in reply to Very slow regex substitution on Unicode string

What’s going on here is that you’ve hit an old 5.8 bug from more than five years ago. If you change the first argument to timethese to 0 so that it calculates them for you, here are the results:
% perl5.8 ~/906678.pl
Is_utf8 My €0.02 of info
Is_utf8 الأنشطة
Benchmark: running ara nocmt 1step, ara nocmt 2step, ara w-cmt 1step, ara w-cmt 2step, lat nocmt 1step, lat nocmt 2step, lat w-cmt 1step, lat w-cmt 2step for at least 3 CPU seconds...
ara nocmt 1step:  4 wallclock secs ( 3.58 usr +  0.01 sys =  3.59 CPU) @  0.56/s (n=2)
            (warning: too few iterations for a reliable count)
ara nocmt 2step:  3 wallclock secs ( 3.14 usr +  0.01 sys =  3.15 CPU) @ 983039.68/s (n=3096575)
ara w-cmt 1step:  4 wallclock secs ( 3.58 usr +  0.01 sys =  3.59 CPU) @  0.56/s (n=2)
            (warning: too few iterations for a reliable count)
ara w-cmt 2step:  3 wallclock secs ( 3.11 usr +  0.01 sys =  3.12 CPU) @ 840413.46/s (n=2622090)
lat nocmt 1step:  4 wallclock secs ( 3.13 usr +  0.02 sys =  3.15 CPU) @ 1664822.54/s (n=5244191)
lat nocmt 2step:  3 wallclock secs ( 3.12 usr +  0.01 sys =  3.13 CPU) @ 923365.81/s (n=2890135)
lat w-cmt 1step:  4 wallclock secs ( 3.09 usr +  0.00 sys =  3.09 CPU) @ 1437420.39/s (n=4441629)
lat w-cmt 2step:  4 wallclock secs ( 3.11 usr +  0.01 sys =  3.12 CPU) @ 789479.17/s (n=2463175)

% perl5.10 ~/906678.pl
Is_utf8 My €0.02 of info
Is_utf8 الأنشطة
Benchmark: running ara nocmt 1step, ara nocmt 2step, ara w-cmt 1step, ara w-cmt 2step, lat nocmt 1step, lat nocmt 2step, lat w-cmt 1step, lat w-cmt 2step for at least 3 CPU seconds...
ara nocmt 1step:  3 wallclock secs ( 3.17 usr +  0.02 sys =  3.19 CPU) @ 1592576.80/s (n=5080320)
ara nocmt 2step:  4 wallclock secs ( 3.16 usr +  0.00 sys =  3.16 CPU) @ 803844.94/s (n=2540150)
ara w-cmt 1step:  3 wallclock secs ( 3.14 usr + -0.01 sys =  3.13 CPU) @ 1481335.46/s (n=4636580)
ara w-cmt 2step:  4 wallclock secs ( 3.27 usr +  0.01 sys =  3.28 CPU) @ 695634.45/s (n=2281681)
lat nocmt 1step:  3 wallclock secs ( 3.25 usr +  0.01 sys =  3.26 CPU) @ 1511156.44/s (n=4926370)
lat nocmt 2step:  4 wallclock secs ( 3.20 usr +  0.01 sys =  3.21 CPU) @ 791326.79/s (n=2540159)
lat w-cmt 1step:  3 wallclock secs ( 3.13 usr +  0.01 sys =  3.14 CPU) @ 1336099.36/s (n=4195352)
lat w-cmt 2step:  3 wallclock secs ( 3.18 usr +  0.01 sys =  3.19 CPU) @ 686363.01/s (n=2189498)

% perl5.12 ~/906678.pl
Is_utf8 My €0.02 of info
Is_utf8 الأنشطة
Benchmark: running ara nocmt 1step, ara nocmt 2step, ara w-cmt 1step, ara w-cmt 2step, lat nocmt 1step, lat nocmt 2step, lat w-cmt 1step, lat w-cmt 2step for at least 3 CPU seconds...
ara nocmt 1step:  4 wallclock secs ( 3.21 usr +  0.00 sys =  3.21 CPU) @ 1937381.31/s (n=6218994)
ara nocmt 2step:  3 wallclock secs ( 3.11 usr +  0.01 sys =  3.12 CPU) @ 766259.94/s (n=2390731)
ara w-cmt 1step:  3 wallclock secs ( 3.07 usr +  0.01 sys =  3.08 CPU) @ 1481614.61/s (n=4563373)
ara w-cmt 2step:  3 wallclock secs ( 3.12 usr +  0.01 sys =  3.13 CPU) @ 670184.03/s (n=2097676)
lat nocmt 1step:  2 wallclock secs ( 3.05 usr +  0.01 sys =  3.06 CPU) @ 1992276.47/s (n=6096366)
lat nocmt 2step:  4 wallclock secs ( 3.08 usr +  0.01 sys =  3.09 CPU) @ 750254.05/s (n=2318285)
lat w-cmt 1step:  3 wallclock secs ( 3.07 usr +  0.01 sys =  3.08 CPU) @ 1647204.22/s (n=5073389)
lat w-cmt 2step:  3 wallclock secs ( 3.08 usr +  0.01 sys =  3.09 CPU) @ 678859.55/s (n=2097676)

% perl5.14 ~/906678.pl
Is_utf8 My €0.02 of info
Is_utf8 الأنشطة
Benchmark: running ara nocmt 1step, ara nocmt 2step, ara w-cmt 1step, ara w-cmt 2step, lat nocmt 1step, lat nocmt 2step, lat w-cmt 1step, lat w-cmt 2step for at least 3 CPU seconds...
ara nocmt 1step:  4 wallclock secs ( 3.00 usr +  0.01 sys =  3.01 CPU) @ 2013724.92/s (n=6061312)
ara nocmt 2step:  3 wallclock secs ( 3.26 usr +  0.01 sys =  3.27 CPU) @ 828593.88/s (n=2709502)
ara w-cmt 1step:  4 wallclock secs ( 3.16 usr +  0.00 sys =  3.16 CPU) @ 1714874.68/s (n=5419004)
ara w-cmt 2step:  4 wallclock secs ( 3.32 usr +  0.01 sys =  3.33 CPU) @ 678051.05/s (n=2257910)
lat nocmt 1step:  2 wallclock secs ( 3.08 usr +  0.01 sys =  3.09 CPU) @ 2063201.94/s (n=6375294)
lat nocmt 2step:  4 wallclock secs ( 3.30 usr + -0.01 sys =  3.29 CPU) @ 823556.84/s (n=2709502)
lat w-cmt 1step:  3 wallclock secs ( 3.17 usr + -0.01 sys =  3.16 CPU) @ 1607695.89/s (n=5080319)
lat w-cmt 2step:  4 wallclock secs ( 3.19 usr +  0.01 sys =  3.20 CPU) @ 684215.00/s (n=2189488)

See? No more problem. Not only that, it’s getting faster, too.