A system and method to replace a defective tile of an array of tiles on a die is disclosed. Each tile in includes a plurality of processing cores. Each of the tiles are in communication with each other through a network. A defective tile is detected and a spare tile external to the tiles is selected. Packets sent on the network to the defective tile are directed to the spare tile. The function of the defective tile is assigned to the spare tile.
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 13/14 - Gestion de demandes d'interconnexion ou de transfert
G06F 13/38 - Transfert d'informations, p. ex. sur un bus
A system and method to replace a defective tile of an array of tiles on a die is disclosed. Each tile in includes a plurality of processing cores. Each of the tiles are in communication with each other through a network. A defective tile is detected and a spare tile external to the tiles is selected. Packets sent on the network to the defective tile are directed to the spare tile. The function of the defective tile is assigned to the spare tile.
G06F 11/20 - Détection ou correction d'erreur dans une donnée par redondance dans le matériel en utilisant un masquage actif du défaut, p. ex. en déconnectant les éléments défaillants ou en insérant des éléments de rechange
3.
METHOD AND SYSTEM FOR PROVIDING EFFICIENT QUANTUM PROOF ENCRYPTION USING BOOLEAN GATES WITH REDUCED BOOTSTRAPPING
A system and method to reduce the number of bootstraps required for performing Boolean operations on ciphertexts using Fully Homomorphic Encryption over the Torus (tFHE). A modulo is determined for a plurality of encrypted ciphertexts. False is assigned to a first value and true is assigned to a second value. A Boolean expression including the plurality of ciphertexts is converted into a mathematical function and each of the possible outputs of the function is mapped to the first value or the second value. A first outcome value is assigned to the result of the mathematical function when the Boolean expression evaluates to true. A second outcome value is assigned to each of a set of results of the mathematical function when the Boolean expression evaluates to false. The number of bootstraps required is reduced based on only having to bootstrap the first and second outcome values.
The disclosure is directed toward a method and system to efficiently determine an attention function by reducing bootstrapping steps and ciphertext matrix operations. A combined key and query matrix in plaintext is pre-calculated. A ciphertext-plaintext matrix multiplication of the combined key and query matrix with a ciphertext input is performed. The output of the resulting ciphertext-plaintext matrix multiplication of the combined key and query matrix is bootstrapped. A ciphertext-plaintext matrix multiplication of the value matrix with the ciphertext input is performed. The resulting ciphertext-plaintext matrix multiplication of the value matrix is bootstrapped.
H04L 9/06 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité l'appareil de chiffrement utilisant des registres à décalage ou des mémoires pour le codage par blocs, p. ex. système DES
H04L 9/00 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité
5.
SYSTEM AND METHOD FOR ESTABLISHING HIGH SPEED CONNECTIONS FOR DEVICES IN A HOST BASED ARCHITECTURE
A system and method to configure a high-speed interconnection network between devices coupled to a host is disclosed. A bus having a plurality of lanes is coupled to the host. Each of the devices are coupled to one or more of the lanes of the bus allowing communication to the host. Each of the devices is coupled to two neighboring devices via cables coupled to high-speed input output ports to form a ring interconnection between the devices. The devices are identified by designating a first device and sending packets to the other devices to identify the other devices to allow high speed data traffic to be sent between the devices on the ring interconnection.
The disclosure is directed toward a method to efficiently perform a large language model operation by reducing the memory and computation resources for operating on a matrix such as a weight matrix. The weight matrix may be included in operations such as the head attention function of the large language model. The weight matrix is decomposed into multiple submatrices that each include a set of weights from the weight matrix. An input of the large language model is provided. A submatrix solution for the submatrix is determined from the input and the set of weights of the submatrix. The determining of a solution is repeated for each of the submatrices to provide multiple submatrix solutions. The resulting submatrix solutions are combined to obtain a matrix solution for the weight matrix.
Systems and methods for configuring a reduced instruction set computer processor architecture to execute fully homomorphic encryption (FHE) logic gates as a streaming topology. The method includes parsing sequential FHE logic gate code, transforming the FHE logic gate code into a set of code modules that each have in input and an output that is a function of the input and which do not pass control to other functions, creating a node wrapper around each code module, configuring at least one of the primary processing cores to implement the logic element equivalents of each element in a manner which operates in a streaming mode wherein data streams out of corresponding arithmetic logic units into the main memory and other ones of the plurality arithmetic logic units.
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 7/48 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
9.
METHOD AND SYSTEM FOR REDUCING FFT CALCULATIONS IN FHE BOOTSTRAPPING
A system and method to reduce fast Fourier transforms (FFT) required for bootstrapping in a Fully Homomorphic Encryption process. Ciphertext is separated into a vector of n samples. A fast Fourier transfer (FFT) is performed over a first vector of the samples. An FFT is performed for each of n polynomial terms multiplied by a bootstrap key. A point wise multiplication of each of the FFT outputs of the FFTs of the polynomial terms and the output of the FFT over the vector of the n samples is performed. The result of the FFT over the vector of the n samples is added to the results of the set of pointwise multiplications. An inverse FFT (IFFT) is performed on the FFT over the vector of n samples and the accumulated results of the point-wise multiplications to obtain a bootstrapping result of the ciphertext.
H04L 9/06 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité l'appareil de chiffrement utilisant des registres à décalage ou des mémoires pour le codage par blocs, p. ex. système DES
G06F 17/14 - Transformations de Fourier, de Walsh ou transformations d'espace analogues
H04L 9/00 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité
10.
METHOD AND SYSTEM FOR REDUCING FFT CALCULATIONS IN FHE BOOTSTRAPPING
A system and method to reduce fast Fourier transforms (FFT) required for bootstrapping in a Fully Homomorphic Encryption process. Ciphertext is separated into a vector of n samples. A fast Fourier transfer (FFT) is performed over a first vector of the samples. An FFT is performed for each of n polynomial terms multiplied by a bootstrap key. A point wise multiplication of each of the FFT outputs of the FFTs of the polynomial terms and the output of the FFT over the vector of the n samples is performed. The result of the FFT over the vector of the n samples is added to the results of the set of pointwise multiplications. An inverse FFT (IFFT) is performed on the FFT over the vector of n samples and the accumulated results of the point-wise multiplications to obtain a bootstrapping result of the ciphertext.
A system to output a response to a query via processing a large language model is disclosed. The system includes an array of processing cores arranged in a grid allowing each processing core to communicate directly to a neighboring processing core. An interconnection network is coupled to each of the processing cores allowing communication between the processing cores. A first processing core of the array of processing cores is configured to receive an encrypted query. A second processing core of the array of processing cores is configured to input the encrypted query to a large language model; execute the large language model having general weights in plaintext; and provide an encrypted output of the large language model.
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 21/57 - Certification ou préservation de plates-formes informatiques fiables, p. ex. démarrages ou arrêts sécurisés, suivis de version, contrôles de logiciel système, mises à jour sécurisées ou évaluation de vulnérabilité
G06N 3/10 - Interfaces, langages de programmation ou boîtes à outils de développement logiciel, p. ex. pour la simulation de réseaux neuronaux
12.
METHOD, APPARATUS, AND COMPUTER-READABLE MEDIUM FOR PARALLELIZATION OF A COMPUTER PROGRAM ON A PLURALITY OF COMPUTING CORES
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
A method and system to provide secure fine-tune training and inferences for large language models using a Fully Homomorphic Encryption (FHE) process is disclosed. A trained public large language model with a matrix of public weights is stored in plaintext. A private training set of proprietary data is stored as a set of encrypted weights. An encrypted query is received. The encrypted query is processed via an encrypted data path with the set of encrypted weights for trained large language model. The encrypted query is processed via a plaintext data path with the set of public weights in plaintext for the trained large language model. A low rank adaptation (LoRA) approach may be used in fine-tuning the model to reduce computational complexity and memory access. Encrypted results are obtained from the processing via the encrypted data path and the plaintext weight path.
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
15.
METHOD AND SYSTEM FOR COMPRESSING APPLICATION DATA FOR OPERATIONS ON MULTI-CORE SYSTEMS
A system and method to compress application control data, such as weights for a layer of a convolutional neural network, is disclosed. A multi-core system for executing at least one layer of the convolutional neural network includes a storage device storing a compressed weight matrix of a set of weights of the at least one layer of the convolutional network and a decompression matrix. The compressed weight matrix is formed by matrix factorization and quantization of a floating point value of each weight to a floating point format. A decompression module is operable to obtain an approximation of the weight values by decompressing the compressed weight matrix through the decompression matrix. A plurality of cores executes the at least one layer of the convolutional neural network with the approximation of weight values to produce an inference output.
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
G06F 7/483 - Calculs avec des nombres représentés par une combinaison non linéaire de nombres codés, p. ex. nombres rationnels, système de numération logarithmique ou nombres à virgule flottante
G06F 18/2135 - Extraction de caractéristiques, p. ex. en transformant l'espace des caractéristiquesSynthétisationsMappages, p. ex. procédés de sous-espace basée sur des critères d'approximation, p. ex. analyse en composantes principales
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/77 - Traitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
H03M 7/02 - Conversion en, ou à partir de codes pondérés, c.-à-d. le poids donné à un chiffre dépendant de sa position dans le bloc ou dans le mot-code
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
16.
CACHE ARCHITECTURE FOR A MASSIVELY PARALLEL PROCESSING ARRAY
A cache architecture for an array of identical cores arranged in a grid. Each of the cores include interconnections to neighboring cores in the grid, a memory, and an algorithmic logic unit. A first core of the array is configured to receive a memory access request for data from at least one core of the array of cores configured to perform a computational operation. A second core of the array is configured to determine whether the requested data is present in a cache memory via a cache index including addresses in the cache memory. A third core of the array is configured as the cache memory. The memory of the third core is used as the cache memory. An address of the requested data from the cache index is passed to the third core to output the requested data.
A system and method to allocate serial interconnection lanes on a die to multiple communication protocols is disclosed. The die has at least one processing core. The die incudes a first communication subsystem including a controller, a protocol coding sublayer (PCS) for interchanging data, and a data interface coupled to the core. The die includes a second communication subsystem including a controller, a PCS for interchanging data, and a data interface coupled to the core. A mode input selects at least one of the first or second communication protocol. A data router has an input coupled to the PCS of the first communication subsystem and an input coupled to the PCS of the second communication subsystem. The data router has an output coupled to the set of serial interconnection lanes, and a selection input coupled to the mode input to allocate some of the lanes for the selected protocol.
A cache architecture for an array of identical cores arranged in a grid. Each of the cores include interconnections to neighboring cores in the grid, a memory, and an algorithmic logic unit. A first core of the array is configured to receive a memory access request for data from at least one core of the array of cores configured to perform a computational operation. A second core of the array is configured to determine whether the requested data is present in a cache memory via a cache index including addresses in the cache memory. A third core of the array is configured as the cache memory. The memory of the third core is used as the cache memory. An address of the requested data from the cache index is passed to the third core to output the requested data.
A cache architecture for an array of identical cores arranged in a grid. Each of the cores include interconnections to neighboring cores in the grid, a memory, and an algorithmic logic unit. A first core of the array is configured to receive a memory access request for data from at least one core of the array of cores configured to perform a computational operation. A second core of the array is configured to determine whether the requested data is present in a cache memory via a cache index including addresses in the cache memory. A third core of the array is configured as the cache memory. The memory of the third core is used as the cache memory. An address of the requested data from the cache index is passed to the third core to output the requested data.
A method and system for providing robust streaming of data from a multi-core die is disclosed. The techniques include using a high bandwidth memory (HBM) device as retransmit buffers for large amounts of data to ensure robust communication in relatively high round trip-transmission time (RTT) transmission. Another technique is supporting two or more Ethernet ports between components to both transmit the same data packets on the two ports to insure robustness. Another technique is to use sequence numbers and send data packets from the different ports in a round robin fashion and reorder the packets upon receipt of an external device. Another technique is dynamically adding and removing paths for data packets between devices with multiple ports based on the quality of the path.
A system and method to create a robust topology of a layout of cores for performing a function on an array of cores arranged in a grid is disclosed. A defective core file of location of defective cores in the array and an optimal ideal topology of a configuration layout of at least some of the cores is input. The location of at least one defective core of the array is determined. At least some of the cores in the array of cores are assigned to the optimal initial topography of cores in the array. It is determined whether at least one defective core is in the optimal initial topography. The functions of the cores in the row and the column of the at least one defective core are assigned to additional neighboring cores in the array of cores to create the robust topology.
A system and method to create a robust topology of a layout of cores for performing a function on an array of cores arranged in a grid is disclosed. A defective core file of location of defective cores in the array and an optimal ideal topology of a configuration layout of at least some of the cores is input. The location of at least one defective core of the array is determined. At least some of the cores in the array of cores are assigned to the optimal initial topography of cores in the array. It is determined whether at least one defective core is in the optimal initial topography. The functions of the cores in the row and the column of the at least one defective core are assigned to additional neighboring cores in the array of cores to create the robust topology.
G06F 11/20 - Détection ou correction d'erreur dans une donnée par redondance dans le matériel en utilisant un masquage actif du défaut, p. ex. en déconnectant les éléments défaillants ou en insérant des éléments de rechange
G06F 11/07 - Réaction à l'apparition d'un défaut, p. ex. tolérance de certains défauts
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
25.
Method, apparatus, and computer-readable medium for parallelization of a computer program on a plurality of computing cores
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
A system and method to efficiently configure an array of processing cores to perform functions of a program. A function of the program is converted to a configuration of cores. The configuration is laid out in a first subset of the array of cores. The configuration is stored. The configuration is replicated to perform the function on a second subset of the array of cores.
A system and method to efficiently configure an array of processing cores to perform functions of a program. A function of the program is converted to a configuration of cores. The configuration is laid out in a first subset of the array of cores. The configuration is stored. The configuration is replicated to perform the function on a second subset of the array of cores.
A system and method to allocate serial interconnection lanes on a die to multiple communication protocols is disclosed. The die has at least one processing core. The die incudes a first communication subsystem including a controller, a protocol coding sublayer (PCS) for interchanging data, and a data interface coupled to the core. The die includes a second communication subsystem including a controller, a PCS for interchanging data, and a data interface coupled to the core. A mode input selects at least one of the first or second communication protocol. A data router has an input coupled to the PCS of the first communication subsystem and an input coupled to the PCS of the second communication subsystem. The data router has an output coupled to the set of serial interconnection lanes, and a selection input coupled to the mode input to allocate some of the lanes for the selected protocol.
Systems and methods for configuring a reduced instruction set computer processor architecture to execute fully homomorphic encryption (FHE) logic gates as a streaming topology. The method includes parsing sequential FHE logic gate code, transforming the FHE logic gate code into a set of code modules that each have in input and an output that is a function of the input and which do not pass control to other functions, creating a node wrapper around each code module, configuring at least one of the primary processing cores to implement the logic element equivalents of each element in a manner which operates in a streaming mode wherein data streams out of corresponding arithmetic logic units into the main memory and other ones of the plurality arithmetic logic units.
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 7/48 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés
G06F 17/14 - Transformations de Fourier, de Walsh ou transformations d'espace analogues
A system and method to compress application control data, such as weights for a layer of a convolutional neural network, is disclosed. A multi-core system for executing at least one layer of the convolutional neural network includes a storage device storing a compressed weight matrix of a set of weights of the at least one layer of the convolutional network and a decompression matrix. The compressed weight matrix is formed by matrix factorization and quantization of a floating point value of each weight to a floating point format. A decompression module is operable to obtain an approximation of the weight values by decompressing the compressed weight matrix through the decompression matrix. A plurality of cores executes the at least one layer of the convolutional neural network with the approximation of weight values to produce an inference output.
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
G06F 7/483 - Calculs avec des nombres représentés par une combinaison non linéaire de nombres codés, p. ex. nombres rationnels, système de numération logarithmique ou nombres à virgule flottante
G06F 18/2135 - Extraction de caractéristiques, p. ex. en transformant l'espace des caractéristiquesSynthétisationsMappages, p. ex. procédés de sous-espace basée sur des critères d'approximation, p. ex. analyse en composantes principales
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/77 - Traitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
H03M 7/02 - Conversion en, ou à partir de codes pondérés, c.-à-d. le poids donné à un chiffre dépendant de sa position dans le bloc ou dans le mot-code
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 21/44 - Authentification de programme ou de dispositif
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
A method and system for providing robust streaming of data from a multi-core die is disclosed. The techniques include using a high bandwidth memory (HBM) device as retransmit buffers for large amounts of data to ensure robust communication in relatively high round trip-transmission time (RTT) transmission. Another technique is supporting two or more Ethernet ports between components to both transmit the same data packets on the two ports to insure robustness. Another technique is to use sequence numbers and send data packets from the different ports in a round robin fashion and reorder the packets upon receipt of an external device. Another technique is dynamically adding and removing paths for data packets between devices with multiple ports based on the quality of the path.
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 7/76 - Dispositions pour le réagencement, la permutation ou la sélection de données selon des règles prédéterminées, indépendamment du contenu des données
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
34.
Method and apparatus for configuring a reduced instruction set computer processor architecture to execute a fully homomorphic encryption algorithm
Systems and methods for configuring a reduced instruction set computer processor architecture to execute fully homomorphic encryption (FHE) logic gates as a streaming topology. The method includes parsing sequential FHE logic gate code, transforming the FHE logic gate code into a set of code modules that each have in input and an output that is a function of the input and which do not pass control to other functions, creating a node wrapper around each code module, configuring at least one of the primary processing cores to implement the logic element equivalents of each element in a manner which operates in a streaming mode wherein data streams out of corresponding arithmetic logic units into the main memory and other ones of the plurality arithmetic logic units.
G06F 7/48 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés
Systems and methods for reconfiguring a reduced instruction set computer processor architecture are disclosed. Exemplary implementations may: provide a primary processing core consisting of a RISC processor; provide a node wrapper associated with each of the plurality of secondary cores, the node wrapper comprising access memory associates with each secondary core, and a load/unload matrix associated with each secondary core; operate the architecture in a manner in which, for at least one core, data is read from and written to the at least cache memory in a control-centric mode; the secondary cores are selectively partitioned to operate in a streaming mode wherein data streams out of the corresponding secondary core into the main memory and other ones of the plurality of secondary cores.
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
37.
Parallel processing of data having data dependencies for accelerating the launch and performance of operating systems and other computing applications
Representative embodiments are disclosed for a rapid and highly parallel decompression of compressed executable and other files, such as executable files for operating systems and applications, having compressed blocks including run length encoded (“RLE”) data having data-dependent references. An exemplary embodiment includes a plurality of processors or processor cores to identify a start or end of each compressed block; to partially decompress, in parallel, a selected compressed block into independent data, dependent (RLE) data, and linked dependent (RLE) data; to sequence the independent data, dependent (RLE) data, and linked dependent (RLE) data from a plurality of partial decompressions of a plurality of compressed blocks, to obtain data specified by the dependent (RLE) data and linked dependent (RLE) data, and to insert the obtained data into a corresponding location in an uncompressed file. The representative embodiments are also applicable to other types of data processing for applications having data dependencies.
H03M 7/46 - Conversion en, ou à partir de codes à longueur de série, c.-à-d. par représentation du nombre de chiffres successifs ou groupes de chiffres de même type à l'aide d'un mot-code et d'un chiffre représentant ce type
G06F 3/06 - Entrée numérique à partir de, ou sortie numérique vers des supports d'enregistrement
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
H03M 7/42 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse utilisant une table pour le procédé de codage ou de décodage, p. ex. utilisant une mémoire morte
A method and system for providing robust streaming of data from a multi-core die is disclosed. The techniques include using a high bandwidth memory (HBM) device as retransmit buffers for large amounts of data to ensure robust communication in relatively high round trip-transmission time (RTT) transmission. Another technique is supporting two or more Ethernet ports between components to both transmit the same data packets on the two ports to insure robustness. Another technique is to use sequence numbers and send data packets from the different ports in a round robin fashion and reorder the packets upon receipt of an external device. Another technique is dynamically adding and removing paths for data packets between devices with multiple ports based on the quality of the path.
A system and method to compress application control data, such as weights for a layer of a convolutional neural network, is disclosed. A multi-core system for executing at least one layer of the convolutional neural network includes a storage device storing a compressed weight matrix of a set of weights of the at least one layer of the convolutional network and a decompression matrix. The compressed weight matrix is formed by matrix factorization and quantization of a floating point value of each weight to a floating point format. A decompression module is operable to obtain an approximation of the weight values by decompressing the compressed weight matrix through the decompression matrix. A plurality of cores executes the at least one layer of the convolutional neural network with the approximation of weight values to produce an inference output.
G06F 7/483 - Calculs avec des nombres représentés par une combinaison non linéaire de nombres codés, p. ex. nombres rationnels, système de numération logarithmique ou nombres à virgule flottante
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
G06K 9/62 - Méthodes ou dispositions pour la reconnaissance utilisant des moyens électroniques
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 7/57 - Unités arithmétiques et logiques [UAL], c.-à-d. dispositions ou dispositifs pour accomplir plusieurs des opérations couvertes par les groupes ou pour accomplir des opérations logiques
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 7/57 - Unités arithmétiques et logiques [UAL], c.-à-d. dispositions ou dispositifs pour accomplir plusieurs des opérations couvertes par les groupes ou pour accomplir des opérations logiques
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
A representative reconfigurable processing circuit and a reconfigurable arithmetic circuit are disclosed, each of which may include input reordering queues; a multiplier shifter and combiner network coupled to the input reordering queues; an accumulator circuit; and a control logic circuit, along with a processor and various interconnection networks. A representative reconfigurable arithmetic circuit has a plurality of operating modes, such as floating point and integer arithmetic modes, logical manipulation modes, Boolean logic, shift, rotate, conditional operations, and format conversion, and is configurable for a wide variety of multiplication modes. Dedicated routing connecting multiplier adder trees allows multiple reconfigurable arithmetic circuits to be reconfigurably combined, in pair or quad configurations, for larger adders, complex multiplies and general sum of products use, for example.
G06F 21/44 - Authentification de programme ou de dispositif
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
H03K 19/21 - Circuits OU EXCLUSIF, c.-à-d. donnant un signal de sortie si un signal n'existe qu'à une seule entréeCircuits à COÏNCIDENCES, c.-à-d. ne donnant un signal de sortie que si tous les signaux d'entrée sont identiques
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 5/01 - Procédés ou dispositions pour la conversion de données, sans modification de l'ordre ou du contenu des données maniées pour le décalage, p. ex. la justification, le changement d'échelle, la normalisation
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 7/544 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés pour l'évaluation de fonctions par calcul
44.
CONFIGURING A REDUCED INSTRUCTION SET COMPUTER PROCESSOR ARCHITECTURE TO EXECUTE A FULLY HOMOMORPHIC ENCRYPTION ALGORITHM
Systems and methods for configuring a reduced instruction set computer processor architecture to execute fully homomorphic encryption (FHE) logic gates as a streaming topology. The method includes parsing sequential FHE logic gate code, transforming the FHE logic gate code into a set of code modules that each have in input and an output that is a function of the input and which do not pass control to other functions, creating a node wrapper around each code module, configuring at least one of the primary processing cores to implement the logic element equivalents of each element in a manner which operates in a streaming mode wherein data streams out of corresponding arithmetic logic units into the main memory and other ones of the plurality arithmetic logic units.
Systems and methods for configuring a reduced instruction set computer processor architecture to execute fully homomorphic encryption (FHE) logic gates as a streaming topology. The method includes parsing sequential FHE logic gate code, transforming the FHE logic gate code into a set of code modules that each have in input and an output that is a function of the input and which do not pass control to other functions, creating a node wrapper around each code module, configuring at least one of the primary processing cores to implement the logic element equivalents of each element in a manner which operates in a streaming mode wherein data streams out of corresponding arithmetic logic units into the main memory and other ones of the plurality arithmetic logic units.
G06F 7/48 - Méthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs n'établissant pas de contact, p. ex. tube, dispositif à l'état solideMéthodes ou dispositions pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. en utilisant une représentation binaire, ternaire, décimale utilisant des dispositifs non spécifiés
G06F 17/14 - Transformations de Fourier, de Walsh ou transformations d'espace analogues
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
Systems and methods for reconfiguring a reduced instruction set computer processor architecture are disclosed. Exemplary implementations may: provide a primary processing core consisting of a RISC processor; provide a node wrapper associated with each of the plurality of secondary cores, the node wrapper comprising access memory associates with each secondary core, and a load/unload matrix associated with each secondary core; operate the architecture in a manner in which, for at least one core, data is read from and written to the at least cache memory in a control-centric mode; the secondary cores are selectively partitioned to operate in a streaming mode wherein data streams out of the corresponding secondary core into the main memory and other ones of the plurality of secondary cores.
G06F 9/30 - Dispositions pour exécuter des instructions machines, p. ex. décodage d'instructions
G06F 15/78 - Architectures de calculateurs universels à programmes enregistrés comprenant une seule unité centrale
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 7/57 - Unités arithmétiques et logiques [UAL], c.-à-d. dispositions ou dispositifs pour accomplir plusieurs des opérations couvertes par les groupes ou pour accomplir des opérations logiques
09 - Appareils et instruments scientifiques et électriques
Produits et services
Integrated circuits, digital signal processors and data processors having a processor core configured into a plurality of computational units and state machines for use in instruction processing and data flow operations
09 - Appareils et instruments scientifiques et électriques
Produits et services
Computer hardware, downloadable software, downloadable software programming models and programming language, and downloadable software development kits (SDKs), all for programming and configuring integrated circuits and processors for instruction processing and data flow operations, configuration of processor cores into a plurality of computational units for instruction processing and data flow operations, for use in the fields of artificial intelligence, machine learning, neural networks, deep learning, deep neural networks, database processing, edge processing, Internet of Things, security, biometrics, and predictive technology
42 - Services scientifiques, technologiques et industriels, recherche et conception
Produits et services
Design and development of computer hardware and software; Product research, development, design, and engineering of integrated circuits, processors, semiconductor processors, electronic circuit boards, rack-mountable chassis having processors, software-configurable processors, software development kits (SDKs), cloud computing software, software platform for software development featuring software development kits (SDKs), software and software-programmable models all for programming and configuring integrated circuits and processors for instruction processing and data flow operations for use in the fields of artificial intelligence, machine learning, neural networks, deep learning, deep neural networks, database processing, edge processing, Internet of Things, security, biometrics, and predictive technology; Design, development, customization, integration, technical support, namely, monitoring of technological functions of computer systems, and maintenance of computer software; Providing temporary use of on-line non-downloadable software and on-line non-downloadable software programs all for use in programming and configuring integrated circuits and processors for instruction processing and data flow operations for use in the fields of artificial intelligence, machine learning, neural networks, deep learning, deep neural networks, database processing, edge processing, Internet of Things, security, biometrics, and predictive technology
50.
Method and apparatus for a multi-core system for implementing stream-based computations having inputs from multiple streams
A method and system of efficient use and programming of a multi-processing core device. The system includes a programming construct that is based on stream-domain code. A programmable core based computing device is disclosed. The computing device includes a plurality of processing cores coupled to each other. A memory stores stream-domain code including a stream defining a stream destination module and a stream source module. The stream source module places data values in the stream and the stream conveys data values from the stream source module to the stream destination module. A runtime system detects when the data values are available to the stream destination module and schedules the stream destination module for execution on one of the plurality of processing cores.
G06F 9/46 - Dispositions pour la multiprogrammation
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
51.
Parallel processing of data having data dependencies for accelerating the launch and performance of operating systems and other computing applications
Representative embodiments are disclosed for a rapid and highly parallel decompression of compressed executable and other files, such as executable files for operating systems and applications, having compressed blocks including run length encoded (“RLE”) data having data-dependent references. An exemplary embodiment includes a plurality of processors or processor cores to identify a start or end of each compressed block; to partially decompress, in parallel, a selected compressed block into independent data, dependent (RLE) data, and linked dependent (RLE) data; to sequence the independent data, dependent (RLE) data, and linked dependent (RLE) data from a plurality of partial decompressions of a plurality of compressed blocks, to obtain data specified by the dependent (RLE) data and linked dependent (RLE) data, and to insert the obtained data into a corresponding location in an uncompressed file. The representative embodiments are also applicable to other types of data processing for applications having data dependencies.
H03M 7/46 - Conversion en, ou à partir de codes à longueur de série, c.-à-d. par représentation du nombre de chiffres successifs ou groupes de chiffres de même type à l'aide d'un mot-code et d'un chiffre représentant ce type
G06F 3/06 - Entrée numérique à partir de, ou sortie numérique vers des supports d'enregistrement
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
H03M 7/42 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse utilisant une table pour le procédé de codage ou de décodage, p. ex. utilisant une mémoire morte
52.
Method, apparatus, and computer-readable medium for parallelization of a computer program on a plurality of computing cores
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
A method and system of compiling and linking source stream programs for efficient use of multi-node devices. The system includes a compiler, a linker, a loader and a runtime component. The process converts a source code stream program to a compiled object code that is used with a programmable node based computing device having a plurality of processing nodes coupled to each other. The programming modules include stream statements for input values and output values in the form of sources and destinations for at least one of the plurality of processing nodes and stream statements that determine the streaming flow of values for the at least one of the plurality of processing nodes. The compiler converts the source code stream based program to object modules, object module instances and executables. The linker matches the object module instances to at least one of the multiple cores. The loader loads the tasks required by the object modules in the nodes and configure the nodes matched with the object module instances. The runtime component runs the converted program.
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06Q 40/00 - FinanceAssuranceStratégies fiscalesTraitement des impôts sur les sociétés ou sur le revenu
Representative embodiments are disclosed for a rapid and highly parallel decompression of compressed executable and other files, such as executable files for operating systems and applications, having compressed blocks including run length encoded (“RLE”) data having data-dependent references. An exemplary embodiment includes a plurality of processors or processor cores to identify a start or end of each compressed block; to partially decompress, in parallel, a selected compressed block into independent data, dependent (RLE) data, and linked dependent (RLE) data; to sequence the independent data, dependent (RLE) data, and linked dependent (RLE) data from a plurality of partial decompressions of a plurality of compressed blocks, to obtain data specified by the dependent (RLE) data and linked dependent (RLE) data, and to insert the obtained data into a corresponding location in an uncompressed file. The representative embodiments are also applicable to other types of data processing for applications having data dependencies.
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
G06F 17/30 - Recherche documentaire; Structures de bases de données à cet effet
H03M 7/46 - Conversion en, ou à partir de codes à longueur de série, c.-à-d. par représentation du nombre de chiffres successifs ou groupes de chiffres de même type à l'aide d'un mot-code et d'un chiffre représentant ce type
H03M 7/42 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse utilisant une table pour le procédé de codage ou de décodage, p. ex. utilisant une mémoire morte
56.
Method, apparatus, and computer-readable medium for parallelization of a computer program on a plurality of computing cores
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06Q 40/00 - FinanceAssuranceStratégies fiscalesTraitement des impôts sur les sociétés ou sur le revenu
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
57.
Method, apparatus, and computer-readable medium for parallelization of a computer program on a plurality of computing cores
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
G06Q 40/00 - FinanceAssuranceStratégies fiscalesTraitement des impôts sur les sociétés ou sur le revenu
58.
Method, apparatus, and computer-readable medium for parallelization of a computer program on a plurality of computing cores
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
G06Q 40/00 - FinanceAssuranceStratégies fiscalesTraitement des impôts sur les sociétés ou sur le revenu
59.
Parallel processing of data having data dependencies for accelerating the launch and performance of operating systems and other computing applications
Representative embodiments are disclosed for a rapid and highly parallel decompression of compressed executable and other files, such as executable files for operating systems and applications, having compressed blocks including run length encoded (“RLE”) data having data-dependent references. An exemplary embodiment includes a plurality of processors or processor cores to identify a start or end of each compressed block; to partially decompress, in parallel, a selected compressed block into independent data, dependent (RLE) data, and linked dependent (RLE) data; to sequence the independent data, dependent (RLE) data, and linked dependent (RLE) data from a plurality of partial decompressions of a plurality of compressed blocks, to obtain data specified by the dependent (RLE) data and linked dependent (RLE) data, and to insert the obtained data into a corresponding location in an uncompressed file. The representative embodiments are also applicable to other types of data processing for applications having data dependencies.
H03M 7/40 - Conversion en, ou à partir de codes de longueur variable, p. ex. code Shannon-Fano, code Huffman, code Morse
H03M 7/30 - CompressionExpansionÉlimination de données inutiles, p. ex. réduction de redondance
H03M 7/46 - Conversion en, ou à partir de codes à longueur de série, c.-à-d. par représentation du nombre de chiffres successifs ou groupes de chiffres de même type à l'aide d'un mot-code et d'un chiffre représentant ce type
G06F 3/06 - Entrée numérique à partir de, ou sortie numérique vers des supports d'enregistrement
60.
Method and apparatus for a compiler and related components for stream-based computations for a general-purpose, multiple-core system
A method and system of compiling and linking source stream programs for efficient use of multi-node devices. The system includes a compiler, a linker, a loader and a runtime component. The process converts a source code stream program to a compiled object code that is used with a programmable node based computing device having a plurality of processing nodes coupled to each other. The programming modules include stream statements for input values and output values in the form of sources and destinations for at least one of the plurality of processing nodes and stream statements that determine the streaming flow of values for the at least one of the plurality of processing nodes. The compiler converts the source code stream based program to object modules, object module instances and executables. The linker matches the object module instances to at least one of the multiple cores. The loader loads the tasks required by the object modules in the nodes and configure the nodes matched with the object module instances. The runtime component runs the converted program.
An apparatus, computer-readable medium, and computer-implemented method for parallelization of a computer program on a plurality of computing cores includes receiving a computer program comprising a plurality of commands, decomposing the plurality of commands into a plurality of node networks, each node network corresponding to a command in the plurality of commands and including one or more nodes corresponding to execution dependencies of the command, mapping the plurality of node networks to a plurality of systolic arrays, each systolic array comprising a plurality of cells and each non-data node in each node network being mapped to a cell in the plurality of cells, and mapping each cell in each systolic array to a computing core in the plurality of computing cores.
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
G06Q 40/00 - FinanceAssuranceStratégies fiscalesTraitement des impôts sur les sociétés ou sur le revenu
62.
Method and apparatus for a general-purpose multiple-core system for implementing stream-based computations
A method and system of efficient use and programming of a multi-processing core device. The system includes a programming construct that is based on stream-domain code. A programmable core based computing device is disclosed. The computing device includes a plurality of processing cores coupled to each other. A memory stores stream-domain code including a stream defining a stream destination module and a stream source module. The stream source module places data values in the stream and the stream conveys data values from the stream source module to the stream destination module. A runtime system detects when the data values are available to the stream destination module and schedules the stream destination module for execution on one of the plurality of processing cores.
G06F 9/46 - Dispositions pour la multiprogrammation
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
G06F 15/00 - Calculateurs numériques en généralÉquipement de traitement de données en général
G06F 9/38 - Exécution simultanée d'instructions, p. ex. pipeline ou lecture en mémoire
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
63.
Method and apparatus for a compiler and related components for stream-based computations for a general-purpose, multiple-core system
A method and system of compiling and linking source stream programs for efficient use of multi-node devices. The system includes a compiler, a linker, a loader and a runtime component. The process converts a source code stream program to a compiled object code that is used with a programmable node based computing device having a plurality of processing nodes coupled to each other. The programming modules include stream statements for input values and output values in the form of sources and destinations for at least one of the plurality of processing nodes and stream statements that determine the streaming flow of values for the at least one of the plurality of processing nodes. The compiler converts the source code stream based program to object modules, object module instances and executables. The linker matches the object module instances to at least one of the multiple cores. The loader loads the tasks required by the object modules in the nodes and configure the nodes matched with the object module instances. The runtime component runs the converted program.
A method and system of efficient use and programming of a multi-processing core device. The system includes a programming construct that is based on stream-domain code. A programmable core based computing device is disclosed. The computing device includes a plurality of processing cores coupled to each other. A memory stores stream-domain code including a stream defining a stream destination module and a stream source module. The stream source module places data values in the stream and the stream conveys data values from the stream source module to the stream destination module. A runtime system detects when the data values are available to the stream destination module and schedules the stream destination module for execution on one of the plurality of processing cores.
G06F 9/46 - Dispositions pour la multiprogrammation
G06F 9/45 - Compilation ou interprétation de langages de programmation évolués
G06F 15/76 - Architectures de calculateurs universels à programmes enregistrés
G06F 9/40 - Dispositions pour exécuter des sous-programmes, c. à d. des combinaisons de plusieurs instructions
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
