摘要:
The present invention provides a technique for synchronisation between pipelines in a data processing apparatus. The data processing apparatus comprises a main processor operable to execute a sequence of instructions, the main processor comprising a first pipeline having a first plurality of pipeline stages, and a coprocessor operable to execute coprocessor instructions in said sequence of instructions. The coprocessor comprises a second pipeline having a second plurality of pipeline stages, and each coprocessor instruction is arranged to be routed through both the first pipeline and the second pipeline. Furthermore, at least one synchronising queue is provided coupling a predetermined pipeline stage in one of the pipelines with a partner pipeline stage in the other of the pipelines, the predetermined pipeline stage being operable to cause a token to be placed in the synchronising queue when processing a coprocessor instruction, and the partner pipeline stage being operable to process that coprocessor instruction upon receipt of the token from the synchronising queue. By this approach, the first and second pipelines are synchronised between the predetermined pipeline stage and the partner pipeline stage, and hence ensures that the pipelines are correctly synchronised for crucial transfers of information without requiring that strict synchronisation at all stages is necessary.
摘要:
A wireless device for processing streams of media data in a wireless bi-directional communications network. The bi-directional communications network is capable of transmitting and receiving media data streams which comprise a combination of at least two of audio, video, radio, graphics, encryption, authentication, and networking information. The wireless device has at least one programmable media processor (12) for receiving, processing and transmitting the stream of media data over the bi-directional communications network. The processor executes group instructions to read a plurality of data elements of the media data stream from a register file (110), to perform, on the data elements, group operations including both group integer and group floating point operations capable of dynamically partitioning the data by each specifying one of a plurality of data element sizes, and to write concatenated results in the register file.
摘要:
A digital baseband processor is provided for concurrent operation with different wireless systems. The digital baseband processor includes a digital signal processor for executing digital signal processor instructions, a microcontroller for executing microcontroller instructions, and a timing and event processor controlled by the digital signal processor and the microcontroller for executing timing-sensitive instructions. The timing and event processor includes a plurality of instruction sequencers for executing timing-sensitive instruction threads and a time base generator for generating timing signals for initiating execution of the instruction threads on each of the plurality of instruction sequencers.
摘要:
A general purpose, programmable media processor (12) for processing and transmitting a media data streams. The media processor (12) incorporates an execution unit (100) that maintains substantially peak data through out of media data streams. The execution unit (100) includes a dynamically partionable multi-precision arithmetic unit (102), programmable switch (104) and programmable extended mathematical element (106). A high bandwidth external interface (124) supplies media data streams at substantially peak rates to a general purpose register file (110) and the execution unit. A memory management unit, and instruction and data cache/buffers (118, 120) are provided. The general purpose, programmable media processor (12) is disposed in a network fabric consisting of fiber optic cable, coaxial cable and twisted pair wires to transmit, process and receive single or unified media data streams.
摘要:
Automatic selective power and energy control of one or more processing elements matches a degree of parallelism to a monitored condition, in a highly parallel programmable data processor. For example, logic of the parallel processor detects when program operations (e.g. for a particular task or due to a detected temperature) require less than the full width of the data path. In response, the control logic automatically sets a mode of operation requiring a subset of the parallel processing capacity. At least one parallel processing element, that is not needed, can be shut down, to conserve energy and/or to reduce heating (i.e., power consumption). At a later time, when operation of the added capacity is appropriate, the logic detects the change in processing conditions and automatically sets the mode of operation to that of the wider data path, typically the full width. The mode change reactivates the previously shut-down processing element.
摘要:
The present invention relates to a data pipeline management system and more particularly to a minimum memory solution for unidirectional data pipeline management in a situation where both the Producer and Consumer need asynchronous access to the pipeline, data is non-atomic, and only the last complete (and validated) received message is relevant and once a data read from/write to the pipeline is initiated, that data must be completely processed. The data pipeline management system according to the invention can be implemented as a circular queue of as little as three entries and an additional handshake mechanism, implemented as a set of indices that can fit in a minimum of six bits (2 x 2 + 2 x 1). Both the Producer and Consumer will have a 2 bit index indicating where they are in the queue, and a 1 bit binary value indicating a special situation. Both parties can read all the indices but can only write their own, i.e. P and wrapP for the Producer and C and wrapC for the Consumer. For management of the handshakes a set of rules is provided.
摘要:
The present invention provides apparatus and methods to perform thermal management in a computing environment. In one embodiment, thermal attributes are associated with operations and/or processing components, and the operations are scheduled for processing by the components so that a thermal threshold is not exceeded. In another embodiment, hot and cool queues are provided for selected operations, and the processing components can select operations from the appropriate queue so that the thermal threshold is not exceeded.
摘要:
The invention relates to a method and device, for operand processing in a processor unit, comprising at least two execution units, operated with a given cycle, whereby the execution units are controlled by means of control signals, for the processing of the operands and which can be switched between a first operating mode and a second operating mode, characterised in that, in the first operating mode, both execution units are supplied with the same operands and, in the second operating mode, both execution units are supplied with differing operands and, in the first operating mode, both execution units are controlled with the same control signal for processing the operands and, in the second operating mode, both execution units are provided with differing control signals for processing the operands.
摘要:
A reconfigurable processor (100) calculates execution times of configuration for executing pipeline processing from hardware configuration information, and fixes a clock cycle until processing ends. A counter (108) compares the fixed clock cycle with the actual number of elapsed clocks, and, when the number of elapsed clocks equals the clock cycle, it is determined that pipeline processing has ended, and a configuration controller (106) is notified of this.
摘要:
A general purpose, programmable media processor (12) for processing and transmitting a media data streams. The media processor (12) incorporates an execution unit (100) that maintains substantially peak data through out of media data streams. The execution unit (100) includes a dynamically partionable multi-precision arithmetic unit (102), programmable switch (104) and programmable extended mathematical element (106). A high bandwidth external interface (124) supplies media data streams at substantially peak rates to a general purpose register file (110) and the execution unit. A memory management unit, and instruction and data cache/buffers (118, 120) are provided. The general purpose, programmable media processor (12) is disposed in a network fabric consisting of fiber optic cable, coaxial cable and twisted pair wires to transmit, process and receive single or unified media data streams.