Abstract: Light contribution information can be determined and cached for use in rendering image frames for a scene. In at least one embodiment, a spatial hash data structure can be used to split the scene into regions, such as octahedral voxels. Using cast light rays, an average light contribution can be computed for each individual voxel. Those light values can then be used to build a cumulative distribution function for each voxel that can be used to select which lights to sample for a given frame during rendering. The sampling for a region or voxel can be based at least in part upon the number of contributing lights for that region, as well as the relative contributions of those lights. Such an approach can be very bandwidth and cache efficient, while providing high image quality.

Abstract: Embodiments relate to methods for efficiently encoding sensor data captured by an autonomous vehicle and building a high definition map using the encoded sensor data. The sensor data can be LiDAR data which is expressed as multiple image representations. Image representations that include important LiDAR data undergo a lossless compression while image representations that include LiDAR data that is more error-tolerant undergo a lossy compression. Therefore, the compressed sensor data can be transmitted to an online system for building a high definition map. When building a high definition map, entities, such as road signs and road lines, are constructed such that when encoded and compressed, the high definition map consumes less storage space. The positions of entities are expressed in relation to a reference centerline in the high definition map. Therefore, each position of an entity can be expressed in fewer numerical digits in comparison to conventional methods.

Abstract: Embodiments are disclosed for driving a vertical cavity surface emitting laser (VCSEL). An example method includes injecting, via a universal driver, a direct current (DC) bias current to a VCSEL. The VCSEL is configured to convert the modulated signal into an optical signal encoding one or more bits. The example method further includes providing a modulated signal to the VCSEL. The modulated signal encodes a digital sequence comprising the one or more bits using a modulation method.

Abstract: Various embodiments include an error correction code (ECC) system that provides protection against various errors in addition to data bit errors. In general, ECC codes protect against data bit errors, where one or more data bits in a data word contain the wrong value. The ECC code is based on the original data bits, such that a data bit error results in a data word that is inconsistent with the ECC code generated for and stored with the data word. The present embodiments generate ECC codes based on address information and/or sequencing information in addition to the data bits in the data word. As a result, the present embodiments detect bit errors in this address information and/or sequencing information. Such errors include write address decoding errors, read address decoding errors, write enable errors, and stale data errors.

Abstract: A system calibrates one or more sensors mounted to an autonomous vehicle. From the one or more sensors, the system identifies a primary sensor and a secondary sensor. The system determines a reference angle for the primary sensor, and based on that reference angle for the primary sensor, a scan-start time representing a start of a scan and a scan-end time representing an end of a scan. The system receives, from the primary sensor, a primary set of scan data recorded from the scan-start time to the scan-end time. The system receives, from the secondary sensor, a secondary set of sensor data recorded from the scan-start time to the scan-end time. The system calibrates the primary and secondary sensors by determining a relative transform for transforming points between the first set of scan data and the second set of scan data.

Abstract: A method for rendering a light field comprises projecting rays from a viewpoint positioned at a first side of a spatial light modulator (SLM) to a clipping plane positioned at an opposing side of the SLM to form an elemental view frustum within a three-dimensional scene and rendering objects within the elemental view frustum to generate components of a first elemental image for the first elemental region. The SLM may include a tiled array of non-overlapping elemental regions and a top edge and a bottom edge of a first elemental region of the non-overlapping elemental regions are intersected by the rays to form the elemental view frustum. Furthermore, the light field may include the first elemental image and additional elemental images corresponding to the array of elemental regions and each one of the additional elemental images is rendered using an additional elemental view frustum.

Abstract: Systems and methods for cooling a datacenter are disclosed. In at least one embodiment, a first flow controller within a cooling manifold is associated with a second flow controller and with a tube there between; and the first flow controller is movable in at least one direction relative to dimensions of the cooling manifold so that it can be positioned for mating with a server tray or box and so that the second flow controller can be mated with a rack manifold.

Abstract: Bit generating cells are subjected to processes that accelerate aging-related characteristics before they are configured for use in the field (enrolled). Aging improves the reliability of the cells by shifting device characteristic in a direction that improves the cell behavior with respect not only to aging but also environment variations. Outputs of the cells are read, and the cells are reconfigured with a bias to output an opposite value, and then aged for enrollment.

Abstract: Systems and methods to detect one or more segments of one or more objects within one or more images based, at least in part, on a neural network trained in an unsupervised manner to infer the one or more segments. Systems and methods to help train one or more neural networks to detect one or more segments of one or more objects within one or more images in an unsupervised manner.

Abstract: Systems, processes, and techniques to automatically detect and enlarge a speaking one of plurality of participants on one side of a video conference. In at least one embodiment, the speaking participant is identified using one or more heuristics and/or one or more neural networks.

Abstract: Candidate grasping models of a deformable object are applied to generate a simulation of a response of the deformable object to the grasping model. From the simulation, grasp performance metrics for stress, deformation controllability, and instability of the response to the grasping model are obtained, and the grasp performance metrics are correlated with robotic grasp features.

Abstract: A ring oscillator circuit with a frequency that is sensitive to the timing of a clock-to-Q (clk2Q) propagation delay of one or more flip-flops utilized in the ring oscillator. The clock2Q is the delay between the clock signal arriving at the clock pin on the flop and the Q output reflecting the state of the input data signal to the flop. Clk2q delay measurements are made based on measurement of the ring oscillator frequency, leading to more accurate estimates of clk2Q for different types of flip-flops and flip-flop combinations, which may in turn enable improvements in circuit layouts, performance, and area.

Abstract: A computer based system and method for sending data packets over a data network may include: preparing data packets and packet descriptors on one or more graphical processing units (GPUs); associating packets with a packet descriptor, which may determine a desired transmission time of the packets associated with that descriptor; receiving an indication of a clock time; and physically transmitting packets via an output interface, at a clock time corresponding to the desired transmission time. A computer based system and method for GPU-initiated communication over a 5G data network may include allocating one or more memory buffers in GPU memory; performing at least one 5G signal processing procedure by a GPU; preparing descriptors for a plurality of packets, where each packet includes allocated memory buffers, and where the descriptors provide scheduling instructions for the packets; and triggering the sending of packets over the network based on prepared descriptors.

Abstract: A computer based system and method for sending data packets over a data network may include: preparing data packets and packet descriptors on one or more graphical processing units (GPUs); associating packets with a packet descriptor, which may determine a desired transmission time of the packets associated with that descriptor; receiving an indication of a clock time; and physically transmitting packets via an output interface, at a clock time corresponding to the desired transmission time. A computer based system and method for GPU-initiated communication over a 5G data network may include allocating one or more memory buffers in GPU memory; performing at least one 5G signal processing procedure by a GPU; preparing descriptors for a plurality of packets, where each packet includes allocated memory buffers, and where the descriptors provide scheduling instructions for the packets; and triggering the sending of packets over the network based on prepared descriptors.

Abstract: In various examples, as a user is speaking or presenting content during an online video conference, the data stream may be processed to generate a textual representation (e.g., transcript) of the audio and/or information relating to the video. The textual representation and/or video related information may then be processed to determine a context or one or more topic(s) of discussion. Based on the determined context/topic(s), a corresponding neural network(s) may be selected. Once a neural network has been selected, comments may be retrieved from a chat feature of the application and applied to the neural network. The neural network may then output data to indicate the relevance of the comments to the determined discussion topic. Based on the relevance of the comment, the comment may be allowed, prioritized, deleted, de-emphasized, or otherwise filtered in the chat feature.

Abstract: In various examples, sensor data—such as masked sensor data—may be used as input to a machine learning model to determine a confidence for object to person associations. The masked sensor data may focus the machine learning model on particular regions of the image that correspond to persons, objects, or some combination thereof. In some embodiments, coordinates corresponding to persons, objects, or combinations thereof, in addition to area ratios between various regions of the image corresponding to the persons, objects, or combinations thereof, may be used to further aid the machine learning model in focusing on important regions of the image for determining the object to person associations.

Abstract: According to an aspect of an embodiment, operations may comprise accessing an HD map of a region comprising information describing an intersection of two or more roads and describing lanes of the two or more roads that intersect the intersection, automatically identifying constraints on the lanes at the intersection, automatically calculating, based on the constraints on the lanes at the intersection, lane connectivity for the intersection, displaying, on a user interface, the automatically calculated lane connectivity for the intersection, receiving, from a user through the user interface, confirmation that the automatically calculated lane connectivity for the intersection is an actual lane connectivity for the intersection, and adding the actual lane connectivity for the intersection to the information describing the intersection in the HD map.

Abstract: Various embodiments include a memory device that is capable of performing write training operations, to determine that certain timing conditions are met, without storing data patterns in memory. Prior approaches for write training involve storing a long data pattern into the memory followed by reading the long data pattern to determine whether the data was written to memory correctly. Instead, the disclosed memory device generates a data pattern within the memory device that matches the data pattern being transmitted to the memory device by an external memory controller. If the data pattern generated by the memory device matches the data pattern received from the memory controller, then the memory device stores a pass status in a register. If the data patterns do not match, then the memory device stores a pass status in a register. The memory controller reads the register to determine whether the write training passed or failed.

Abstract: Various embodiments include a memory device that is capable of performing command address interface training operations, to determine that certain timing conditions are met, with fewer I/O pins relative to prior approaches. Prior approaches for command address interface training involve loading data via a set of input pins, a clock signal, and a clock enable signal that identifies when the input pins should be sampled. Instead, the disclosed memory device generates a data pattern within the memory device that matches the data pattern continuously being transmitted to the memory device by an external memory controller. The memory device compares the generated data pattern with the received data pattern and transmits the result of the comparison on one or more data output pins. The memory controller receives and analyzes the result of the comparison to determine whether the command address interface training passed or failed.

Abstract: A system for managing virtual memory. The system includes a first processing unit configured to execute a first operation that references a first virtual memory address. The system also includes a first memory management unit (MMU) associated with the first processing unit and configured to generate a first page fault upon determining that a first page table that is stored in a first memory unit associated with the first processing unit does not include a mapping corresponding to the first virtual memory address. The system further includes a first copy engine associated with the first processing unit. The first copy engine is configured to read a first command queue to determine a first mapping that corresponds to the first virtual memory address and is included in a first page state directory. The first copy engine is also configured to update the first page table to include the first mapping.