Table 1 shows the five model input parameters.
Table 1: CACTI input parameters
In addition, there are six array organization parameters
that are used to estimate the cache access and cycle time.
In the basic organization discussed by Wada [2],
a single set shares a common wordline. Figure 2-a shows this
organization, where B is the block size (in bytes), A is the
associativity, and S is the number of sets
( 
).
Clearly, such an organization could result in an array that is much larger
in one direction than the other, causing either the bitlines or wordlines
to be very slow.
To alleviate this problem, Wada describes how the array can be broken
horizontally and vertically and defines two parameters,
and
,
which indicates to what extent the array has been divided.
Figure 2-b introduces another organization parameter,
.
This parameter indicates how many sets are mapped to a single wordline,
and allows the overall access time of the array to be changed without
breaking it into smaller subarrays.
The optimum values of , , and
depend on the cache and block sizes, as well as the associativity.
Figure 2: Cache organization parameter
We assume that the tag array can be configured independently of the data array.
Thus, there are also three tag array
parameters: 
,
, and 
.