Slow trunk

[Boudewijn Rempt]

On Saturday 06 February 2010, Boudewijn Rempt wrote:
> On Sat, 6 Feb 2010, Dmitry Kazakov wrote:
> > Hi, All.
> > 
> > It seems some commit made the trunk very slow. The painting became slow
> > even with predefined brushes.
> > The reference revision that was fast: r1083631
> > Slow revision: r1085960
> > 
> > Something bad happened between them =)
> 
> The interesting thing is that, for the rest of us, painting became a lot
> faster.

To be exact:

trunk before walkers

RESULT : KisStrokeBenchmark::benchmarkStroke():
     226.9 msec per iteration (total: 2269, iterations: 10)
RESULT : KisStrokeBenchmark::benchmarkRandomLines():
     1,460.0 msec per iteration (total: 14600, iterations: 10)
 
 
r1083597 (about the closest I could find for your base revision)
 
RESULT : KisStrokeBenchmark::benchmarkStroke():
        1,930.7 msec per iteration (total: 19307, iterations: 10)
RESULT : KisStrokeBenchmark::benchmarkRandomLines():
     13,380.2 msec per iteration (total: 133802, iterations: 10)

After committing the walkers patch:

RESULT : KisStrokeBenchmark::benchmarkStroke():
     246.0 msec per iteration (total: 2460, iterations: 10)
PASS   : KisStrokeBenchmark::benchmarkStroke()
RESULT : KisStrokeBenchmark::benchmarkRandomLines():
     1,541.0 msec per iteration (total: 15411, iterations: 10)

(Though that's probably normal variation)
-- 
Boudewijn Rempt | http://www.valdyas.org