Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
>> Joshua D. Drake wrote:
>>> I would also question the 64KB at a time. Why not a 1024KB (arbitrary)
>>> at a time? Is it a resource issue? In the old days when we actually
>>> had people trying to run postgresql on 128 and 256 megs of ram, o.k.
>>> but now?
>
>> It would be simple enough to change. Try it and see if it actually makes
>> a difference. All you have to change is the define of RAW_BUF_SIZE.
>
> Seems unlikely that making it bigger than (a fraction of) L2 cache
> would be a smart move.
O.k. these CPUs have 1meg of L2 so I will try with 512k.
Joshua D. Drake
>
> regards, tom lane
>
