Re: spinlock contention

On Tue, Jun 28, 2011 at 5:55 PM, Florian Pflug <fgp(at)phlo(dot)org> wrote:
> On Jun28, 2011, at 23:48 , Robert Haas wrote:
>> On Tue, Jun 28, 2011 at 5:33 PM, Merlin Moncure <mmoncure(at)gmail(dot)com> wrote:
>>> On Tue, Jun 28, 2011 at 3:18 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>>>> user-32: none(1.0),atomicinc(14.4),pg_lwlock_cas(22.1),cmpxchng(41.2),pg_lwlock(588.2),spin(1264.7)
>>>
>>> I may not be following all this correctly, but doesn't this suggest a
>>> huge potential upside for the cas based patch you posted upthread when
>>> combined with your earlier patches that were bogging down on spinlock
>>> contentionl?
>>
>> Well, you'd think so, but in fact that patch makes it slower.  Don't
>> ask me why, 'cuz I dunno.  :-(
>
> Huh? Where do you see your CAS patch being slower than unpatched LWLocks
> in these results? Or are you referring to pgbench runs you made, not
> to these artificial benchmarks?

pgbench -S

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company