]>Fine Control of Local Whitespace in Placement : Figure 6
517919.fig.006a
(a)
517919.fig.006b
(b)
Figure 6: ICCAD 2004 IBM-MSwPins benchmark [27] ibm01 before (left) and after (right) optimal whitespace allocation via network flows. The HPWL improvement for this placement is 2.61% and takes only 10 CPU seconds.