regards,
Sergey Kaplun