Classically, digital computers have had two basic modes of sequencing instructions. In the first, normal sequencing, each instruction has a unique successor which may be defined by an instruction counter or by a next instruction address within the insm~ction itself. The second mode is the seleetior~ of an alternate sequence by a branching, skipping, or suppression operation. A third mode of sequencing, program interruption, has been recognized more recently , although computers as early as UN~vxc I contained rudimentary interruption provisions. In this mode, execution of a sequenee of instructions may be interrupted at an arbitrary point, and specification of the next instruction in this case is completely independent of the last instruction executed. Provision may or may nol~ be made for saving the current location in the interrupted sequence. In branching, the selection of an alternate instrttction implies the selection of a new sequence, i.e., the alternate instruction specifies or implies (through an instruction counter) its own successor. The same may be true of interruption; or, an interruptior~ syst, em may be defined so that the interrupting instruction does not change the instruction counter. In this case the interrupting instruction, if a branch, m a y specify its own successor, but if it does not, ~,he successor is implied by the last instruction of the interrupted sequence . I t is often desirable for an instruction sequence ~co execute a single, noninterrupt, ing instruction which does not specify or imply its own successor. For this purpose the Execute operations have been independently developed by several groups. These may be considered a fourth mode of instruction sequencing. The four modes can be summarized briefly by stating the four possible relationships between an original sequence A and a second sequence B:
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=367168.