Via Technologies, Inc.

Via Alliance Semiconductor Co., Ltd.

Ip-first, LLC

I.p.-first, L.l.c.

Integrated Device Technology, Inc.

Shanghai Zhaoxin Semiconductor Co., Ltd.

Other

Centaur Technologies

Jp First LLC

G Glenn Henry

Terry J Parks

Rodney E Hooker

Gerard M Col

Colin Eddy

Dinesh K Jain

Thomas A Crispin

Brent Bean

Stephan Gaskins

Darius D Gaskins

Thomas C McDonald

Timothy A Elliott

Douglas Raye Reed

Kim C Houck

Arturo Martin-de-Nicolas

Kyle T O'Brien

Parviz Palangpour

Albert J Loper, Jr

Jui-Shuan Chen

Jason Chen

James R Lundberg

Matthew Daniel Day

Daniel G Miner

Darius Gaskins

Charles John Holthaus

John D Bunda

Daruis D Gaskins

A method for operating an apparatus that includes a program memory, a data memory and a status register that holds a status, wherein the status has fields including: a program memory address at which a most recent instruction is fetched from the program memory, a data memory access address at which data has most recently been accessed in the data memory by the apparatus and a repeat count indicating a number of times an operation specified in a current program instruction remains to be performed, the apparatus further including a condition register having condition fields corresponding to the status fields held in the status register, the method including: writing the condition register with a condition including the condition fields; and generating an interrupt request to a processing core in response to detecting that the status held in the status register satisfies the condition specified in the condition register.

Neural network unit that interrupts processing core upon condition

A programmable apparatus includes a program memory that holds instructions of a program fetched and executed by the apparatus, a data memory that holds data processed by the instructions, a status register that holds a status having fields: a program memory address at which a most recent instruction is fetched from the program memory, a data memory access address at which data has most recently been accessed in the data memory by the apparatus and a repeat count that indicates a number of times an operation specified in a current program instruction remains to be performed. A condition register has condition fields corresponding to the status register fields. Control logic generates an interrupt request to a processing core in response to detecting that the status held in the status register satisfies the condition specified in the condition register.

An apparatus includes a first memory, processing units that access the first memory, and a counter that, for each period of a sequence of periods, holds an indication of accesses to the first memory during the period; and control logic that, for each period of the sequence of periods, monitors the indication to determine whether it exceeds the threshold and, if so, stalls the processing units from accessing the first memory for a remaining portion of the period.

Neural network unit that manages power consumption based on memory accesses per period

A processor including a memory controller for interfacing an external memory and a programmable functional unit (PFU). The PFU is programmed by a PFU program to modify operation of the memory controller, in which the PFU includes programmable logic elements and programmable interconnectors. For example, the PFU is programmed by the PFU program to add a function or otherwise to modify an existing function of the memory controller enhance its functionality during operation of the processor. In this manner, the functionality and/or operation of the memory controller is not fixed once the processor is manufactured, but instead the memory controller may be modified after manufacture to improve efficiency and/or enhance performance of the processor, such as when executing a corresponding process.

Processor with memory controller including dynamically programmable functional unit

A hardware processing unit is provided. The hardware processing unit includes: an accumulator; a multiplier-adder receives first and second factors and receives an addend, the multiplier-adder generates a sum of the addend and a product of the first and second factors and provides the sum; a first multiplexer receives a first operand, a positive one, and a negative one and selects one of them for provision as the first factor to the multiplier-adder; a second multiplexer receives a second operand, a positive one, and a negative one and selects one of them for provision as the second factor to the multiplier-adder; a third multiplexer, having an output, that receives the first operand and the second operand and selects one of them for provision on its output; and a fourth multiplexer receives the third multiplexer output and the sum and selects one of them for provision to the accumulator.

Neural network unit

A neural network unit includes a register programmable with a control value, a plurality of neural processing units (NPU), and a plurality of activation function units (AFU). Each NPU includes an arithmetic logic unit (ALU) that performs arithmetic and logical operations on a sequence of operands to generate a sequence of results and an accumulator into which the ALU accumulates the sequence of results as an accumulated value. Each AFU includes a first module that performs a first function on the accumulated value to generate a first output, a second module that performs a second function on the accumulated value to generate a second output, the first function is distinct from the second function, and a multiplexer that receives the first and second outputs and selects one of the two outputs based on the control value programmed into the register.

Neural network unit with plurality of selectable output functions

A first data storage holds cache lines, an accelerator has a second data storage that selectively holds accelerator data and cache lines evicted from the first data storage, a tag directory holds tags for cache lines stored in the first and second data storages, and a mode indicator indicates whether the second data storage is operating in a first or second mode in which it respectively holds cache lines evicted from the first data storage or accelerator data. In response to a request to evict a cache line from the first data storage, in the first mode the control logic writes the cache line to the second data storage and updates a tag in the tag directory to indicate the cache line is present in the second data storage, and in the second mode the control logic instead writes the cache line to a system memory.

Processor with selective data storage (of accelerator) operable as either victim cache data storage or accelerator memory and having victim cache tags in lower level cache wherein evicted cache line is stored in said data storage when said data storage is in a first mode and said cache line is stored in system memory rather then said data store when said data storage is in a second mode

A neural network unit has a first memory that holds elements of a data matrix and a second memory that holds elements of a convolution kernel. An array of neural processing units (NPU) each have a multiplexed register that receives a corresponding element of a row from the first memory and that also receives the multiplexed register output of an adjacent NPU. A register receives a corresponding element of a row from the second memory. An arithmetic unit receives the outputs of the register, the multiplexed register and an accumulator and performs a multiply-accumulate operation on them. For each sub-matrix of a plurality of sub-matrices of the data matrix, each arithmetic unit selectively receives either the element from the first memory or the adjacent NPU multiplexed register output and performs a series of the multiply-accumulate operations to accumulate into the accumulator a convolution of the sub-matrix with the convolution kernel.

Neural network unit that performs convolutions using collective shift register among array of neural processing units

A processor comprising a mode indicator, a plurality of processing cores, and a neural network unit (NNU), comprising a memory array, an array of neural processing units (NPU), cache control logic, and selection logic that selectively couples the plurality of NPUs and the cache control logic to the memory array. When the mode indicator indicates a first mode, the selection logic enables the plurality of NPUs to read neural network weights from the memory array to perform computations using the weights. When the mode indicator indicates a second mode, the selection logic enables the plurality of processing cores to access the memory array through the cache control logic as a cache memory.

Processor with memory array operable as either cache memory or neural network unit memory

A processor with an expandable instruction set architecture for dynamically configuring execution resources. The processor includes a programmable execution unit (PEU) that may be programmed to perform a user-defined function in response to a user-defined instruction (UDI). The PEU includes programmable logic elements and programmable interconnectors that are collectively programmed to perform at least one processing operation. A UDI loader is responsive to a UDI load instruction that specifies a UDI and a location of programming information that is used to program the PEU. The PEU may be programmed for one or more UDIs for one or more processes. An instruction table stores each UDI and corresponding information to identify the UDI and possibly to reprogram the PEU if necessary. A UDI handler consults the instruction table to identify a received UDI and to send corresponding information to the PEU to execute the corresponding user-defined function.

Processor with an expandable instruction set architecture for dynamically configuring execution resources

A microprocessor includes a plurality of processing cores and a configuration register configured to indicate whether each of the plurality of processing cores is enabled or disabled. Each enabled one of the plurality of processing cores is configured to read the configuration register in a first instance to determine which of the plurality of processing cores is enabled or disabled and generate a respective configuration-related value based on the read of the configuration register in the first instance. The configuration register is updated to indicate that a previously enabled one of the plurality of processing cores is disabled. Each enabled one of the plurality of processing cores is configured to read the configuration register in a second instance to determine which of the plurality of processing cores is enabled or disabled and generate the respective configuration-related value based on the read of the configuration register in the second instance.

Dynamic reconfiguration of multi-core processor

A memory holds D rows of N words and receives an address having logD bits and an extra bit. Each of N processing units (PU) of index J has first and second registers, an accumulator, an arithmetic unit that performs an operation thereon to accumulate a result, and multiplexing logic receiving memory word J, and for PUs 0 to (N/2)&#x2212;1 also memory word J+(N/2). In a first mode, the multiplexing logic of PUs 0 to N&#x2212;1 selects word J to output to the first register. In a second mode: when the extra bit is a zero, the multiplexing logic of PUs 0 to (N/2)&#x2212;1 selects word J to output to the first register, and when the extra bit is a one, the multiplexing logic of PUs 0 through (N/2)&#x2212;1 selects word J+(N/2) to output to the first register.

Neural network unit with re-shapeable memory

A processor includes a front-end portion that issues instructions to execution units that execute the issued instructions. A hardware neural network unit (NNU) execution unit includes a first memory that holds data words associated with artificial neural networks (ANN) neuron outputs, a second memory that holds weight words associated with connections between ANN neurons, and a third memory that holds a program comprising NNU instructions that are distinct, with respect to their instruction set, from the instructions issued to the NNU by the front-end portion of the processor. The program performs ANN-associated computations on the data and weight words. A first instruction instructs the NNU to transfer NNU instructions of the program from architectural general purpose registers to the third memory. A second instruction instructs the NNU to invoke the program stored in the third memory.

Processor with hybrid coprocessor/execution unit neural network unit

First/second memories hold rows of N weight/data words. Each of N processing units (PU) of index J have a register, an accumulator having an output, an arithmetic unit that performs an operation thereon to accumulate a result, the first input receives the output of the accumulator, the second input receives a respective first memory weight word, the third input receives a respective data word output by the register, and multiplexing logic receives a respective second memory data word and a data word output by the register of PU J&#x2212;1 and outputs a selected data word to the register. PU J&#x2212;1 for PU 0 is PU N&#x2212;1. The multiplexing logic of PU N/4 also receives the data word output by the register of PU (3N/4)&#x2212;1. The multiplexing logic of PU 3N/4 also receives the data word output by the register of PU (N/4)&#x2212;1.

Neural network unit with segmentable array width rotator

First/second memories hold rows of N weight/data words. Each of N processing units (PU) of index J have a register, an accumulator having an output, an arithmetic unit that performs an operation thereon to accumulate a result, the first input receives the output of the accumulator, the second input receives a respective first memory weight word, the third input receives a respective data word output by the register, and multiplexing logic receives a respective second memory data word and a data word output by the register of PU J&#x2212;1 and outputs a selected data word to the register. PU J&#x2212;1 for PU 0 is PU N&#x2212;1. The multiplexing logic of PU 0 also receives the data word output by the register of PU (N/2)&#x2212;1. The multiplexing logic of PU N/2 also receives the data word output by the register of PU N&#x2212;1.

Growing community of inventors

Austin, TX, United States of America

G Glenn Henry