US 4,903,196 AGranted

Method and apparatus for guaranteeing the logical integrity of data in the general purpose registers of a complex multi-execution unit uniprocessor

AssigneeInternational Business Machines Corporation

InventorsJames H. Pomerene, Thomas R. Puzak, Rudolph N. Rechtschaffen, Frank J. Sparacio

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method and apparatus for controlling access to its general purpose registers (GPRs) by a high end machine configuration including a plurality of execution units within a single CPU. The invention allows up to "N" execution units to be concurrently executing up to "N" instructions using the GPR sequentially or different GPR's concurrently as either SINK or SOURCE while at the same time preserving the logical integrity of the data supplied to the execution units. The use of the invention allows a higher degree of parallelism in the execution of the instructions than would otherwise be possible if only sequential operations were performed. A series of special purpose tags are associated with each GPR and execution unit. These tags are used together with control circuitry both within the GPR's, within the individual execution units and within the instruction decode unit, which permit the multiple use of the registers to be accomplished while maintaining the requisite logical integrity.

Claims20

Dependent claims

In a complex high speed-multi-execution unit uniprocessor system having N-execution units and a plurality of General Purpose Registers (GPRs) each of which is accessible to each of the execution units, said system including an instruction decoder for decoding a sequence of program instructions to be executed by said system during an instruction decoding cycle, said N-execution units concurrently executing up to N decoded instructions during subsequent execution cycles, said execution units using the same GPR sequentially or different GPR's concurrently, the improvement comprising;

means for preserving the logical integrity of the data within the individual GPRs during simultaneous execution of sequential instructions, including supplementary storage means associated with each GPR for storing the unique IDs of execution units which are assigned usage of the specific GPR,
means operative during the decoding cycle for assigning specific GPR Registers to be used as sources and sinks for each instruction,
means for assigning an execution unit to each instruction by means of an identifying tag (ID) unique to such execution unit, stored in the storage means associated with respective GPRs,
means operative during the execution cycle of said system for allowing an instruction to be processed by an execution unit when it is determined that the instruction is independent of the execution of one or more preceding instructions,
said means for determining if an instruction is independent comprising means for determining if an identified GPR register to be used as a Source or Sink for a particular instruction being executed has an execution unit ID value stored in said supplementary storage means as the result of the execution of a preceeding instruction, that matches an execution unit ID value stored in the current execution unit and wherein,
said means for decoding operates substantially independently of the execution of dependent instructions by the execution units.

3 more elements

In a complex high speed multi-execution unit uniprocessor system having N-execution units and a plurality of General Purpose Registers (GPRs) each of which is accessible to each of the execution units, said system including an instruction decoder for decoding a sequence of program instructions to be executed by said system during an instruction decoding cycle, said "N" execution units concurrently executing up to "N" decoded instructions during subsequent execution cycles, said execution units using the same GPR sequentially or different GPR's concurrently, the improvement comprising a means for preserving the logical integrity of the data within the individual GPRs during simultaneous execution of sequential instructions, said means comprising:

at least four special tag fields associated with each GPR, a SINK FORWARD TAG, a SINK EXECUTE TAG, and at least one SOURCE TAG/SOURCE CRT pair, said tag fields controlling access by an execution unit to a GPR during said execution cycle,
each execution unit having associated therewith instruction register storage means having, in addition to fields for storing an instruction to be executed by the execution unit, at least two special purpose tag fields for storing an identification (ID) tag unique to that execution unit and at least one SINK V SOURCE TAG, and
a control mechanism including:
assigning means for assigning instruction to execution units, loading means for loading said special purpose tag fields in said execution units and said GPRs during said instruction decoding cycle of the system operation, said control mechanism thereby preparing the execution units to execute instructions, said loading means loading a particular GPR's SINK FORWARD TAG field with the ID TAG of a first execution unit assigned an instruction which uses that GPR as a SINK and further loading the first execution unit's SINK V SOURCE TAG field with the ID TAG of a second execution unit which uses the particular GPR as a SINK immediately prior to the first execution units use of the GPR, as a source,
means operable, as each of the execution units complete execution of an assigned instruction, to load a particular GPR's SINK EXECUTE TAG with an ID TAG of the execution unit which has just finished using it, thereby allowing the execution unit whose SINK V SOURCE TAG matches the particular GPR's current SINK EXECUTE TAG to access the particular GPR and wherein,
said means for assigning operating substantially independently of the execution of dependent instructions by the execution units.

3 more elements

Dependencies are inferred from the claim text.