Multi-cycle key compare units. A compare unit includes a comparator, additional compare logic and at least one pair of buffers which provide input to the comparator. The compare unit sorts variable length records in streaming mode without the need for complex state machines to maintain state relating to the comparing. A record may have a variable length key and optional variable length data. The record and/or the key is split into fixed, pre-defined lengths. The total key and record lengths are unknown to the comparator of the compare unit.

A system can include a memristive crossbar array, which can include row lines and column lines intersecting the row lines. Resistive memory elements can be coupled between the row lines and the column lines at the junctions formed by the row and column lines. The resistive memory elements represent the values of the matrix. The system can further include an analogue circuit. The system can be configured to perform an exponentiation of the values of the vector in accordance with a first exponent. The crossbar array can be configured to apply the resulting values of the vector to the resistive elements thereby generating currents. The analogue circuit can be configured to perform an exponentiation of the generated currents in accordance with a second exponent.

A first record group and a second record group having a tree structure are merge-sorted. The first record group and the second record group are acquired, and depth information indicative of the hierarchical depth of the tree structure is attached as metadata to key information on each node of each record included in the acquired first record group and second record group. The depth information is compared in preference to the key information to perform merge sort sequentially.

A first record group and a second record group having a tree structure are merge-sorted. The first record group and the second record group are acquired, and depth information indicative of the hierarchical depth of the tree structure is attached as metadata to key information on each node of each record included in the acquired first record group and second record group. The depth information is compared in preference to the key information to perform merge sort sequentially.

A computational search method for retrieving computer information related to a query includes transforming a plurality of candidate answers to candidate answer recurrent binary embedding (RBE) embeddings using a trained RBE model. A query is transformed to a query RBE embedding using the trained RBE model. The query RBE embedding is compared to each candidate answer RBE embedding of a plurality of candidate answer RBE embeddings using a similarity function. The candidate answers are sorted based on the comparisons made using the similarity function, and returning a plurality of the top candidate answers.

A computational search method for retrieving computer information related to a query includes transforming a plurality of candidate answers to candidate answer recurrent binary embedding (RBE) embeddings using a trained RBE model. A query is transformed to a query RBE embedding using the trained RBE model. The query RBE embedding is compared to each candidate answer RBE embedding of a plurality of candidate answer RBE embeddings using a similarity function. The candidate answers are sorted based on the comparisons made using the similarity function, and returning a plurality of the top candidate answers.

A security code input may be obfuscated from a thermal imaging device by randomly heating a random set of inputs of an input device. The security code is inputted on an input device, which communicates with a security system to grant or deny access to a user based on an entry of the security code. The input device includes a plurality of hearing elements. The input device may receive an input from the user. A random set of heating elements including one or more heating elements, are generated from the plurality of heating elements. A temperature is determined for the one or more heating elements of the random set of heating elements. The temperature is then applied to the one or more heating elements of the random set of heating elements of the input device.

Disclosed herein are techniques for accelerating convolution operations or other matrix multiplications in applications such as neural network. In one example, an apparatus comprises a first circuit, a second circuit, and a third circuit. The first circuit is configured to: receive first values in a first format, the first values being generated from one or more asymmetric quantization operations of second values in a second format, and generate difference values based on subtracting a third value from each of the first values, the third value representing a zero value in the first format. The second circuit is configured to generate a sum of products in the first format using the difference values. The third circuit is configured to convert the sum of products from the first format to the second format based on scaling the sum of products with a scaling factor.

Certain aspects provide systems and methods for performing an operation on a B.sup.ε-tree. A method comprises writing a message associated with the operation to a first slot in a first buffer of a first non-leaf node of the B.sup.ε-tree in an append-only manner, wherein a first filter associated with the first slot is used for query operations associated with the first slot. The method further comprises determining that the first buffer is full and, upon determining to flush the message to a non-leaf child node, flushing the message in an append-only manner to a second slot in a second buffer of the non-leaf child node, wherein a second filter associated with the second slot is used for query operations associated with the second slot. The method further comprises, upon determining to flush the message to a leaf node, flushing the message to the leaf node in a sorted manner.

Certain aspects provide systems and methods for performing an operation on a B.sup.ε-tree. A method comprises writing a message associated with the operation to a first slot in a first buffer of a first non-leaf node of the B.sup.ε-tree in an append-only manner, wherein a first filter associated with the first slot is used for query operations associated with the first slot. The method further comprises determining that the first buffer is full and, upon determining to flush the message to a non-leaf child node, flushing the message in an append-only manner to a second slot in a second buffer of the non-leaf child node, wherein a second filter associated with the second slot is used for query operations associated with the second slot. The method further comprises, upon determining to flush the message to a leaf node, flushing the message to the leaf node in a sorted manner.