Apple Inc.

Applied Micro Circuits Corporation

Intrinsity, Inc.

Timothy A Olson

James S Blomgren

Jon A Loschke

Terence M Potter

Terrence Matthew Potter

Andrew M Havlir

Robert A Drebin

Michael A Geary

Donald W Glowka

Jeffrey B Reed

Thomas W Rudwick, Iii

Christophe Harle

Douglas C Youngwith

Instructions may require one or more operands to be executed, which may be provided from a register file. In the context of a GPU, however, a register file may be a relatively large structure, and reading from the register file may be energy and/or time intensive An operand cache may be used to store a subset of operands, and may use less power and have quicker access times than the register file. Hint values may be used in some embodiments to suggest that a particular operand should be stored in the operand cache (so that is available for current or future use). In one embodiment, a hint value indicates that an operand should be cached whenever possible. Hint values may be determined by software, such as a compiler, in some embodiments. One or more criteria may be used to determine hint values, such as how soon in the future or how frequently an operand will be used again.

Hint values for use with an operand cache

A system and method for efficiently accessing operands in a datapath. An apparatus includes a data operand register file and an execution pipeline with multiple stages. In addition, the apparatus includes a result bypass cache configured to store data results conveyed by at least the final stage of the execution pipeline stage. Control logic is included which is configured to determine whether source operands for an instruction entering the pipeline are available in the last stage of the pipeline or in the result bypass cache. If the source operands are available in the last stage of the pipeline or the result bypass cache, they may be obtained from one of those locations rather than reading from the register file. If the source operands are not available from the last stage or the result bypass cache, then they may be obtained from the data operand register file.

Result bypass cache

Instructions may require one or more operands to be executed, which may be provided from a register file. In the context of a GPU, however, a register file may be a relatively large structure, and reading from the register file may be energy and/or time intensive An operand cache may store a subset of operands, and may use less power and have quicker access times than the register file. In some embodiments, intelligent operand prefetching may speed execution by reducing memory bank conflicts (e.g., conflicts within a register file containing multiple memory banks). An unused operand slot for another instruction (e.g., an instruction that does not require a maximum number of source operands allowed by an instruction set architecture) may be used to prefetch an operand for another instruction in one embodiment. Prefetched operands may be stored in an operand cache, and prefetching may occur based on software-provided information.

Intelligent caching for an operand cache

Instructions may require one or more operands to be executed, which may be provided from a register file. In the context of a GPU, however, a register file may be a relatively large structure, and reading from a register file may be energy and/or time intensive An operand cache may be used to store a subset of operands, and may use less power and have quicker access times than the register file. Selectors (e.g., multiplexers) may be used to read operands from the operand cache. Power savings may be achieved in some embodiments by activating only a subset of the selectors, which may be done by activators (e.g. flip-flops). Operands may also be concurrently provided to two or more locations via forwarding, which may be accomplished via a source selection unit in some embodiments. Operand forwarding may also reduce power and/or speed execution in one or more embodiments.

Operand cache design

A system and method are provided for updating a speculative global history prediction record in a microprocessor system using pipelined instruction processing. The method accepts microprocessor instructions with consecutive operations, including a conditional branch operation with an associated first branch address. A speculative global history record (SGHR) of conditional branch resolutions and predictions is accessed and hashed with the first branch address, creating a first hash result. The first hash result is used to index a branch history table (BHT) of previous first branch resolutions. As a result, a first branch prediction is made, and the SGHR is updated with the first branch prediction. A non-speculative global history record (NSGHR) of branch resolutions is updated with the resolution of the first branch operation, and if the first branch prediction is incorrect, the SGHR is corrected using the NSGHR.

Method for repairing a speculative global history record

System and method for repairing a speculative global history record

A system and method are provided for updating a global history prediction record in a microprocessor system using pipelined instruction processing. The method accepts a microprocessor instruction of consecutive operations, including a conditional branch operation with an associated branch address, at a first stage in a pipelined microprocessor execution process. A global history record (GHR) of conditional branch resolutions and predictions is accessed and hashed with the branch address, creating a first hash result. The first hash result is used to access an indexed branch history table (BHT) of branch direction counts and the BHT is used to make a branch prediction. If the branch prediction being 'taken', the current GHR value is left-shifted and hashed with the branch address, creating a second hash result which is used in creating an updated GHR.

Global history branch prediction updating responsive to taken branches

System and method for speculative global history prediction updating

The matching algorithm of the layout synthesis method and apparatus disclosed locates transistor pattern matches in a design, links a parameterized tile to each identified match, and adjusts certain variable parameters of the linked parameterized tile to meet the physical design requirements of each located match. Each transistor pattern corresponds to a parameterized tile, which is an actual physical representation of the corresponding pattern and includes variable parameters, which may include transistor size. The matching algorithm locates matches in the design for an ordered list of patterns, names each located match, links the proper parameterized tile to each named match, and adjusts the tile's variable parameters as required. Transistors in the design are included in one and only one named located match.

Physical realization of dynamic logic using parameterized tile partitioning

This invention discloses a group of instructions, blockand blockin a matrix processorthat rearranges data between vector and matrix forms of an A&#xd7;B matrix of datawhere the data matrix includes one or more 4&#xd7;4 sub-matrices of dataThe instructions of this invention simultaneously swaps row or columns between the firstsecondthirdand fourthmatrix registers according to the instructions that perform predefined matrix tensor operations on the data matrix that includes one of the following group of operations: swapping rows between the different individual matrix registers, or swapping columns between the different individual matrix registers. Additionally, successive iterations or combinations of the blockand or blockinstructions perform standard tensor matrix operations from the following group of matrix operations: transpose, shuffle, and deal.

Rearranging data between vector and matrix forms in a SIMD matrix processor

A method and apparatus that minimizes instruction gaps behind a branch instruction in a multistage pipelined processor is disclosed. A pre-branch instruction that corresponds to a branch instruction to inserted into the instruction stream a sufficient number of instructions ahead of the branch instruction to insure that the pre-branch instruction exits the decode stage of the pipeline at the same time the branch instruction exits the first instruction fetch stage of the pipeline. The pre-branch instruction is decoded and causes the instruction fetch unit either to begin fetching instructions at a target address, where the branch is known or predicted to be taken, or to continue fetching instructions along the main execution path, the branch is known or predicted to be not taken.

Growing community of inventors

Austin, TX, United States of America

Timothy A Olson