Browse Prior Art Database

Conditionally Ending the Microprogram Sequence of an Instruction

IP.com Disclosure Number: IPCOM000085696D
Original Publication Date: 1976-May-01
Included in the Prior Art Database: 2005-Mar-02
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Curlee, TO: AUTHOR [+4]

Abstract

The first cycle control store (FCCS) described in U.S. Patent 3,800,293 provides immediate access to the first control word of the next instruction. The technique described takes advantage of a FCCS to allow for the conditional end operation (CEOP) of an instruction. Further, the technique can be accomplished in one machine cycle, so no time is lost between instructions.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 89% of the total text.

Page 1 of 1

Conditionally Ending the Microprogram Sequence of an Instruction

The first cycle control store (FCCS) described in U.S. Patent 3,800,293 provides immediate access to the first control word of the next instruction. The technique described takes advantage of a FCCS to allow for the conditional end operation (CEOP) of an instruction. Further, the technique can be accomplished in one machine cycle, so no time is lost between instructions.

Basically, the output of the FCCS is treated as one of five target words for the next cycle (the other four coming from the control store buffer). This is a change from U.S. Patent 3,800,293 wherein the first cycle word replaced one of the four branch target words. The procedure now would be to bypass the conventional branching mechanism entirely for EOP handling.

The advantage of this scheme is that it is faster. The conventional branch logic must be able to handle a large number of conditional branches; hence it is slow. But the choice to bypass it or not can be made very quickly, if the number of conditions upon which the End-Op will be done is few (as is normally the case). In fact, it is fast enough to avoid loss of cycles between instructions.

The need for this technique arises from the use of a two-cycle buffer. This type of buffer requires a "prefetch" of the next control word. Specifying, CEOP under the old technique caused a "prefetch" to be done before the (slower) branch logic actually determined that the End-Op would be...