What Is The Three Address Instruction Format?

What Is The Three Address Instruction Format?

Nothing much, really. Everything would just be in multiples of 16 rather than multiples of 8. Most machines today operate on 32-bit or 64-bit data anyway so t would continue to do so but the number of bytes per word would be halved. The main consequence could be that of cost. As the minimum unit of computation is 16 bits, many of the higher order bits would get wasted for operations that involve very small numbers. For example, if I had an array of several integers which I knew were in the range of say 0-50 then I would like to declare an array of the smallest unit (byte in Java or char in C) possible, but this would take more memory if the unit was 16-bits instead of 8-bits. As physical memory cost is proportional to the number of bits (not bytes), you would be paying roughly the same cost for 2GB RAM in a 16-bit per byte machine as you would for 4GB RAM in a 8-bit per byte machine. Thus, the number of addressable units you get to work with would be less for the same cost. That said, it isn't necessarily better to work with a smaller unit like 4-bit bytes because then your 64-bit machine has to support one more level of operation (e.g. adding 64-bit integers, 32-bit integers, 16-bit integers, 8-bit integers AND 4-bit integers). This would cause the instruction set and ALU circuitry to be slightly more complex. In the end, I suppose it's just a trade-off but it's always good to have a unit that is a power of 2 so that addressing the bits becomes efficient.