Re: spinlock contention

On Jul8, 2011, at 22:27 , Stefan Kaltenbrunner wrote:
> On 07/08/2011 04:21 PM, Tom Lane wrote:
>> Florian Pflug <fgp(at)phlo(dot)org> writes:
>>> Patch attached.
>>
>>> Beware that it needs at least GCC 4.1, otherwise it'll use a per-partition
>>> spin lock instead of "locked xadd" to increment the shared counters.
>>
>> That's already sufficient reason to reject the patch. Not everyone
>> uses gcc, let alone very recent versions of gcc.
>
> hmm - 4.1.0 was released in february 2006, which will be much older than
> even the 5 year support policy we have on pg when 9.2 will be released,
> not sure how much it will matter if we don't support as specific
> optimization on a gcc that old..

Still, it's not really hard to support older Versions, at least on
x86 and x86-64. All it takes is some inline assembly. I just don't
want to put effort into this until we know whether or not the whole
approach is worthwhile or not.

Should someone want to test this patch, but can't because of the GCC
version restriction, please speak up.

best regards,
Florian Pflug