On Fri, Sep 26, 2014 at 3:26 PM, Andres Freund <[email protected]> wrote:
> Neither, really. The hash calculation is visible in the profile, but not
> that pronounced yet. The primary thing noticeable in profiles (besides
> cache efficiency) is the comparison of the full tag after locating a
> possible match in a bucket. 20 byte memcmp's aren't free.
I'm not arguing against a more efficient hash table, but one simple
win would be to have a buffer tag specific comparison function. I mean
something like the attached patch. This way we avoid the loop counter
overhead, can check the blocknum first and possibly have better branch
prediction.
Do you have a workload where I could test if this helps alleviate the
comparison overhead?
Regards,
Ants Aasma
--
Cybertec Schönig & Schönig GmbH
Gröhrmühlgasse 26
A-2700 Wiener Neustadt
Web: http://www.postgresql-support.de
diff --git a/src/backend/storage/buffer/buf_table.c b/src/backend/storage/buffer/buf_table.c
index 7a38f2f..ba37b5f 100644
--- a/src/backend/storage/buffer/buf_table.c
+++ b/src/backend/storage/buffer/buf_table.c
@@ -34,6 +34,7 @@ typedef struct
static HTAB *SharedBufHash;
+static int BufTableCmpBufferTag(BufferTag *a, BufferTag *b, int n);
/*
* Estimate space needed for mapping hashtable
@@ -46,6 +47,19 @@ BufTableShmemSize(int size)
}
/*
+ * Compare contents of two buffer tags. We use this instead of the default
+ * memcmp to minimize comparison overhead.
+ */
+static int
+BufTableCmpBufferTag(BufferTag *a, BufferTag *b, int n)
+{
+ Assert(offsetof(BufferTag, blockNum) == 2*sizeof(uint64));
+ return (a->blockNum == b->blockNum
+ && ((uint64*)a)[0] == ((uint64*)b)[0]
+ && ((uint64*)a)[1] == ((uint64*)b)[1]) ? 0 : 1;
+}
+
+/*
* Initialize shmem hash table for mapping buffers
* size is the desired hash table size (possibly more than NBuffers)
*/
@@ -61,6 +75,7 @@ InitBufTable(int size)
info.entrysize = sizeof(BufferLookupEnt);
info.hash = tag_hash;
info.num_partitions = NUM_BUFFER_PARTITIONS;
+ info.match = BufTableCmpBufferTag;
SharedBufHash = ShmemInitHash("Shared Buffer Lookup Table",
size, size,
--
Sent via pgsql-hackers mailing list ([email protected])
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
