Am 30.12.2013 15:06, schrieb Daniel Kozak:
This is cool!
Can you post somewhere your code and data which you use for this test? I
really like to test it on my new AA implementation :)
Here is the post that describes how to get the data:
http://forum.dlang.org/post/yglwnmawrvbhpszds...@forum.dlang.org
And here the post how to convert it to tabular text data:
http://forum.dlang.org/post/rhyycuwlxsoqbtpzp...@forum.dlang.org
After dumping the "relationship" table from mysql I just removed the
first column because it only contains the ID of the entry. I then used
the remaining two columns for profiling. You will have to get the data
yourself, as I can't repost it here (because its not my data).
Here is the post that describes my implementation:
http://forum.dlang.org/post/l9kcm2$2i4i$1...@digitalmars.com
For profiling the lookups, I generated 100_000 random numbers in the
range range (0, 200_000] and saved them to a file. I then profile the
lookup of those 100_000 random numbers. Out of all these numbers rughly
75% are valid indices and the rest are invalid indices to profile both
the lookup of valid and invalid keys.
For profiling your implementation I just used the code you posted.
Kind Regards
Benjamin Thaut
