The current cache algorithm traverses the array starting always from
the beginning, so the average number of comparisons needed to perform
a lookup is proportional to the size of the array.
By using a hash of the offset as the starting point, lookups are
faster and independent from the array size.
The hash is computed using the cluster number of the table, multiplied
by 4 to make it perform better when there are collisions.
In my tests, using a cache with 2048 entries, this reduces the average
number of comparisons per lookup from 430 to 2.5.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
BDRVQcowState *s = bs->opaque;
int i;
int ret;
+ int lookup_index;
uint64_t min_lru_counter = UINT64_MAX;
int min_lru_index = -1;
offset, read_from_disk);
/* Check if the table is already cached */
- for (i = 0; i < c->size; i++) {
+ i = lookup_index = (offset / s->cluster_size * 4) % c->size;
+ do {
const Qcow2CachedTable *t = &c->entries[i];
if (t->offset == offset) {
goto found;
min_lru_counter = t->lru_counter;
min_lru_index = i;
}
- }
+ if (++i == c->size) {
+ i = 0;
+ }
+ } while (i != lookup_index);
if (min_lru_index == -1) {
/* This can't happen in current synchronous code, but leave the check