Hello Robert,
>> Good point. One simple idea here would be to use a custom pgbench
>> script that has no SQL commands and just calculates the values of some
>> parameters to measure the impact without depending on the backend,
>> with a fixed number of transactions.
>
> Sure, we could do that. But whether it materially changes pgbench -S
> results, say, is a lot more important.
Indeed. Several runs on my laptop:
~ 400000-540000 tps with master using: \set naccounts 100000 * :scale \setrandom aid 1 :naccounts
~ 430000-530000 tps with full function patch using: \set naccounts 100000 * :scale \setrandom aid 1 :naccounts
~ 730000-890000 tps with full function patch using: \set aid random(1, 100000 * :scale)
The performance is pretty similar on the same script. The real pain is
variable management, avoiding some is a win.
However, as you suggest, the tps impact even with -M prepared -S is
nought, because the internal scripting time in pgbench is much smaller
than the time to do actual connecting and querying.
--
Fabien.