Method and apparatus for delaying writing back the results of instructions to a processor
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06F-009/312
G06F-009/38
출원번호
US-0479627
(1990-02-14)
발명자
/ 주소
Patel Piyush G. (Fremont CA)
출원인 / 주소
Intel Corporation (Santa Clara CA 02)
인용정보
피인용 횟수 :
40인용 특허 :
0
초록▼
The present invention describes an integer execution unit register file having one fewer write port by employing delayed writeback for data transfer instructions in a high speed processor. The integer execution unit comprises a register file with 32 separate registers, each 32-bits long. The registe
The present invention describes an integer execution unit register file having one fewer write port by employing delayed writeback for data transfer instructions in a high speed processor. The integer execution unit comprises a register file with 32 separate registers, each 32-bits long. The register file is a write through register file. A four-stage instruction pipeline is employed to execute all integer instructions. The four stages are (1) Fetch, (2) Decode, (3) Execute, and (4) Writeback. For data transfer type of instructions such as, a load instruction, one extra instruction stage is usually required. The prior art processors add one extra write port to accommodate such data transfer type of instructions. The present invention delays the writing of the data transfer type instruction until the writeback stage of the next data transfer instruction. The result of the data transfer type instruction returns at the end of the writeback stage. The result is held in a temporary register. All references to the result of such a data type transfer instruction will be bypassed from the temporary register to the proper execution block. The data from the temporary register is written back into the register file only at the writeback stage of the next data transfer type instruction. Thus, the present invention resolves the conflict of resources for the write port and at the same time saves one port to the register file. The present invention not only reduces the amount of silicon area required for high speed processors, it also reduces the layout complexity. It follows that the present invention improves the overall speed of the processor.
대표청구항▼
In a processor having a plurality of memory cells, an integer execution unit, and a bus control unit for processing instructions, including data transfer instructions, in a parallel instruction pipeline, said pipeline including Fetch, Decode, Execute, and Writeback stages, the integer execution unit
In a processor having a plurality of memory cells, an integer execution unit, and a bus control unit for processing instructions, including data transfer instructions, in a parallel instruction pipeline, said pipeline including Fetch, Decode, Execute, and Writeback stages, the integer execution unit comprising: a register file comprising at least one write port for writing data and at least one read port for reading data, said register file further having a plurality of registers; a bypass control unit, connected to said read and write ports, for delaying writing of a result of a data transfer instruction into the register file through said write port until the writeback stage of a next data transfer instruction; a shifter, connected to said control bypass unit, for performing logical and arithmetic shift operations; an arithmetic logic unit (“ALU”), connected to said shifter, for calculating effective addresses for said instructions; an internal bus over which data is communicated; an instruction pointer unit, connected to the internal bus, for pointing to the effective address of a next instruction; an instruction control logic unit, connected to said control bypass unit over said internal bus, for storing the next instruction to be executed by said integer execution unit.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허를 인용한 특허 (40)
Nakada Tatsumi (Kawasaki JPX), Apparatus for operand data bypassing having previous operand storage register connected between arithmetic input selecto.
Haines Ralph Warren ; O'Neill Dan Craig ; Pries Stephen C. ; Miller William V. ; Waterson Kent B. ; Weinman David S. ; Shay Michael J. ; Pang Jianhua Helen ; Herrington Daniel R. ; Marley Brian J. ; , Integrated circuit with multiple functions sharing multiple internal signal buses according to distributed bus access an.
Lauritzen Mogens ; Weiss Richard A., Method and apparatus for communicating integer and floating point data over a shared data path in a single instruction.
Naffziger, Samuel D., Multi-port computer register file having shared word lines for read and write ports and storage elements that power down or enter a high-impendance state during write operations.
Johannes Wang ; Sanjiv Garg ; Trevor Deosaran, System and method for retiring approximately simultaneously a group of instructions in a superscalar microprocessor.
Wang Johannes ; Garg Sanjiv ; Dcosaran Trevor, System and method for retiring approximately simultaneously a group of instructions in a superscalar microprocessor.
Wang, Johannes; Garg, Sanjiv; Deosaran, Trevor, System and method for retiring approximately simultaneously a group of instructions in a superscalar microprocessor.
Wang, Johannes; Garg, Sanjiv; Deosaran, Trevor, System and method for retiring approximately simultaneously a group of instructions in a superscalar microprocessor.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.