On Sat, 2023-12-09 at 18:52 +0700, John Naylor wrote:
> > I tested using the new hash function APIs for my search path cache,
> > and
> > there's a significant speedup for cases not benefiting from
> > a86c61c9ee.
> > It's enough that we almost don't need a86c61c9ee. So a definite +1
> > to
> > the new APIs.
>
> Do you have a new test?
Still using the same basic test here:
https://www.postgresql.org/message-id/04c8592dbd694e4114a3ed87139a7a04e4363030.camel%40j-davis.com
What I did is:
a. add your v5 patches
b. disable optimization in a86c61c9ee
c. add attached patch to use new hash APIs
I got a slowdown between (a) and (b), and then (c) closed the gap about
halfway. It started to get close to test noise at that point -- I could
get some better numbers out of it if it's helpful.
Also, what I'm doing in the attached path is using part of the key as
the seed. Is that a good idea or should the seed be zero or come from
somewhere else?
Regards,
Jeff Davis
From a30e5f0ea580fb5038eb90e862f697b557627f32 Mon Sep 17 00:00:00 2001
From: Jeff Davis <j...@j-davis.com>
Date: Fri, 8 Dec 2023 12:14:27 -0800
Subject: [PATCH v1] Use new hash APIs for search path cache
---
src/backend/catalog/namespace.c | 16 ++++++++++++----
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/src/backend/catalog/namespace.c b/src/backend/catalog/namespace.c
index 5027efc91d..af815a889d 100644
--- a/src/backend/catalog/namespace.c
+++ b/src/backend/catalog/namespace.c
@@ -42,6 +42,7 @@
#include "catalog/pg_type.h"
#include "commands/dbcommands.h"
#include "common/hashfn.h"
+#include "common/hashfn_unstable.h"
#include "funcapi.h"
#include "mb/pg_wchar.h"
#include "miscadmin.h"
@@ -247,11 +248,18 @@ static bool MatchNamedCall(HeapTuple proctup, int nargs, List *argnames,
static inline uint32
spcachekey_hash(SearchPathCacheKey key)
{
- const unsigned char *bytes = (const unsigned char *) key.searchPath;
- int blen = strlen(key.searchPath);
+ const char *buf = key.searchPath;
+ fasthash64_state hs;
- return hash_combine(hash_bytes(bytes, blen),
- hash_uint32(key.roleid));
+ fasthash64_init(&hs, key.roleid);
+
+ while (*buf)
+ {
+ fasthash64_accum_byte(&hs, *buf);
+ buf++;
+ }
+
+ return fasthash64_final32(&hs);
}
static inline bool
--
2.34.1
