Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > I have applied this patch to inline MemSet for newNode.
> > I will make another commit to make more general use of palloc0.
>
> Refresh my memory on why either of these was a good idea?
You are talking about the use of palloc0 in place of palloc/MemSet(0),
not the use of palloc0 in newNode, right?
> > This was already discussed on hackers.
>
> And this was not the approach agreed to, IIRC. What you've done has
> eliminated the possibility of optimizing away the controlling tests
> in MemSet.
I see now:
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&selm=200210120006.g9C06i502964%40candle.pha.pa.us
I thought someone had tested that there was little/no performance
difference between these two statements:
MemSet(ptr, 0, 256)
vs.
i = 256;
MemSet(ptr, 0, i)
However, seeing no email reply, I now assume no test was done.
Neil, can you run such a test and let us know. It has to be with a
compiler that optimizes out the MemSet length test when the len is a
constant. I can't remember what compiler version/settings that was, but
it may have been -O3 gcc 3.20.
I can back out my changes, but it would be easier to see if there is a
benefit before doing that. Personally, I am going to be surprised if a
single conditional tests in MemSet (which is not in the assignment loop)
will make any measurable difference.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
