jyapayne/swig
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Adjust the DOH string hash function

Browse source 
The one we're currently using only considers the last five characters
plus the least significant bit of the last-but-sixth character, which
unsurprisingly generates a lot of many-way collisions.

This change seems to give about a 4% reduction in wallclock time for
processing li_std_list_wrap.i from the testsuite for Python.  The
hash collision rate for this example drops from 39% to 0!

Closes #2303
This commit is contained in:
Olly Betts 2022-07-07 11:50:00 +12:00
commit 5a96a39aa4
1 changed files with 14 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

20
Source/DOH/string.c
View file

@ -180,19 +180,27 @@ static int String_hash(DOH *so) {
if (s->hashkey >= 0) { if (s->hashkey >= 0) {
return s->hashkey; return s->hashkey;
} else { } else {
char *c = s->str; /* We use the djb2 hash function: https://theartincode.stanis.me/008-djb2/
*
* One difference is we use initial seed 0. It seems the usual seed value
* is intended to help spread out hash values, which is beneficial if
* linear probing is used but DOH Hash uses a chain of buckets instead, and
* grouped hash values are probably more cache friendly. In tests using
* 0 seems slightly faster anyway.
*/
const char *c = s->str;
unsigned int len = s->len > 50 ? 50 : s->len; unsigned int len = s->len > 50 ? 50 : s->len;
unsigned int h = 0; unsigned int h = 0;
unsigned int mlen = len >> 2; unsigned int mlen = len >> 2;
unsigned int i = mlen; unsigned int i = mlen;
for (; i; --i) { for (; i; --i) {
h = (h << 5) + *(c++); h = h + (h << 5) + *(c++);
h = (h << 5) + *(c++); h = h + (h << 5) + *(c++);
h = (h << 5) + *(c++); h = h + (h << 5) + *(c++);
h = (h << 5) + *(c++); h = h + (h << 5) + *(c++);
} }
for (i = len - (mlen << 2); i; --i) { for (i = len - (mlen << 2); i; --i) {
h = (h << 5) + *(c++); h = h + (h << 5) + *(c++);
} }
h &= 0x7fffffff; h &= 0x7fffffff;
s->hashkey = (int)h; s->hashkey = (int)h;