My speed tests compare different aspects of the languages, but focus on how close pure perl6 can get to perl5 + XS. As perl6 has no XS, that is not really a fair compare, but it aims on getting a general idea of how the perl 6 language itself optimizes performance over time.

Comparing pure perl5 (0.676) to pure perl6 (2.596) shows a factor of 3.8 *today* (2018-01-02). A comparison to perl 5 + XS shows a factor of 130, which is *much* closer than the 350 from August, and that is speed improvement in the language itself. If a process is to be written purely for speed, that might be a consideration to stick to perl 5. If your process however only parses a few thousand lines en development of the program itself is more important that an extra second in parsing the CSV (most processes parse CSV as part of a bigger picture), than reconsidering a look at perl 6 might surprise you. It is quite stable by now.

I also compare the same test to other languages on a daily basis.

One speed improvement in perl 6 that has been really noticable is in the area of parallelization. Something that my tests do not show, as one cannot parse CSV data from a single stream in parallel.