Re: performance of first exec of prepared statement

> On Apr 17, 2020, at 7:34 AM, Ted Toth <txtoth(at)gmail(dot)com> wrote:
>
> 
>
>
>> On Fri, Apr 17, 2020 at 8:28 AM Ted Toth <txtoth(at)gmail(dot)com> wrote:
>>
>>
>>> On Thu, Apr 16, 2020 at 8:09 PM Rob Sargent <robjsargent(at)gmail(dot)com> wrote:
>>> On 4/16/20 6:15 PM, Adrian Klaver wrote:
>>> > On 4/16/20 4:59 PM, Ted Toth wrote:
>>> >>
>>> >>
>>> >> On Thu, Apr 16, 2020 at 6:29 PM Ted Toth <txtoth(at)gmail(dot)com
>>> >> <mailto:txtoth(at)gmail(dot)com>> wrote:
>>> >>
>>> >> I've noticed that the first exec of an INSERT prepared statement
>>> >> takes ~5 time longer (I'm using libpq in C and wrapping the calls to
>>> >> time them) then subsequent exec's is this the expected behavior and
>>> >> if so is there any thing I can do to mitigate this affect?
>>> >>
>>> >> Ted
>>> >>
>>> >>
>>> >> For example (in my environment) I'm seeing the prepare take ~10ms,
>>> >> the first exec take ~30 ms and subsequent exec's take ~4 ms.
>>> >>
>>> >
>>> > I don't have an answer. I believe though that to help those that might
>>> > it would be helpful to show the actual code.
>>> >
>>> >
>>> You expect the subsequent calls to benefit from the cached query parse
>>> and planning. What does you query cost without begin wrapped in a
>>> prepared statement (preferably from a cold start).
>>>
>>>
>>>
>> I thought that's what the PQprepare call was supposed to do i.e. parsing/planning.
>>
>> It's a bit difficult to get an unprepared query cost since there are a lot of columns :(
>> #define INSERT_SQL "INSERT INTO t (<column names>) VALUES ($1,$2,$3,$4,$5,$6,$7,$8,$9,$10,$11,$12,$13,$14,$15,$16,$17,$18,$19,$20,$21,$22,$23,$24,$25,$26,$27,$28,$29,$30,$31,$32,$33,$34,$35,$36,$37,$38,$39,$40,$41,$42,$43,$44,$45,$46,$47,$48,$49,$50,$51,$52,$53,$54,$55,$56,$57,$58,$59,$60,$61,$62,$63,$64,$65,$66,$67,$68,$69,$70,$71,$72,$73,$74,$75,$76,$77,$78,$79,$80,$81,$82,$83,$84,$85,$86,$87);"
>>
> Ah prepare does the parsing and execute does the planning.
>
> Another related question is that my implementation uses strings for all values is there a performance benefit to using actual values and specifying their datatype?

The cost in this example is 87-N toSting() calls on client plus 87-N valueOf(Sting) calls on server where N is the number of textual columns
>