MIPS architecture has a “zero” one of these things to increase speed. In (15[1])Verilog, declared wires are replaced with variables representing these things when using sequential logic. Rematerialization is used while allocating these things during compilation in order to prevent spilling. Connecting a series of flip-flops so the output of one feeds into the input of another creates the (*) “shift” type of these things, which are used in the counters that make up ALUs. The majority of these things in 64-bit architectures like Intel’s x86 are 64 bits wide. In a load-store architecture, arithmetic operations can only be performed on values from these things after they are read from memory. For 10 points, machine code instructions write to what CPU memory units that are smaller and faster than caches? ■END■