Benedikt, could you please elaborate on how can one print those assembler listings for different browsers?
P. S. By the way, I think that this way sometimes can be more favorable over an obscure micro-benchmarking, because the latter does not give you a clue why the particular implementation outperforms as opposed to the generated code examination. Also, it is going to be easier to monitor the performance regressions in performance-centric microlibraries if someone makes the assembly code comparisons to run in automatic mode between runtime engine upgrades. Just thinking aloud.
Thanks for the good food for thought.