Re: visibility maps

On Thu, Dec 11, 2008 at 8:09 PM, Heikki Linnakangas
<heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
> Pavan Deolasee wrote:
>
>
>> I can do some if we don't have already.
>
> Oh, yes please!
>

I did some tests today with pgbench on a decent SMP machine. The goal
was to see if multiple clients (20 in the test) tries to update tuples
in different data blocks, if the EX lock on the VM page causes any
contention.

I can confirm that I haven't seen any drop in the tps with VM. I guess
since the bit reset is a very small code compared to the entire UPDATE
code path, may be its less likely than I thought previously that
multiple clients attempt to reset the bit at the same time. I'll do
some more tests to see if setting the bit in HOT-prune path leads to
any contention or not.

I can send details of the test I did, if anyone is interested.

Thanks,
Pavan

--
Pavan Deolasee
EnterpriseDB http://www.enterprisedb.com