International Business Machines Corporation

Apple Inc.

Intel Corporation

Mary Douglass Brown

Christopher Michael Abernathy

Dung Q Nguyen

Hung Qui Le

Todd Alan Venton

William Elton Burky

James Allan Kahle

Susan Elizabeth Eisen

Joel Abraham Silberman

Ronald P Hall

Robert Allen Cordes

James Wilson Bishop

Maureen Anne Delaney

Mahesh K Reddy

Jafar Nahidi

Balaji Kadambi

Bartholomew Blaner

Sundeep Chadha

Jeffrey Carl Brownscheidle

Balaram Sinharoy

Ian D Kountanis

David James Williamson

Brett Stanley Feero

Jared Warner Stark, Iv

John Barry Griswell, Jr

Saiful Islam

Sam G Chu

Niket K Choudhary

Jonathan Jay Tyler

Douglas C Holman

Bjorn P Christensen

Ethan R Schuchman

Ilhyun Kim

Abhishek Kumar

Siavash Zangeneh Kamali

A cache subsystem is disclosed. The cache subsystem includes a cache configured to store information in cache lines arranged in a plurality of ways. A requestor circuit generates a request to access a particular cache line in the cache. A prediction circuit is configured to generate a prediction of which of the ways includes the particular cache line. A comparison circuit verifies the prediction by comparing a particular address tag associated with the particular cache line to a cache tag corresponding to a predicted one of the ways. Responsive to determining that the prediction was correct, a confirmation indication is stored indicating the correct prediction. For a subsequent request for the particular cache line, the cache is configured to forego a verification of the prediction that the particular cache line is included in the one of the ways based on the confirmation indication.

Prediction confirmation for cache subsystem

An apparatus includes an instruction cache circuit and an instruction fetch circuit. The instruction fetch circuit is configured to retrieve, from the instruction cache circuit, a fetch group that includes a plurality of instructions for execution by a processing circuit, and to make a determination that the fetch group includes a control transfer instruction that is predicted to be taken. A target address associated with the control transfer instruction is directed to an instruction within the fetch group. The instruction fetch circuit is further configured to, based on the determination, alter instructions within the fetch group in a manner that is based on a type of the control transfer instruction.

Using a next fetch predictor circuit with short branches and return fetch groups

A processor includes a first level register file, second level register file, and register file mapper. The first and second level register files are comprised of physical registers, with the first level register file more efficiently accessed relative to the second level register file. The register file mapper is coupled with the first and second level register files. The register file mapper comprises a mapping structure and register file mapper controller. The mapping structure hosts mappings between logical registers and physical registers of the first level register file. The register file mapper controller determines whether to map a destination logical register of an instruction to a physical register in the first level register file. The register file mapper controller also determines, based on metadata associated with the instruction, whether to write data associated with the destination logical register to one of the physical registers of the second level register file.

Processor for avoiding reduced performance using instruction metadata to determine not to maintain a mapping of a logical register to a physical register in a first level register file

A system and process for managing thread execution includes providing two data register sets coupled to a processor and using, by the processor, the two register sets as first-level registers for thread execution. A portion of main memory or cache memory is assigned as second-level registers where the second-level registers serve as registers of at least one of the two data register sets for executing the threads. Data for the threads may be moved between the first-level registers and second-level registers for different modes of thread processing.

Thread transition management

A system and process for managing thread transitions includes providing two data register sets coupled to a processor and using, by the processor, the two register sets as first-level registers for thread execution. A determination is made whether a quantity of the first-level registers needed to execute one or more threads exceeds a quantity of the first-level registers of the two data register sets. Responsive to determining that the quantity of the first-level registers needed to execute the one or more threads exceeds the quantity of the first-level registers of the two data register sets, a portion of main memory or cache memory is assigned as second-level registers where the second-level registers serve as registers of at least one of the two data register sets for executing the one or more threads.

A system and process for managing thread transitions includes determining that a transition is to be made regarding the relative use of two data register sets where the two data register sets are used by a processor as first-level registers for thread execution. Based on the transition determination, a determination is made whether to move thread data in at least one of the first-level registers to second-level registers. Responsive to determining to move the thread data, a portion of main memory or cache memory is assigned as the second-level registers where the second-level registers serve as registers of at least one of the two data register sets for executing a thread. The thread data from the at least one first-level register is moved to the second-level registers based on the move determination.

In an embodiment, an apparatus includes a cache memory and a control circuit. The control circuit may be configured to pre-fetch and store a first quantity of instruction data in response to a determination that a first pre-fetch operation request is received after a reset and prior to a first end condition. The first end condition may depend on an amount of unused storage in the cache memory. The control circuit may be further configured to pre-fetch and store a second quantity of instruction data in response to a determination that a second pre-fetch operation request is received after the first end condition. The second quantity may be less than the first quantity.

Methods for core recovery after a cold start

A register file bypass controller in communication with a set of one or more bypass registers, the register file bypass controller configured to receive a register file bypass request; determine whether to grant the register file bypass request; determine whether data identified by the register file bypass request is present in the set of one or more bypass registers in response to determining to grant the register file bypass request; determine a selected bypass register in the set of one or more bypass registers in response to determining the data identified by the register file bypass request is not present in the set of one or more bypass registers; determine to store the data identified by the register file bypass request in the selected bypass register; and notify an execution unit to cancel instruction execution associated with the data identified by the register file bypass request.

Bypassing a higher level register file in a processor having a multi-level register file and a set of bypass registers

Various systems, processes, products, and techniques may be used to manage thread transitions. In particular implementations, a system and process for managing thread transitions may include the ability to determine that a transition is to be made regarding the relative use of two data register sets and determine, based on the transition determination, whether to move thread data in at least one of the data register sets to second-level registers. The system and process may also include the ability to move the thread data from at least one data register set to second-level registers based on the move determination.

A processor includes an execution unit, a first level register file, a second level register file, a plurality of storage locations and a register file bypass controller. The first and second level register files are comprised of physical registers, with the first level register file more efficiently accessed relative to the second level register file. The register file bypass controller is coupled with the execution unit and second level register file. The register file bypass controller determines whether an instruction indicates a logical register is unmapped from a physical register in the first level register file. The register file controller also loads data into one of the storage locations and selects one of the storage locations as input to the execution unit, without mapping the logical register to one of the physical registers in the first level register file.

Efficient usage of a multi-level register file utilizing a register file bypass

A processor includes an instruction fetch unit, an issue queue coupled to the instruction fetch unit, an execution unit coupled to the issue queue, and a multi-level register file including a first level register file having lower access latency and a second level register file having higher access latency. Each of the first and second level register files includes a plurality of physical registers for holding operands that is concurrently shared by a plurality of threads. The processor further includes a mapper that, at dispatch of an instruction specifying a source logical register from the instruction fetch unit to the issue queue, initiates a swap of a first operand associated with the source logical register that is in the second level register file with a second operand held in the first level register file. The issue queue, following the swap, issues the instruction to the execution unit for execution.

Multi-level register file supporting multiple threads

Growing community of inventors

Austin, TX, United States of America

Mary Douglass Brown