Use countTrailingZeros for indexOfTheOnlyBit

[alexbiehl]

This is an experiment for indexOfTheOnlyBit. I haven't benchmarked yet. But will do so tonight.

[alexbiehl]

I forgot about this one! Here you can see the benchmarks https://gist.github.com/alexbiehl/ed33c0cc7f009fc647606e9b88bf9e9c/revisions (left is original, right is with these patches)

[treeowl]

@alexbiehl, what's with the regressions in list conversion functions?

[treeowl]

I'm sorry I lost track of this. The benchmark numbers you posted look very good, but unfortunately I haven't been able to reproduce your results. Part of the problem may be the general flakiness of the incredibly short benchmarks. Can you suggest a procedure for getting good results reliably?

[alexbiehl]

Me neither! I went ahead and further "optimized" the folds by using Count{Leading, Trailing}Zeros but the effect was basically nil. I suggest we abandon this for now.

Edit: Yes I could try to make longer, more reliable benchmarks first, will do that.

[treeowl]

All right. Let me know. We could use more reliable benchmarks regardless of what happens to this PR.

[treeowl]

Do you think you could rebase this? @Fuuzetsu did some benchmarking and found that it at least doesn't seem to hurt. It's definitely less messy and cuts code size in key places.

[alexbiehl]

Yeah sure, will do this week.

…

On Mon, Jul 22, 2019, 20:40 David Feuer ***@***.***> wrote: Do you think you could rebase this? @Fuuzetsu <https://github.com/Fuuzetsu> did some benchmarking and found that it at least doesn't seem to hurt. It's definitely less messy and cuts code size in key places. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#379?email_source=notifications&email_token=AAOKFCNFQF4GIBFI3NFY673QAX5KJA5CNFSM4C3F6N5KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD2QZAIQ#issuecomment-513904674>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAOKFCIUZXZZZIIFUDTMFNLQAX5KJANCNFSM4C3F6N5A> .

[alexbiehl]

@treeowl Done!

[treeowl]

Thanks!