Link Details

Branch Delay Slots are one of the awkward features of RISC architectures. RISC CPUs are pipelined by definition, so while the current instruction is in execution, the following instruction(s) will be in the pipeline already. If there is for example a conditional branch in the instruction stream, the CPU cannot know whether the next instruction is the one following the branch or the instruction at the target location until it has evaluated the branch. This would cause a bubble in the pipeline; therefore some RISC architectures have a branch delay slot: The instruction after the branch will always be executed, no matter whether the branch is taken or not. So in practice, you can put the instruction that would be before the branch right after the branch, if this instruction is independent of the branch instruction, i.e. doesn’t access the same registers. Otherwise, you can fill it with a NOP. Out-of-order architectures can do this reordering at runtime, so there would be no need for a delay slot. Nevertheless, the delay slot is a feature of the architecture, not the implementation.