Re: optimizing a cpu-heavy query

2011/4/27 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> Joel Reymont <joelr1(at)gmail(dot)com> writes:
>> On Apr 26, 2011, at 5:00 PM, Tom Lane wrote:
>>> For another couple orders of magnitude, convert the sub-function to C code.  (I don't think you need
>>> a whole data type, just a function that does the scalar product.)
>
>> That's a 30x speedup, from 12 minutes down to 38s. Thanks Tom!
>
> Huh, I would've bet on a lot more actually.  The nodeFunctionscan and
> nodeAgg code must not be as inefficient as it looks on the surface ...

Did you mean in that case you can optimize it by collapsing those
nodes into one?

Regards,

--
Hitoshi Harada