Abstract: Certain aspects of the present disclosure provide techniques for processing machine learning model data with a machine learning task accelerator, including: configuring one or more signal processing units (SPUs) of the machine learning task accelerator to process a machine learning model; providing model input data to the one or more configured SPUs; processing the model input data with the machine learning model using the one or more configured SPUs; and receiving output data from the one or more configured SPUs.

Abstract: Provided are louver blade, louvered roof, and viewing shelter. Drainage groove is arranged on one side of body of louver blade; and includes groove wall located on outer side of body. Baffle plate is arranged on top of groove wall. When body is arranged horizontally, top end of groove wall is not higher than top surface of louver blade. Baffle plate can switch between natural and stressed states. Top end of baffle plate in natural state is higher than top end of groove wall, and has height that can be reduced in stressed state. Louvered roof includes shelter frame and multiple louver blades. When multiple louver blades are closed on shelter frame, other side of one louver blade relative to drainage groove presses against baffle plate of adjacent louver blade. Baffle plate switches to stressed state. When louver blades rotate to open, baffle plate switches to natural state.

Abstract: Provided are a low-noise air conditioner and a viewing shelter. The low-noise air conditioner includes a housing and a heat exchange module. A partition is arranged inside the housing to divide the housing into an upper cavity and at least one lower cavity. The heat exchange module is arranged inside the upper cavity, and the lower cavity includes a first channel and a second channel, wherein the first channel and the second channel are respectively configured to communicate indoor air with the upper cavity. The viewing shelter includes a structural frame and the low-noise air conditioner mentioned above. The low-noise air conditioner is mounted on the structural frame.

Abstract: Certain aspects provide an apparatus for signal processing in a neural network. The apparatus generally includes first computation in memory (CIM) cells configured as a first kernel for a neural network computation, the first set of CIM cells comprising one or more first columns and a first plurality of rows of a CIM array. The apparatus also include a second set of CIM cells configured as a second kernel for the neural network computation, the second set of CIM cells comprising the one or more first columns and a second plurality of rows of the CIM array. The first plurality of rows may be different than the second plurality of rows.

Abstract: The utility model relates to the technical field of tables, and particularly to a folding table which comprises a tabletop and hinging pieces, wherein the tabletop comprises at least two table plates, and adjacent table plates are hinged by the hinging pieces to make the tabletop in an unfolded state or a folded state; twist locks are arranged on the adjacent table plates close to hinging positions; the operating part is connected with the sliding rod and is used for poking the sliding rod to slide. The utility model solves the problem that assembly and disassembly of an existing folding table is relatively cumbersome.

Abstract: Devices and systems for controlling pressure during sensing are disclosed. Such sensors and/or systems may be embodied in a wearable user device. The wearable user device may include a cuff configured to apply a pressure to a portion of a user at a plurality of discrete pressure levels over a plurality of time periods; a biometric sensor configured to obtain a plurality of sensor measurements associated with a blood vessel of the user at corresponding ones of the plurality of discrete pressure levels during corresponding ones of the plurality of time periods; and a wearable structure including the cuff and the biometric sensor.

Abstract: Certain aspects of the present disclosure provide techniques for performing machine learning computations in a compute in memory (CIM) array comprising a plurality of bit cells, including: determining that a sparsity of input data to a machine learning model exceeds an input data sparsity threshold; disabling one or more bit cells in the CIM array based on the sparsity of the input data prior to processing the input data; processing the input data with bit cells not disabled in the CIM array to generate an output value; applying a compensation to the output value based on the sparsity to generate a compensated output value; and outputting the compensated output value.

Abstract: A lower vehicle body rear structure includes a rear floor integrally formed in a die-casting mode, rear floor longitudinal beam rear sections fixedly connected to the rear floor, a shelf, and a spare tire cabin. Rear floor longitudinal beams arranged on the two sides of the rear floor, and a rear floor middle cross beam and a rear floor rear cross beam which are connected between the rear floor longitudinal beams are formed on the rear floor, and rear wheel covers connected to the rear floor longitudinal beams are respectively arranged on the two sides of the rear floor. The shelf is located above the rear floor, and the two ends of the shelf are connected to the rear wheel covers by means of connecting plates; and the rear floor, the shelf and the two connecting plates are connected to form an annular structure.

Abstract: Certain aspects of the present disclosure provide techniques for performing machine learning computations in a compute in memory (CIM) array comprising a plurality of bit cells, including: determining that a sparsity of input data to a machine learning model exceeds an input data sparsity threshold; disabling one or more bit cells in the CIM array based on the sparsity of the input data prior to processing the input data; processing the input data with bit cells not disabled in the CIM array to generate an output value; applying a compensation to the output value based on the sparsity to generate a compensated output value; and outputting the compensated output value.

Abstract: An Ethernet data transmission circuit, an Ethernet data transmission system and an Ethernet data transmission method are provided. The Ethernet data transmission circuit includes: a polarity processing circuit for processing a polarity carried by Ethernet data into a preset polarity; and an encoder for receiving the Ethernet data and the preset polarity carried by the Ethernet data, and encoding the Ethernet data. On the one hand, the security of Ethernet in a transmission process can be improved; on the other hand, without increasing workload of the encoder, the polarity processing circuit of the Ethernet data transmission circuit can be used to process the Ethernet data to be with a preset polarity, to facilitate the encoder to code.

Abstract: Devices and systems for controlling pressure during sensing are disclosed. Such sensors and/or systems may be embodied in a wearable user device. The wearable user device may include a cuff configured to apply a pressure to a portion of a user at a plurality of discrete pressure levels over a plurality of time periods; a biometric sensor configured to obtain a plurality of sensor measurements associated with a blood vessel of the user at corresponding ones of the plurality of discrete pressure levels during corresponding ones of the plurality of time periods; and a wearable structure including the cuff and the biometric sensor.

Abstract: This specification provides an LSM tree-based data storage method and a related device, applied to an LSM tree-based data storage system. The method includes: determining whether the first storage layer meets a merge condition for merging with the second storage layer, and if yes, selecting a to-be-merged target file from the at least one first file stored at the first storage layer, where the target file includes data corresponding to a target type; and searching the plurality of second sub-files for a target sub-file that includes data corresponding to the target type, and merging the target file and the target sub-file.

Abstract: Synchronized sensors and systems are disclosed. Techniques involving a synchronized sensor system for determining a physiological parameter of a user may include: obtaining one or more first measurements at a first location of the user via a first sensor; obtaining one or more second measurements at a second location of the user via a second sensor; and determining the physiological parameter of the user based on the one or more first measurements, the one or more second measurements, and a distance between the first location and the second location, the distance between the first location and the second location determined based on acoustic communication between the first sensor and the second sensor. In some implementations, acoustic communication may include ultrasound signals between the first sensor and the second sensor, which may be time synchronized by exchanging timestamps.

Abstract: This application discloses a method for monitoring SH of an epitaxial layer of a CMOS device, fabricating a measurement reference device, AA and STI structures of the measurement reference device being the same as those of the CMOS device to be monitored, polysilicon lines being formed on the AA and STI structures of the measurement reference device, the polysilicon lines on the AA structures being connected in series to obtain a first polysilicon line string, the polysilicon lines on the STI structures being connected in series to obtain a second polysilicon line string; obtain current I1 flowing through the first polysilicon line string and current I2 flowing through the second polysilicon line string; and performing calculation according to I1 and I2 to obtain SH of an epitaxial layer of the measurement reference device as SH of an epitaxial layer of the CMOS device to be monitored.

Abstract: Certain aspects of the present disclosure provide a method, including: storing a depthwise convolution kernel in a first one or more columns of a CIM array; storing a fused convolution kernel in a second one or more columns of the CIM array; storing pre-activations in one or more input data buffers associated with a plurality of rows of the CIM array; processing the pre-activations with the depthwise convolution kernel in order to generate depthwise output; modifying one or more of the pre-activations based on the depthwise output to generate modified pre-activations; and processing the modified pre-activations with the fused convolution kernel to generate fused output.

Abstract: Certain aspects provide an apparatus for signal processing in a neural network. The apparatus generally includes computation circuitry configured to perform a convolution operation, the computation circuitry having multiple input rows, and an activation buffer having multiple buffer segments coupled to the multiple the multiple input rows of the computation circuitry, respectively. In some aspects, each of the multiple buffer segments comprises a first multiplexer having a plurality of multiplexer inputs, and each of the plurality of multiplexer inputs of one of the first multiplexers on one of the multiple buffer segments is coupled to a data output of the activation buffer on another one of the multiple buffer segments.