Abstract: A memory array may include a read bit line (RBL), a complimentary read bit line (RBLb), a plurality of storage cells each selectably coupled to the RBL and the RBLb such that an XNOR of a read enable (RE) signal and a content of the respective storage cell is output to the RBL in response to the RE signal and an XOR of the RE signal and the content of the respective storage cell is output to the RBLb in response to the RE signal, and a sensing circuit coupled to the RBL and the RBLb and configured to compare a signal on the RBL to a signal on the RBLb and output a comparison result.

Abstract: A system and method for single cycle binary matrix multiplication in neural network computations is disclosed. The system includes a memory array storing binary weights, an input unit for activating rows based on a binary activation vector, and per-column majority sense amplifiers. The system performs binary matrix multiplication in a single cycle, enabling efficient implementation of binary neural networks. The memory array may include sections for weights and inverse weights, with corresponding activation register sections. Differential sense amplifiers may implement the majority function. The system can be applied to convolutional neural networks, using SRAM arrays for image storage and processing. Methods for determining majority votes and counting activated bits using iterative modification of the activation vector are also described.

Abstract: A semiconductor package assembly includes an interposer mounted on a package substrate, a column parallel processor mounted on and electrically connected to the interposer, and a high bandwidth memory (HBM) stack mounted on the parallel processor. The parallel processor includes a memory array with rows and columns, with operations occurring in the columns. Columns of the HBM stack are electrically connected to the columns of the parallel processor. The column parallel processor includes an associative processing unit (APU), a switch fabric for managing data routing, a local SRAM for temporary storage, and a buffer for managing data flow between the HBM stack and processing elements. The assembly is configured to process large language models and perform pattern searches within large datasets stored in the HBM stack.

Abstract: A device and method for facilitating orthogonal data transposition during data transfers to/from a processing array and a storage memory since the data words processed by the processing array (using computational memory cells) are stored orthogonally to how the data words are stored in storage memory. Thus, when data words are transferred between storage memory and the processing array, a mechanism orthogonally transposes the data words.

Abstract: A write data processing method associated with computational memory cells formed as a memory/processing array provides the ability to shift data between adjacent bit lines in each section of the memory/processing array or the same relative bit lines in adjacent sections of the memory/processing array.

Abstract: A read register is provided that captures and stores the read result on a read bit line connected to a set of computational memory cells. The read register may be implemented in the set of computational memory cell to enable the logical XOR, logical AND, and/or logical OR accumulation of read results in the read register. The set of computational memory cells with the read register provides a mechanism for performing complex logical functions across multiple computational memory cells connected to the same read bit line.

Abstract: A read and write data processing apparatus and method associated with computational memory cells formed as a memory/processing array (having a plurality of bit line sections) provides a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections.

Abstract: A processing array that performs one cycle full adder operations. The processing array may have different bit line read/write logic that permits different operations to be performed.

Abstract: A write data processing apparatus and method associated with computational memory cells formed as a memory/processing array provides the ability to shift data between adjacent bit lines in each section of the memory/processing array or the same relative bit lines in adjacent sections of the memory/processing array. The memory/processing array has one or more sections and each section has its own unique set of “n” bit lines.

Abstract: A read and write data processing apparatus and method associated with computational memory cells formed as a memory/processing array (having a plurality of bit line sections) provides a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections.

Abstract: A device and method for facilitating orthogonal data transposition during data transfers to/from a processing array and a storage memory since the data words processed by the processing array (using computational memory cells) are stored orthogonally to how the data words are stored in storage memory. Thus, when data words are transferred between storage memory and the processing array, a mechanism orthogonally transposes the data words.

Abstract: A processing array that performs one cycle full adder operations. The processing array may have different bit line read/write logic that permits different operations to be performed.

Abstract: A processing array that performs one cycle full adder operations. The processing array may have different bit line read/write logic that permits different operations to be performed.

Abstract: A write data processing method associated with computational memory cells formed as a memory/processing array provides the ability to shift data between adjacent bit lines in each section of the memory/processing array or the same relative bit lines in adjacent sections of the memory/processing array.

Abstract: A read and write data processing apparatus and method associated with computational memory cells formed as a memory/processing array (having a plurality of bit line sections) provides a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections.

Abstract: A processing array that performs one cycle full adder operations. The processing array may have different bit line read/write logic that permits different operations to be performed.

Abstract: A device and method for facilitating orthogonal data transposition during data transfers to/from a processing array and a storage memory since the data words processed by the processing array (using computational memory cells) are stored orthogonally to how the data words are stored in storage memory. Thus, when data words are transferred between storage memory and the processing array, a mechanism orthogonally transposes the data words.

Abstract: A write data processing apparatus and method associated with computational memory cells formed as a memory/processing array provides the ability to shift data between adjacent bit lines in each section of the memory/processing array or the same relative bit lines in adjacent sections of the memory/processing array. The memory/processing array has one or more sections and each section has its own unique set of “n” bit lines.

Abstract: A read and write data processing apparatus and method associated with computational memory cells formed as a memory/processing array provides the ability for selected write data in a bit line section to be logically combined (e.g. logically ANDed) with the read result on a read bit line, as if the write data were the read data output of another computational memory cell being read during the read operation. When accumulation logic is implemented in the bit line sections, the implementation and utilization of additional read logic circuitry provides a mechanism for selected write data in a bit line section to be used as the data with which the read result on the read bit line accumulates, before the newly accumulated result is captured and stored in the bit line section's read register.

Abstract: A read register is provided that captures and stores the read result on a read bit line connected to a set of computational memory cells. The read register may be implemented in the set of computational memory cell to enable the logical XOR, logical AND, and/or logical OR accumulation of read results in the read register. The set of computational memory cells with the read register provides a mechanism for performing complex logical functions across multiple computational memory cells connected to the same read bit line.