Q. Pros and Cons of general purpose register architecture?
General Purpose Register Architecture: Pros and Cons
- Registers can be employed to store variables as it decreases memory traffic and boosts up execution. It also increases code density as register names are shorter than memory addresses.
- Instructions shouldcomprise bits to specify which register to operate on henceforth large instruction size than accumulator type machines.
- Memory access can be minimized (registers can hold many intermediate values).
- Implementation is complex as compiler writer has to try to maximize register usage.
While most early machines used accumulator architectures or stack in the last 15 years all CPUs made are GPR processors. The 3 main reasons are that registers are faster than memory; the more data that can be kept internally in CPU the faster the program will execute. The third reason is that registers are easier for a compiler to use.
However while CPU's with GPR were clearly better than previous stack and accumulator based CPU's yet they were lacking in various areas. The areas being: Instructions were of varying length from 1 byte to 6-8 bytes. This causes problems with pre-fetching and pipelining of instructions. ALU instructions can have operands which were memory locations since the time to access memory is slower and so does the whole instruction.
So in the early 1980s the concept of RISC was presented. RISC signifies Reduced Instruction Set Computer. Unlike CISC this ISA uses fewer instructions with simple constructs so they can be executed much faster within CPU without having to use memory as often. First RISC CPU, the MIPS 2000 has 32 GPRs. MIPS is a store/load architecture that means only load and store instructions access memory. All other computational instructions preform only on values stored in registers.