| Table 10.1 Baseline Mark I area model. |
|
| Integer Unit Area Components |
|
| Integer ALU | |
| Integer register file with bypassa | |
| Shifter | |
| Incrementer | |
| Program counter unit | |
| TLBb | |
| Decoder hardware | |
| Cache controller | |
| Bus logic | |
| Store buffer | |
| Load/Store byte support | |
| Clock generation | |
| Total integer unit area | |
| Floating Point Unit Area Components |
|
| Register filec | |
| Adder | |
| Multiplier | |
| Division supportd | |
| Total floating-point unit area | |
| Memory Subsystem Area Components |
|
| 16 KBI-cachee | |
| 8 KBD-cachef | |
| Total memory subsystem area | |
| Processor Overhead Area Components |
|
| Integer unit (50%) | |
| Floating-point unit (50%) | |
| Memory subsystem (15%) | |
| Total processor overhead area | |
|
| Total processor area | |
| Die overhead area (20% total processor area) | |
| Total die area | |
| aThe integer register file consists of 32 single-word integer registers that are paired to support double word values. |
| bThere are two TLB's, each of which is 2-way set-associative with LRU replacement and each has 128 entries (64 ´ 2). |
| cThe floating point register file consists of 32 IEEE Single Precision registers that are paired to support Double Precision values. |
| dAdditional area allows multiplier to perform divisiona separate divider element would require 23.3A, which includes both a multiplier and division support. |
| eThe cache is 8kB direct mapped, 4 word per line (16 byte per line), write-through no write-allocate. |
| fThe D-cache is a copyback cache. Both caches have 16 bytes per line and are 4-way set-associative. |