The previous patch set doesn't accept --enable-cassert. The
attached additional one fixes it. It theoretically won't give
degradation but I'll measure the performance change.
At Thu, 21 Jul 2016 18:50:07 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp> wrote in
<20160721.185007.268388411.horiguchi.kyotaro@lab.ntt.co.jp>
> Hello,
>
> At Tue, 12 Jul 2016 11:42:55 +0900 (Tokyo Standard Time), Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp> wrote
in<20160712.114255.156540680.horiguchi.kyotaro@lab.ntt.co.jp>
> After some refactoring, degradation for a simple seqscan is
> reduced to 1.4% and that of "Append(SeqScan())" is reduced to
> 1.7%. The gains are the same to the previous measurement. Scale
> has been changed from previous measurement in some test cases.
>
> t0- (SeqScan()) (2 parallel)
> pl- (Append(4 * SeqScan()))
> pf0 (Append(4 * ForeignScan())) all ForeignScans are on the same connection.
> pf1 (Append(4 * ForeignScan())) all ForeignScans have their own connections.
>
>
> patched-O2 time(ms) stddev(ms) gain from unpatched (%)
> t0 4121.27 1.1 -1.44
> pl 1757.41 0.94 -1.73
> pf0 6458.99 192.4 20.26
> pf1 1747.4 24.81 78.39
>
> unpatched-O2
> t0 4062.6 1.95
> pl 1727.45 9.41
> pf0 8100.47 24.51
> pf1 8086.52 33.53
>
> > > Addition to the aboves, I will try reentrant ExecAsyncWaitForNode
> > > or something.
>
> After some consideration, I found that ExecAsyncWaitForNode
> cannot be reentrant because it means that the control goes into
> async-unaware nodes while having not-ready nodes, that is
> inconsistent state. To inhibit such reentering, I allocated node
> identifiers in depth-first order so that ascendant-descendant
> relationship can be checked (nested-set model) in simple way and
> call ExecAsyncConfigureWait only for the descendant nodes of the
> parameter planstate.
--
Kyotaro Horiguchi
NTT Open Source Software Center