Next: 3.3.2 System Model
Up: 3.3 VLSI Performance Metric
Previous: 3.3 VLSI Performance Metric
Traditional device performance and qualification criteria
are primarily drive current, leakage, and threshold control.
However, these parameters are very indirectly linked to the actual
system performance.
Previous work on systems analysis was usually based on
analytical IV and CV equations with parameters analytically derived
from nominal process parameters [41,42].
However, such analytical models are ever less adequate to capture the
non-ideal behavior of advanced deep-sub-micron devices.
On the other hand, complete device characterization and circuit
simulation to determine the system performance is often not affordable
and, as in the case of deep-sub-micron devices, not a straight-forward
task.
Therefore, a new method was developed by carefully selecting the
relevant parameters and device data sets that can be combined with
a suitable system model.
Figure 3.4 shows the data flow in the performance metric
implementation: technology parameters are used to generate a device
structure (either by process simulation or with analytical dopant
profiles), then a set of device simulations is carried out, and
finally the system performance data are computed from the electrical
device data and system parameters.
Figure 3.4:
Performance metric implementation: data flow
|
Next: 3.3.2 System Model
Up: 3.3 VLSI Performance Metric
Previous: 3.3 VLSI Performance Metric
G. Schrom