Figure 2 depicts a block diagram of the proposed approach
for a L1 cache. The filter consists in comparing only a subset
of bits of the tags, say X. For this purpose the tag array is
decoupled in two structures: one X-bit wide and the other one
N −X bits wide. The TF-Cache employs the least significant
bits of the tags stored in the X-bit wide structure to reduce
the number of accessed ways.