On Jan 10, 2008 2:12 AM, Pavel Stehule <pavel.stehule@xxxxxxxxx> wrote:
> On 10/01/2008, Matthew T. O'Connor <matthew@xxxxxxxx> wrote:
> > Pavel Stehule wrote:
> > > pgbench test - default configuration
> > >
> > > Verze 7.3.15 7.4.13 8.0.8 8.1.4 8.2.beta1 8.3beta1
> > > tps 311 340 334 398 423 585
> > >
> > > but pgbench is simple test and thise numbers hasnot great value.
> >
> > Was that the same version of pgbench each time? Or was it the pgbench
> > that came with each version? I think pgbench has changed a few times,
> > if you are using different versions of pgbench, are these numbers at all
> > meaningful?
> >
>
> I used 8.3 pgbench. I know some problems with it.
>
> pgbench is only one view (one dimension) on PostgreSQL, nothing less
> and nothing more. Some cases can be much faster or equal.
While some queries were no faster in 8.2 than in 7.4 for me, there
were many complex reporting queries that were literally thousands of
times faster. Going from minutes (nearly hours) to a second or less.
The real issue, as usual, is "How much faster is version y than
version x for YOU?" And only you can answer that by testing. In
general, I found that complex reporting queries were greatly sped up.
Updates / inserts / deletes were on the order of a bit faster to about
twice as fast.
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
