26 Apr
2010
26 Apr
'10
2:51 p.m.
On 04/26/2010 07:48 AM, Henning Westerholt wrote:
On Monday 26 April 2010, Alex Balashov wrote:
There is probably a certain point on which further increase of the htable size
make not that much sense anymore. It seems the module only supports from 256
to 16384 buckets in the htable. So depending on the distribution of the hash
function the htable will start to degenerate, up to a worst case complexity to
O(n). The used function is core_case_hash from hash_func.c
Indeed, I asked this question after seeing your git commit to the
documentation that specifies that maximum dimension is 2^14.
Is there a particular reason for the cap to be at 2^14?
--
Alex Balashov - Principal
Evariste Systems LLC
1170 Peachtree Street
12th Floor, Suite 1200
Atlanta, GA 30309
Tel: +1-678-954-0670
Fax: +1-404-961-1892
Web: http://www.evaristesys.com/
i think
this is a static setting, as the memory is allocated on startup
in shared memory. Do you've run into performance problems or memory size
constraints with the htable module?
None whatsoever, I was just curious from a theoretical perspective.
The answer lets me know in an a priori way whether the htable is optimal
for applications involving rapidly shrinking or contracting data sets,
by several factors.