Arm Limited

Jonathan Curtis Beard

Curtis Glenn Dunham

Roxana Rusitoru

Stephan Diestelhorst

Wendy Arnott Elsasser

Jamshed Jalal

Eric Van Hensbergen

Reiley Jeyapaul

Andrew Brookfield Swaine

Andreas Lars Sandberg

Klas Magnus Bruce

Kim Richard Schuttenberg

Prakash Shyamlal Ramrakhyani

Alejandro Rico Carro

Steven Douglas Krueger

Eric Ola Harald Liljedahl

Pavel Shamis

Kar-Lik Kasim Wong

Alexander Donald Charles Chadwick

Gareth James Evans

Syed Ali Mustafa Zaidi

Shibo Wang

Alexander Sebastian Bischoff

Luis Emilio Pena

A hinter data processing apparatus is provided with processing circuitry that determines that an execution context to be executed on a hintee data processing apparatus will require a virtual-to-physical address translation. Hint circuitry transmits a hint to a hintee data processing apparatus to prefetch a virtual-to-physical address translation in respect of an execution context of the further data processing apparatus. A hintee data processing apparatus is also provided with receiving circuitry that receives a hint from a hinter data processing apparatus to prefetch a virtual-to-physical address translation in respect of an execution context of the further data processing apparatus. Processing circuitry determines whether to follow the hint and, in response to determining that the hint is to be followed, causes the virtual-to-physical address translation to be prefetched for the execution context of the data processing apparatus. In both cases, the hint comprises an identifier of the execution context.

Translation hints

Message passing circuitry comprises lookup circuitry responsive to a producer request indicating message data provided on a target message channel by a producer node of a system-on-chip, to obtain, from a channel consumer information structure, selected channel consumer information associated with a given consumer node subscribing to the target message channel. Control circuitry writes the message data to a location associated with an address in a consumer-defined region of address space determined based on the selected channel consumer information. When an event notification condition is satisfied for the target message channel and the given consumer node, and an event notification channel is to be used, event notification data is written to a location associated with an address in a consumer-defined region of address space determined based on event notification channel consumer information associated with the event notification channel.

Message passing circuitry and method

An apparatus comprises at least one processor to execute software processes, a memory system to store data for access by the at least one processor, and checkpointing circuitry to trigger saving, to the memory system, of checkpoints of context state associated with at least one software process executed by the at least one processor. The saving of checkpoints is a background process performed by the checkpointing circuitry in the background of execution of the software processes by the at least one processor.

Checkpoint saving

An apparatus and method for handling stash requests are described. The apparatus has a processing element with an associated storage structure that is used to store data for access by the processing element, and an interface for coupling the processing element to interconnect circuitry. Stash request handling circuitry is also provided that, in response to a stash request targeting the storage structure being received at the interface from the interconnect circuitry, causes a block of data associated with the stash request to be stored within the storage structure. The stash request identifies a given address that needs translating into a corresponding physical address in memory, and also identifies an address space key. Address translation circuitry is used to convert the given address identified by the stash request into the corresponding physical address by performing an address translation that is dependent on the address space key identified by the stash request. The stash request handling circuitry is then responsive to the corresponding physical address determined by the address translation circuitry to cause the block of data to be stored at a location within the storage structure associated with the physical address.

Apparatus and method for handling stash requests

An apparatus comprises memory access circuitry to access a memory system; a plurality of memory mapped registers, including at least an insert register and a producer pointer register; and control circuitry to perform an insert operation in response to receipt of an insert request from a requester device sharing access to the memory system. The insert request specifies an address mapped to the insert register and an indication of a payload. The insert operation includes controlling the memory access circuitry to write the payload to a location in the memory system selected based on a producer pointer value stored in the producer pointer register, and updating the producer pointer register to increment the producer pointer value.

Insert operation

A method and apparatus is provided for processing accelerator instructions in a data processing apparatus, where a block of one or more accelerator instructions is executable on a host processor or on an accelerator device. For an instruction executed on the host processor and referencing a first virtual address, the instruction is issued to an instruction queue of the host processor and executed the instruction by the host processor, the executing including translating, by translation hardware of the host processor, the first virtual address to a first physical address. For an instruction executed on the accelerator device and referencing the first virtual address, the first virtual address is translated, by the translation hardware, to a second physical address and the instruction is sent to the accelerator device referencing the second physical address. An accelerator task may be initiated by writing configuration data to an accelerator job queue.

Accelerator interface mechanism for data processing system

Circuitry comprises a set of data handling nodes comprising: two or more master nodes each having respective storage circuitry to hold copies of data items from a main memory, each copy of a data item being associated with indicator information to indicate a coherency state of the respective copy, the indicator information being configured to indicate at least whether that copy has been updated more recently than the data item held by the main memory; a home node to serialise data access operations and to control coherency amongst data items held by the set of data handling nodes so that data written to a memory address is consistent with data read from that memory address in response to a subsequent access request; and one or more slave nodes including the main memory; in which: a requesting node of the set of data handling nodes is configured to communicate a conditional request to a target node of the set of data handling nodes in respect of a copy of a given data item at a given memory address, the conditional request being associated with an execution condition and being a request that the copy of the given data item is written to a destination node of the data handling nodes; and the target node is configured, in response to the conditional request: (i) when the outcome of the execution condition is successful, to write the data item to the destination node and to communicate a completion-success indicator to the requesting node; and (ii) when the outcome of the execution condition is a failure, to communicate a completion-failure indicator to the requesting node.

Circuitry and method

In a cache stash relay, first data, from a producer device, is stashed in a shared cache of a data processing system. The first data is associated with first data addresses in a shared memory of the data processing system. An address pattern of the first data addresses is identified. When a request for second data, associated with a second data address, is received from a processing unit of the data processing system, any data associated with data addresses in the identified address pattern are relayed from the shared cache to a local cache of the processing unit if the second data address is in the identified address pattern. The relaying may include pushing the data from the shared cache to the local cache or a pre-fetcher of the processing unit pulling the data from the shared cache to the local cache in response to a message.

Cache stash relay

A method and apparatus for monitoring cache transactions in a cache of a data processing system is provided. Responsive to a cache transaction associated with a transaction address, when a cache controller determines that the cache transaction is selected for monitoring, the cache controller retrieves a pointer stored in a register, determines a location in a log memory from the pointer, and writes a transaction identifier to the determined location in the log memory. The transaction identifier is associated with the transaction address and may be a virtual address, for example. The pointer is updated and stored to the register. The architect of the apparatus may include a mechanism for atomically combining data access instructions with an instruction to commence monitoring.

Method and apparatus for architectural cache transaction logging

A system, apparatus and method for accessing an electronic storage medium, such as a memory location storing a page table, or range table. A virtual address of the electronic storage medium is identified that corresponds to designated portions, such as a range of addresses of the electronic storage medium. The virtual address is translated to a corresponding physical address and one or more commands are identified as being excluded from execution in the designated portions of the electronic storage medium. This may be accomplished by using a routine such as mprotect( ). A fault indication, or decoration, is provided to meta-data associated with the physical address, which is associated with the designated portions of the electronic storage medium when excluded commands are provided to the physical address. A mechanism, such as hardware, is actuated when the fault is generated.

System, method and apparatus for fine granularity access protection

A system, apparatus and method for enabling a FIFO-like (first-in-first-out) communication between a plurality of executing processes that are distributed throughout a computing system. Embodiments exploit locality in the hierarchy of the cache memory and communication busses within the computing system to enable the passing of messages or streams of bytes with a low latency and high throughput. In addition, this allows for participating components to be very simple, or very sophisticated, but still benefit from the improved communications patterns.

System, method and apparatus for inter-process communication

A memory system for a data processing apparatus includes a fault management unit, a memory controller (such as a memory management unit or memory node controller), and one or more storage devices accessible via the memory controller and configured for storing critical data. The fault management unit detects and corrects a fault in the stored critical data, a storage device or the memory controller. A data fault may be corrected using a copy of the data, or an error correction code, for example. A level of failure protection for the critical data, such as a number of copies, an error correction code or a storage location in the one or more storage devices, is determined dependent upon a failure characteristic of the device. A failure characteristic, such as an error rate, may be monitored and updated dynamically.

Fault tolerant memory system

A data processing apparatus includes one or more host processors with first processing units, one or more caches with second processing unit, a non-cache memory having a third processing unit and a reorder buffer operable to maintain data order during execution of a program of instructions. An instruction scheduler routes instructions to the processing units. Data coherence is maintained by control logic that blocks access to data locations in use by a selected processing unit other than the selected processing unit until data associated with the data locations are released from the reorder buffer. Data stored in the cache is written to the memory if it is already in a modified state, otherwise the state is set to the modified state. A memory controller may be used to restrict access to memory locations to be operated on.

Method and apparatus for maintaining data coherence in a non-uniform compute device

Data processing apparatus comprises one or more interconnected processing elements; each processing element being configured to execute processing instructions of program tasks; each processing element being configured to save context data relating to a program task following execution of that program task by that processing element; and to load context data, previously saved by that processing element or another of the processing elements, at resumption of execution of a program task; each processing element having respective associated format definition data to define one or more sets of data items for inclusion in the context data; the apparatus comprising format selection circuitry to communicate the format definition data of each of the processing elements with others of the processing elements and to determine, in response to the format definition data for each of the processing elements, a common set of data items for inclusion in the context data.

Data processing

A method and apparatus are provided for dynamically determining when an operation, specified by one or more instructions in a data processing system, is suitable for accelerated execution. Data indicators are maintained, for data registers of the system, that indicate when data-flow from a register derives from a restricted source. In addition, instruction predicates are provided for instructions to indicate which instructions are capable of accelerated execution. From the data indicators and the instruction predicates, the microarchitecture of the data processing system determines, dynamically, when the operation is a thread-restricted function and suitable for accelerated execution in a hardware accelerator. The thread-restricted function may be executed on a hardware processor, such as a vector, neuromorphic or other processor.

Growing community of inventors

Austin, TX, United States of America

Jonathan Curtis Beard