Arm Limited

International Business Machines Corporation

Chris Abernathy

Yasuo Ishii

Joseph Michael Pusdesris

Muhammad Umar Farooq

Abhishek Raja

Michael Filippo

Klas Magnus Bruce

Florent Begon

Adrian Montero

Timothy Hume Heil

Ian Michael Caulfield

Miles Robert Dooley

Jason Nathaniel Dale

Michael David Achenbach

Jeffrey Powers Bradford

Dam Sunwoo

Max John Batley

Huzefa Moiz Sanjeliwala

Balaji Vijayan

Krishnendra Nathella

Nicholas Andrew Plante

Eric Charles Quinnell

Circuitry comprises two or more clusters of execution units, each cluster comprising one or more execution units to execute processing instructions; and scheduler circuitry to maintain one or more queues of processing instructions, the scheduler circuitry comprising picker circuitry to select a queued processing instruction for issue to an execution unit of one of the clusters of execution units for execution; in which: the scheduler circuitry is configured to maintain dependency data associated with each queued processing instruction, the dependency data for a queued processing instruction indicating any source operands which are required to be available for use in execution of that queued processing instruction and to inhibit issue of that queued processing instruction until all of the required source operands for that queued processing instruction are available and is configured to be responsive to an indication to the scheduler circuitry of the availability of the given operand as a source operand for use in execution of queued processing instructions; and the scheduler circuitry is responsive to an indication of availability of one or more last awaited source operands for a given queued processing instruction, to inhibit issue by the scheduler circuitry of the given queued processing instruction to an execution unit in a cluster of execution units other than a cluster of execution units containing an execution unit which generated at least one of those last awaited source operands.

Circuitry and method

A predicated-loop-terminating branch instruction controls, based on whether a loop termination condition is satisfied, whether the processing circuitry should process a further iteration of a predicated loop body or process a following instruction. If at least one unnecessary iteration of the predicated loop body is processed following a mispredicted-non-termination branch misprediction when the loop termination condition is mispredicted as unsatisfied for a given iteration when it should have been satisfied, processing of the at least one unnecessary iteration of the predicated loop body is predicated to suppress an effect of the at least one unnecessary iteration. When the mispredicted-non-termination branch misprediction is detected for the given iteration of the predicated-loop-terminating branch instruction, in response to determining that a flush suppressing condition is satisfied, flushing of the at least one unnecessary iteration of the predicated loop body is suppressed as a response to the mispredicted-non-termination branch misprediction.

Responding to branch misprediction for predicated-loop-terminating branch instruction

A data processing apparatus is provided that includes global-history prediction circuitry that provides a prediction of an outcome of a given control flow instruction based on a result of execution of one or more previous control flow instructions. Correction circuitry provides a corrected prediction of the global-history prediction circuitry in respect of the given control flow instruction and cache circuitry, separate from the correction circuitry, stores the corrected prediction in respect of the given control flow instruction.

Caching override indicators for statistically biased branches to selectively override a global branch predictor

An apparatus is provided that includes a memory hierarchy comprising a plurality of caches and a memory. Prefetch circuitry acquires data from the memory hierarchy before the data is explicitly requested by processing circuitry configured to execute a stream of instructions. Writeback circuitry causes the data to be written back from a higher level cache of the memory hierarchy to a lower level cache of the memory hierarchy and tracking circuitry tracks a proportion of entries that are stored in the lower level cache of the memory hierarchy having been written back from the higher level cache of the memory hierarchy, that are subsequently explicitly requested by the processing circuitry in response to one of the instructions.

Writebacks of prefetched data

Apparatus and a method of operating the same is disclosed. Instruction fetch circuitry is provided to fetch a block of instructions from memory and branch prediction circuitry to generate branch prediction indications for each branch instruction present in the block of instructions. The branch prediction circuitry is responsive to identification of a first conditional branch instruction in the block of instructions that is predicted to be taken to modify a branch prediction indication generated for the first conditional branch instruction to include a subsequent branch status indicator. When there is a subsequent branch instruction after the first conditional branch instruction in the block of instructions that is predicted to be taken the subsequent branch status indicator has a first value, and otherwise the subsequent branch status indicator has a second value. This supports improved handling of a misprediction as taken.

Misprediction of predicted taken branches in a data processing apparatus

Branch prediction circuitry processes blocks of instructions and provides instruction fetch circuitry with indications of predicted next blocks of instructions to be retrieved from memory. Main branch target storage stores branch target predictions for branch instructions in the blocks of instructions. Secondary branch target storage caches the branch target predictions from the main branch target storage. Look-ups in the secondary branch target storage and the main branch target storage are performed in parallel. The main branch target storage is set-associative and an entry in the main branch target storage comprises multiple ways, wherein each way of the multiple ways stores a branch target prediction for one branch instruction. The branch prediction circuitry stores a way prediction for which of the multiple ways contain the branch target predictions for a predicted next block of instructions and stores a flag associated with the way prediction indicating whether all branch target predictions stored for the predicted next block of instructions in the main branch target storage are also stored in the secondary branch target storage. An active value of the flag suppresses the look-up in the main branch target storage for the predicted next block of instructions.

Branch target look up suppression

A data processing apparatus is provided that includes lookup circuitry to provide first prediction data in respect of a first block of instructions and second prediction data in respect of a second block of instructions. First processing circuitry provides a first control flow prediction in respect of the first block of instructions using the first prediction data and second processing circuitry provides a second control flow prediction in respect of the second block of instructions using the second prediction data. The first block of instructions and the second block of instructions collectively define a prediction block and the lookup circuitry uses a reference to the prediction block as at least part of an index to both the first prediction data and the second prediction data.

Apparatus and method for performing multiple control flow predictions

A data processing apparatus is provided that includes a plurality of control flow execution circuits to simultaneously execute a first control flow instruction having a first type and a second control flow instruction having a second type from a plurality of instructions. A control flow prediction update circuit updates at most one of: a prediction of the first control flow instruction based on a result of the first control flow instruction, and a prediction of the second control flow instruction based on a result of the second control flow instruction.

Handling multiple control flow instructions

A variety of data processing apparatuses are provided in which stride determination circuitry determines a stride value as a difference between a current address and a previously received address. Stride storage circuitry stores an association between stride values determined by the stride determination circuitry and a frequency during a training period. Prefetch circuitry causes a further data value to be proactively retrieved from a further address. The further address is the current address modified by a stride value in the stride storage circuitry having a highest frequency during the training period. The variety of data processing apparatuses are directed towards improving efficiency by variously disregarding certain candidate stride values, considering additional further addresses for prefetching by using multiple stride values, using feedback to adjust the training process and compensating for page table boundaries.

Prefetching techniques

The present disclosure is concerned with improvements to cache systems that can be used to improve the performance (e.g. hit performance) and/or bandwidth within a memory hierarchy. For instance, a data processing apparatus is provided that comprises a cache. Access circuitry receives one or more requests for data and when the data is present in the cache the data is returned. Retrieval circuitry retrieves the data and stores the data in the cache, either proactively or in response to the one or more requests for the data. Control circuitry evicts the data from the cache and, in dependence on at least one condition, stores the data in the further cache. The at least one condition comprises a requirement that the data was stored into the cache proactively and that a number of the one or more requests is above a threshold value.

Cache storage techniques

Storage circuitry is provided, that is designed to form part of a memory hierarchy. The storage circuitry comprises receiver circuitry for receiving a request to obtain data from the memory hierarchy. Transfer circuitry causes the data to be stored at a selected destination in response to the request, wherein the selected destination is selected in dependence on at least one selection condition. Tracker circuitry tracks the request while the request is unresolved. If at least one selection condition is met then the destination is the storage circuitry and otherwise the destination is other storage circuitry in the memory hierarchy.

Storage circuitry request tracking

There is provided a data processing apparatus that includes processing circuitry for executing a plurality of instructions. Storage circuitry stores a plurality of entries, each entry relating to an instruction in the plurality of instructions and including a dependency field. The dependency field stores a data dependency of that instruction on a previous instruction in the plurality of instructions. Scheduling circuitry schedules the execution of the plurality of instructions in an order that depends on each data dependency. When the previous instruction is a single-cycle instruction, the dependency field includes a reference to one of the entries that relates to the previous instruction, otherwise, the data dependency field includes an indication of an output destination of the previous instruction.

Scheduling in a data processing apparatus

An apparatus has processing circuitry, register rename circuitry and control circuitry which selects one of first and second move handling techniques for handling a move instruction specifying a source logical register and a destination logical register. In the first technique, the register rename circuitry maps the destination logical register of the move to the same physical register as the source logical register. In the second technique, the processing circuitry writes a data value read from a physical register corresponding to the source logical register to a different physical register corresponding to the destination local register. The second technique is selected when the move instruction specifies the same source logical register as one of the source and destination logical registers as an earlier move instruction handled according to the first technique, and the register mapping used for that register when handling the earlier move instruction is still current.

Handling move instructions via register renaming or writing to a different physical register using control flags

Arbitration circuitry is provided for arbitrating between requests awaiting servicing. The requests require variable numbers of resources and the arbitration circuitry permits the request to be serviced in a different order to the order in which they were received. Checking circuitry prevents a given request other than a oldest request from being serviced when a number of available resources is less than a threshold number of resources. The threshold number is varied based on the number of resources required for at least one other request awaiting servicing.

Arbitration of requests requiring a variable number of resources

A single threaded out-of-order processorincludes an architected register fileand a speculative register file. Speculative register allocation circuitryserves to allocate speculative registers for use in accordance with an allocation sequence and taken from a position determined by a tail point. Read suppression circuitryserves to maintain a boundary pointer corresponding to a position within the allocation sequence such that no speculative register more recently allocated within the allocation sequence than that corresponding to the boundary pointer can have a valid register value. The read suppression circuitryserves to suppress read operations for source operands lying within a read-suppression region delimited by the tail point and the boundary pointer. Separate boundary pointers may be maintained for different types of register values, such as integer register values and floating point register values.

Growing community of inventors

Austin, TX, United States of America

Chris Abernathy