Thanks Zach! I've pulled your changes to 0.3.0-SNAPSHOT. Updated benchmark:
Case: :clojure.zip
Evaluation count : 75480 in 60 samples of 1258 calls.
Execution time mean : 805.666773 µs
Execution time std-deviation : 4.815877 µs
Execution time lower quantile : 797.942766 µs ( 2.5%)
Execution time upper quantile : 816.578299 µs (97.5%)
Found 2 outliers in 60 samples (3.3333 %)
low-severe 2 (3.3333 %)
Variance from outliers : 1.6389 % Variance is slightly inflated by outliers
Case: :fast-zip
Evaluation count : 297900 in 60 samples of 4965 calls.
Execution time mean : 202.892179 µs
Execution time std-deviation : 848.456881 ns
Execution time lower quantile : 201.212286 µs ( 2.5%)
Execution time upper quantile : 204.187311 µs (97.5%)
Found 1 outliers in 60 samples (1.6667 %)
low-severe 1 (1.6667 %)
Variance from outliers : 1.6389 % Variance is slightly inflated by outliers
I did try adding some inline forms, but they didn't seem to help. See this
branch:
https://github.com/akhudek/fast-zip/tree/no-protocol+inline
On Sunday, June 30, 2013 7:46:53 PM UTC-4, Zach Tellman wrote:
>
> This is really cool, thanks for doing this. I was able to eke out another
> 1.8x speedup by replacing '=' with 'identical?' for the keyword comparisons
> [1]. There also might be further room for improvement by defining inline
> forms for some of the smaller functions.
>
> Zach
>
> [1]
> https://github.com/ztellman/fast-zip/commit/ee7a64630389f36a539771658586a093369f7939
>
> On Sunday, June 30, 2013 1:18:05 PM UTC-7, Alexander Hudek wrote:
>>
>> I've updated the clojure.zip implementation to use records internally.
>> This
>> achieves a speedup of roughly 2x. You can find the library below and on
>> clojars:
>>
>> https://github.com/akhudek/fast-zip
>>
>> It's a drop in replacement for clojure.zip in terms of interface and
>> usage.
>> However, since the internal representation has changed, fast-zip
>> locations
>> cannot be used with the clojure.zip implementation. E.g. you cannot
>> create a zipper with fast-zip, then use clojure.zip/next on the result.
>> You
>> must use fast.zip.core/next instead.
>>
>> In addition to the micro-benchmark provided, I've tried this in a larger
>> project
>> that makes extensive use of zippers and found a similar speedup.
>>
>> Comments and suggestions welcome.
>>
>> Alex
>>
>
--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to [email protected]
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
