Re: Performance with new nested-xacts code - Mailing list pgsql-hackers

Hi Tom,

As requested - although the results are all over the place... :-(
One interesting factor in these tests is that the max tps without
the new code was 74.7, with the new code, 85.8.

This is a Sony Laptop, slow IDE disk, Fedora Core 2

[grant@localhost pgsql-HEAD]$ uname -a
Linux localhost.localdomain 2.6.6-1.435 #1 Mon Jun 14 09:09:07 EDT 2004 i686 
i686 i386 GNU/Linux

./bin/postmaster -F

HTH.

Regards,
Grant

-- PRE NESTED XACTS
[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 74.632059 (including connections establishing)
tps = 74.710309 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 61.405658 (including connections establishing)
tps = 61.471754 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 59.702545 (including connections establishing)
tps = 59.754499 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 54.531685 (including connections establishing)
tps = 54.584432 (excluding connections establishing)

-- POST NESTED XACTS
[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 72.656915 (including connections establishing)
tps = 72.732723 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 85.687383 (including connections establishing)
tps = 85.822281 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 59.479127 (including connections establishing)
tps = 59.540478 (excluding connections establishing)

[grant@localhost pgbench]$ ./pgbench -c 5 -t 1000 bench
starting vacuum...end.
transaction type: TPC-B (sort of)
scaling factor: 10
number of clients: 5
number of transactions per client: 1000
number of transactions actually processed: 5000/5000
tps = 51.675145 (including connections establishing)
tps = 51.715526 (excluding connections establishing)

Tom Lane wrote:
[snip]
> 
> Can anyone else reproduce these results?  The test case I'm using is
>     pgbench -i -s 10 bench
> followed by repeated
>     pgbench -c 5 -t 1000 bench
> I've built PG with --enable-debug and --enable-cassert, and am running
> with -F (fsync off) but otherwise absolutely factory-stock
> postgresql.conf.  The hardware is a not-so-new-anymore Dell P4 with
> run-of-the-mill IDE disk drive, running RHL 8.0.  Obviously none of this
> is tuned at all, but the question is why did CVS tip get faster when it
> should by rights be slower.
>