Abstract: A compute-in-memory array is provided that implements a filter for a layer in a neural network. The filter multiplies a plurality of activation bits by a plurality of filter weight bits for each channel in a plurality of channels through a charge accumulation from a plurality of capacitors. The accumulated charge is digitized to provide the output of the filter.

Abstract: A compute-in-memory array is provided that implements a filter for a layer in a neural network. The filter multiplies a plurality of activation bits by a plurality of filter weight bits for each channel in a plurality of channels through a charge accumulation from a plurality of capacitors. The accumulated charge is digitized to provide the output of the filter.

Abstract: Methods and apparatus for performing machine learning tasks, and in particular, to a neural-network-processing architecture and circuits for improved performance through depth parallelism. One example neural-network-processing circuit generally includes a plurality of groups of processing element (PE) circuits, wherein each group of PE circuits comprises a plurality of PE circuits configured to process in parallel an input at a plurality of depths.

Abstract: A dual-mode memory is provided that includes a self-timed clock circuit for asserting a sense enable signal for a sense amplifier. In a low-bandwidth read mode, the self-timed clock circuit asserts the sense enable signal only once during a memory clock cycle. The sense amplifier then senses only a single bit from a group of multiplexed columns. In a high-bandwidth read mode, the self-timed clock circuit successively asserts the sense enable signal so that the sense amplifier successively senses bits from the multiplexed columns.

Abstract: A dual-mode memory is provided that includes a self-timed clock circuit for asserting a sense enable signal for a sense amplifier. In a low-bandwidth read mode, the self-timed clock circuit asserts the sense enable signal only once during a memory clock cycle. The sense amplifier then senses only a single bit from a group of multiplexed columns. In a high-bandwidth read mode, the self-timed clock circuit successively asserts the sense enable signal so that the sense amplifier successively senses bits from the multiplexed columns.

Abstract: A high bandwidth register file circuit that significantly reduces the shared local read bitline RC delay to enable ultra-high performance PRFs with high port counts. In one example, the register file circuit includes read stack nfets in a multiplexer circuit instead of the memory cell causing the local read bitline RC to be independent of the number of read and write ports of the memory cell.

Abstract: A dual-mode memory is provided that includes a self-timed clock circuit for asserting a sense enable signal for a sense amplifier. In a low-bandwidth read mode, the self-timed clock circuit asserts the sense enable signal only once during a memory clock cycle. The sense amplifier then senses only a single bit from a group of multiplexed columns. In a high-bandwidth read mode, the self-timed clock circuit successively asserts the sense enable signal so that the sense amplifier successively senses bits from the multiplexed columns.

Abstract: A dual-mode memory is provided that includes a self-timed clock circuit for asserting a sense enable signal for a sense amplifier. In a low-bandwidth read mode, the self-timed clock circuit asserts the sense enable signal only once during a memory clock cycle. The sense amplifier then senses only a single bit from a group of multiplexed columns. In a high-bandwidth read mode, the self-timed clock circuit successively asserts the sense enable signal so that the sense amplifier successively senses bits from the multiplexed columns.

Abstract: A high bandwidth register file circuit that significantly reduces the shared local read bitline RC delay to enable ultra-high performance PRFs with high port counts. In one example, the register file circuit includes read stack nfets in a multiplexer circuit instead of the memory cell causing the local read bitline RC to be independent of the number of read and write ports of the memory cell.

Abstract: Certain aspects of the present disclosure provide apparatus and methods for performing memory read operations. One example method generally includes precharging a plurality of memory columns during a precharging phase of a read access cycle. The method also includes sensing first data stored in a first memory cell of a first memory column of the plurality of memory columns during a memory read phase of the read access cycle, and sensing second data stored in a second memory cell of a second memory column of the plurality of memory columns during the same memory read phase of the read access cycle.

Abstract: Multi-pump memory system access circuits for sequentially executing parallel memory operations in a memory system are disclosed. A memory system includes a plurality of memory bit cells in a memory array. Each memory bit cell is accessible at a corresponding memory address used by memory read and write operations. The memory system includes ports at which a memory read or a memory write operation is received from a processor in each cycle of a processor clock. To increase memory bandwidth of the memory system without increasing the number of access ports of the memory array within the memory system, a double-pump memory system access circuit double-pumps (i.e., time-multiplexes) the access ports of memory array, effectively doubling the number of ports of the memory array. The double-pump memory system access circuit performs sequential accesses to a port of a memory cell in a memory array within a processor clock period.

Abstract: Certain aspects of the present disclosure provide apparatus and methods for performing memory read operations. One example method generally includes precharging a plurality of memory columns during a precharging phase of a read access cycle. The method also includes sensing first data stored in a first memory cell of a first memory column of the plurality of memory columns during a memory read phase of the read access cycle, and sensing second data stored in a second memory cell of a second memory column of the plurality of memory columns during the same memory read phase of the read access cycle.

Abstract: Read-assist circuits for memory bit cells employing a P-type Field-Effect Transistor (PFET) read port(s) are disclosed. Related memory systems and methods are also disclosed. It has been observed that as node technology is scaled down in size, PFET drive current (i.e., drive strength) exceeds N-type FET (NFET) drive current for like-dimensioned FETs. In this regard, in one aspect, it is desired to provide memory bit cells having PFET read ports, as opposed to NFET read ports, to increase memory read times to the memory bit cells, and thus improve memory read performance. To mitigate or avoid a read disturb condition that could otherwise occur when reading the memory bit cell, read-assist circuits are provided for memory bit cells having PFET read ports.

Abstract: Adaptive power regulation methods and systems are disclosed. In one aspect, one or more process sensors for memory elements are provided, which report information relating to inherent speed characteristics of sub-elements within the memory elements. Based on this reported information, a controller ascertains an appropriate power level to insure a proper data retention voltage (DRV) is applied on voltage rails by a power management unit (PMU) circuit. By using the proper DRV based on the speed characteristics of the sub-elements within the memory elements, power conservation is achieved.

Abstract: Multi-pump memory system access circuits for sequentially executing parallel memory operations in a memory system are disclosed. A memory system includes a plurality of memory bit cells in a memory array. Each memory bit cell is accessible at a corresponding memory address used by memory read and write operations. The memory system includes ports at which a memory read or a memory write operation is received from a processor in each cycle of a processor clock. To increase memory bandwidth of the memory system without increasing the number of access ports of the memory array within the memory system, a double-pump memory system access circuit double-pumps (i.e., time-multiplexes) the access ports of memory array, effectively doubling the number of ports of the memory array. The double-pump memory system access circuit performs sequential accesses to a port of a memory cell in a memory array within a processor clock period.

Abstract: Write-assist circuits for memory bit cells (“bit cells”) employing a P-type Field-Effect transistor (PFET) write port(s) are disclosed. Related methods and systems are also disclosed. It has been observed that as node technology is scaled down in size, PFET drive current (i.e., drive strength) exceeds N-type Field-Effect transistor (NFET) drive current for like-dimensioned FETs. In this regard, in one aspect, it is desired to provide bit cells having PFET write ports, as opposed to NFET write ports, to reduce memory write times to the bit cells, and thus improve memory performance. To mitigate a write contention that could otherwise occur when writing data to bit cells, a write-assist circuit provided in the form of negative wordline boost circuit can be employed to strengthen a PFET access transistor in a memory bit cell having a PFET write port(s).

Abstract: Read-assist circuits for memory bit cells employing a P-type Field-Effect Transistor (PFET) read port(s) are disclosed. Related memory systems and methods are also disclosed. It has been observed that as node technology is scaled down in size, PFET drive current (i.e., drive strength) exceeds N-type FET (NFET) drive current for like-dimensioned FETs. In this regard, in one aspect, it is desired to provide memory bit cells having PFET read ports, as opposed to NFET read ports, to increase memory read times to the memory bit cells, and thus improve memory read performance. To mitigate or avoid a read disturb condition that could otherwise occur when reading the memory bit cell, read-assist circuits are provided for memory bit cells having PFET read ports.

Abstract: Systems and methods for selective refresh of a cache, such as a last-level cache implemented as an embedded DRAM (eDRAM). A refresh bit and a reuse bit are associated with each way of at least one set of the cache. A least recently used (LRU) stack tracks positions of the ways, with positions towards a most recently used position of a threshold comprising more recently used positions and positions towards a least recently used position of the threshold comprise less recently used positions. A line in a way is selectively refreshed if the position of the way is one of the more recently used positions and if the refresh bit associated with the way is set, or the position of the way is one of the less recently used positions and if the refresh bit and the reuse bit associated with the way are both set.

Abstract: P-type Field-effect Transistor (PFET)-based sense amplifiers for reading PFET pass-gate memory bit cells (“bit cells”). Related methods and systems are also disclosed. Sense amplifiers are provided in a memory system to sense bit line voltage(s) of the bit cells for reading the data stored in the bit cells. It has been observed that as node technology is scaled down in size, PFET drive current (i.e., drive strength) exceeds N-type Field-effect Transistor (NFET) drive current due for like-dimensioned FETs. In this regard, in one aspect, PFET-based sense amplifiers are provided in a memory system to increase memory read times to the bit cells, and thus improve memory read performance.

Abstract: Read-assist circuits for memory bit cells employing a P-type Field-Effect Transistor (PFET) read port(s) are disclosed. Related memory systems and methods are also disclosed. It has been observed that as node technology is scaled down in size, PFET drive current (i.e., drive strength) exceeds N-type FET (NFET) drive current for like-dimensioned FETs. In this regard, in one aspect, it is desired to provide memory bit cells having PFET read ports, as opposed to NFET read ports, to increase memory read times to the memory bit cells, and thus improve memory read performance. To mitigate or avoid a read disturb condition that could otherwise occur when reading the memory bit cell, read-assist circuits are provided for memory bit cells having PFET read ports.