1413 results about "Hardware acceleration" patented technology

An integrated circuit for data encryption/decryption and secure key management is disclosed. The integrated circuit may be used in conjunction with other integrated circuits, processors, and software to construct a wide variety of secure data processing, storage, and communication systems. A preferred embodiment of the integrated circuit includes a symmetric block cipher that may be scaled to strike a favorable balance among processing throughput and power consumption. The modular architecture also supports multiple encryption modes and key management functions such as one-way cryptographic hash and random number generator functions that leverage the scalable symmetric block cipher. The integrated circuit may also include a key management processor that can be programmed to support a wide variety of asymmetric key cryptography functions for secure key exchange with remote key storage devices and enterprise key management servers. Internal data and key buffers enable the device to re-key encrypted data without exposing data. The key management functions allow the device to function as a cryptographic domain bridge in a federated security architecture.

Disclosed herein is a method and system for hardware-accelerating the generation of metadata for a data stream using a coprocessor. Using these techniques, data can be richly indexed, classified, and clustered at high speeds. Reconfigurable logic such a field programmable gate arrays (FPGAs) can be used by the coprocessor for this hardware acceleration. Techniques such as exact matching, approximate matching, and regular expression pattern matching can be employed by the coprocessor to generate desired metadata for the data stream.

A compression system is arranged to use software and/or hardware accelerated compression techniques to increase compression speeds and enhance overall data throughput. A logic circuit is arranged to: receive a data stream from a flow control processor, buffer the data stream, select a hardware compressor (e.g., an ASIC), and forward the data to the selected hardware compressor. Each hardware compressor performs compression on the data (e.g., LZ77), and sends the compressed data back to the logic circuit. The logic circuit receives the compressed data, converts the data to another compressed format (e.g., GZIP), and forwards the converted and compressed data back to the flow control processor. History associated with the data stream can be stored in memory by the flow control processor, or in the logic circuit.

Disclosed herein is a method and system for integrating an enterprise's structured and unstructured data to provide users and enterprise applications with efficient and intelligent access to that data. Queries can be directed toward both an enterprise's structured and unstructured data using standardized database query formats such as SQL commands. A coprocessor can be used to hardware-accelerate data processing tasks (such as full-text searching) on unstructured data as necessary to handle a query. Furthermore, traditional relational database techniques can be used to access structured data stored by a relational database to determine which portions of the enterprise's unstructured data should be delivered to the coprocessor for hardware-accelerated data processing.

Error-free state tables are automatically generated from a specification of a group of desired performable functions, such as are provided in a programming language in a formal notation such as Backus-Naur form or a derivative thereof by discriminating tokens corresponding to respective performable functions, identifications, arguments, syntax, grammar rules, special symbols and the like. The tokens may be recursive (e.g. infinite), in which case they are transformed into a finite automata which may be deterministic or non-deterministic. Non-deterministic finite automata are transformed into deterministic finite automata and then into state transitions which are used to build a state table which can then be stored or, preferably, loaded into a finite state machine of a hardware parser accelerator to define its personality.

A system including a storage processing device with an input/output module. The input/output module has port processors to receive and transmit network traffic. The input/output module also has a switch connecting the port processors. Each port processor categorizes the network traffic as fast path network traffic or control path network traffic. The switch routes fast path network traffic from an ingress port processor to a specified egress port processor. The storage processing device also includes a control module to process the control path network traffic received from the ingress port processor. The control module routes processed control path network traffic to the switch for routing to a defined egress port processor. The control module is connected to the input/output module. The input/output module and the control module are configured to interactively support data virtualization, data migration, data journaling, and snapshotting. The distributed control and fast path processors achieve scaling of storage network software. The storage processors provide line-speed processing of storage data using a rich set of storage-optimized hardware acceleration engines. The multi-protocol switching fabric provides a low-latency, protocol-neutral interconnect that integrally links all components with any-to-any non-blocking throughput.

A variety of embodiments for hardware-accelerating the processing of financial market depth data are disclosed. A coprocessor, which may be resident in a ticker plant, can be configured to update order books based on financial market depth data at extremely low latency. Such a coprocessor can also be configured to enrich a stream of limit order events pertaining to financial instruments with data from a plurality of updated order books.

A low-latency virtual reality display system that provides rapid updates to virtual reality displays in response to a user's movements. Display images generated by rendering a 3D virtual world are modified using an image warping approximation, which allows updates to be calculated and displayed quickly. Image warping is performed using various rerendering approximations that use simplified 3D models and simplified 3D to 2D projections. An example of such a rerendering approximation is a shifting of all pixels in the display by a pixel translation vector that is calculated based on the user's movements; pixel translation may be done very quickly, possibly using hardware acceleration, and may be sufficiently realistic particularly for small changes in a user's position and orientation. Additional features include techniques to fill holes in displays generated by image warping, synchronizing full rendering with approximate rerendering, and using prediction of a user's future movements to reduce apparent latency.

A variety of embodiments for hardware-accelerating the processing of financial market depth data are disclosed. A coprocessor, which may be resident in a ticker plant, can be configured to update order books based on financial market depth data at extremely low latency. Such a coprocessor can also be configured to generate a quote event in response to a limit order event being determined to modify the top of an order book.

Hardware assisted transactional memory system with open nested transactions. Some embodiments described herein implement a system whereby hardware acceleration of transactions can be accomplished by implementing open nested transaction in hardware which respect software locks such that a top level transaction can be implemented in software, and thus not be limited by hardware constraints typical when using hardware transactional memory systems.

The invention provides a hardware accelerator and method for realizing a sparse GRU neural network based on FPGA. According to the invention, an apparatus for realizing sparse GRU neural network includes: an input receiving unit which is intended for receiving a plurality of input vectors and distributing the plurality of input vectors to a plurality of computing units; a plurality of computing units which acquire input vectors from the input receiving unit, read weight matrix data of a neural network, decoding the weight matrix data and conduct matrix calculation on the decoded weight matrix data and the input vectors, and output the result of the matrix calculation to a hidden layer state computing module; a hidden layer state computing module which acquires the result of matrix calculation from the computing units PE, and computing the state of the hidden layer; and a control unit which is intended for global control. In addition, the invention also provides a method for realizing sparse GRU neural network through iteration.

A method and apparatus for virtual application of features to electronic messages is disclosed. When a device applies a set of features to an electronic message, one or more of the features may be virtually applied instead of actually applied. For example, instead of encrypting a payload portion of a packet and adding an encryption header, the packet may not be encrypted. However, an appropriate encryption header may still be included in the packet such that the packet appears to have been encrypted when other features are applied. Prior to sending the packet, the payload portion is actually encrypted, such as by using a hardware accelerator. Some implementations may use a dual processor router, in which the input/output processor controls the hardware accelerator, the routing processor performs the virtual application of a feature, and prior to sending the packet the input/output processor actually applies the virtually applied feature.

A method and apparatus of displaying an Electronic Programming Guide (EPG). In one embodiment, an EPG is displayed in a three dimensional virtual mesh, in which independent objects representing television programs are situated. The simplified nature of the three dimensional EPG reduces the amount of processing necessary to display it. In addition, the virtual mesh may be displayed isometrically, so that hardware requirements are further reduced and it may be possible to use a software only three dimensional graphics pipeline. If a user has a set top box (STB) with a hardware accelerated graphics pipeline, the EPG may be displayed in a full three dimensional perspective view. A user can navigate the mesh to find television programs that they wish to view. A user can assign values to types of television programs that they prefer, and these programs will be displayed more prominently.

Multiple hardware accelerators can be used to efficiently perform processes that would otherwise be performed by general purpose hardware running software. The software overhead and bus bandwidth associated with running multiple hardware acceleration processes can be reduced by chaining multiple independent hardware acceleration operations within a circuit card assembly. Multiple independent hardware accelerators can be configured on a single circuit card assembly that is coupled to a computing device. The computing device can generate a playlist of hardware acceleration operations identifying hardware accelerators and associated accelerator options. A task management unit on the circuit card assembly receives the playlist and schedules the hardware acceleration operations such that multiple acceleration operations may be successively chained together without intervening data exchanges with the computing device.

The invention relates to a convolutional neural network hardware acceleration device, a convolution calculation method, and a storage medium. The device comprises an instruction processing unit, a hardware acceleration module and an external data memory unit, wherein the instruction processing unit decodes an instruction set to execute a corresponding operation so as to control the hardware acceleration module; the hardware acceleration module comprises an input caching unit, a data operation unit and an output caching unit, wherein the input caching unit executes the memory access operation of the instruction processing unit, and stores data read from the external data memory unit; the data operation unit executes the operation execution operation of the instruction processing unit, processes the data operation of the convolutional neural network, and controls a data operation process and a data flow direction according to an operation instruction set; the output caching unit executesthe memory access operation of the instruction processing unit, and stores a calculation result which is output by the data operation unit and needs to be written into the external data memory unit;and the external data memory unit stores a calculation result output by the output caching unit and transmits data to the input caching unit according to the reading of the input caching unit.

A computer-based method and system for interactive volume rendering of a large volume data on a conventional personal computer using hardware-accelerated block filtration optimizing uses 3D-textured axis-aligned slices and block filtration. Fragment processing in a rendering pipeline is lessened by passing fragments to various processors selectively in blocks of voxels based on a filtering process operative on slices. The process involves generating a corresponding image texture and performing two-pass rendering, namely a virtual rendering pass and a main rendering pass. Block filtration is divided into static block filtration and dynamic block filtration. The static block filtration locates any view-independent unused signal being passed to a rasterization pipeline. The dynamic block filtration determines any view-dependent unused block generated due to occlusion. Block filtration processes utilize the vertex shader and pixel shader of a GPU in conventional personal computer graphics hardware. The method is for multi-thread, multi-GPU operation.