Mirror
/
cpython
mirror of https://github.com/python/cpython
4
1
Fork 0
Code Issues Releases Wiki Activity

Reduce load factor (from 66% to 60%) to improve effectiveness of linear probing. 

Decreased density gives better collision statistics (average of 2.5 probes in a
full table versus 3.0 previously) and fewer occurences of starting a second
possibly overlapping sequence of 10 linear probes.  Makes resizes a little more
frequent but each with less work (fewer insertions and fewer collisions).
This commit is contained in:
Raymond Hettinger 2017-02-04 02:43:42 -08:00
parent b451f91786
commit 5cd87a8d61
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Objects/setobject.c
View File

@ -234,7 +234,7 @@ set_add_entry(PySetObject *so, PyObject *key, Py_hash_t hash)
so->used++;
entry->key = key;
entry->hash = hash;
if ((size_t)so->fill*3 < mask*2)
if ((size_t)so->fill*5 < mask*3)
return 0;
return set_table_resize(so, so->used);
@ -642,7 +642,7 @@ set_merge(PySetObject *so, PyObject *otherset)
* incrementally resizing as we insert new keys. Expect
* that there will be no (or few) overlapping keys.
*/
if ((so->fill + other->used)*3 >= so->mask*2) {
if ((so->fill + other->used)*5 >= so->mask*3) {
if (set_table_resize(so, so->used + other->used) != 0)
return -1;
}
@ -986,7 +986,7 @@ set_update_internal(PySetObject *so, PyObject *other)
*/
if (dictsize < 0)
return -1;
if ((so->fill + dictsize)*3 >= so->mask*2) {
if ((so->fill + dictsize)*5 >= so->mask*3) {
if (set_table_resize(so, so->used + dictsize) != 0)
return -1;
}