Greg Smith <greg(at)2ndquadrant(dot)com> writes:
> Rob Wultsch wrote:
>> At a minimum I assume that if both of the commands were started at
>> about the same time they would each scan the table in the same
>> direction and whichever creation was slower would benefit from most of
>> the table data it needed being prepopulated in shared buffers. Is this
>> the case?
> This might be optimistic;
No, it's not optimistic in the least, at least not since we implemented
synchronized seqscans (in 8.3 or thereabouts).
regards, tom lane