Abstract: A processing unit can include a performance monitor for monitoring the performance of the processing unit and associated sub-units. The performance monitor can include a state machine. The state machine can be implemented via state machine data entries stored in a memory associated with the performance monitor. A state machine data entry includes information indicating a state transition condition and output signals. The state transition condition includes a current state and input signals required to meet the condition. The output signals include a next state, one or more counter actions, and one or more triggers. The performance monitor implements logic circuits that determine, based on input signals and the state machine data entries, the next state to transition and associated output signals. The state machine data entries can be written and re-written by a user.

Abstract: Various implementations of a current flattening circuit are disclosed, including those utilizing a feedback current regulator, a feedforward current regulator, and a constant current source.

Abstract: In various examples, a multi-sensor fusion machine learning model—such as a deep neural network (DNN)—may be deployed to fuse data from a plurality of individual machine learning models. As such, the multi-sensor fusion network may use outputs from a plurality of machine learning models as input to generate a fused output that represents data from fields of view or sensory fields of each of the sensors supplying the machine learning models, while accounting for learned associations between boundary or overlap regions of the various fields of view of the source sensors. In this way, the fused output may be less likely to include duplicate, inaccurate, or noisy data with respect to objects or features in the environment, as the fusion network may be trained to account for multiple instances of a same object appearing in different input representations.

Abstract: To mitigate pulse shape degradation along a signal route, the signal is driven from two ends. One end of the route is loaded and the other is relatively unloaded. The loaded route and unloaded route may traverse two different metal layers on a printed circuit board. The two routes may thus be related such that the unloaded route has less RC distortion effects on the signal than does the loaded route.

Abstract: Layout techniques for chip packages on printed circuit boards are disclosed that address the multivariate problem of minimizing routing distances for high-speed I/O pins between chip packages while simultaneously providing for the rapid provision of transient power demands to the chip packages. The layout techniques may also enable improved thermal management for the chip packages.

Abstract: A graphics subsystem includes a printed circuit board (PCB), a set of one or more fans, and a heat sink. A graphics processing unit (GPU) is integrated into the PCB. The PCB is shortened to occupy a portion of the width of the graphics subsystem. The heat sink is coupled to the PCB and/or GPU and configured to extend beyond an edge of the PCB, thereby occupying a larger portion of the width of the graphics subsystem compared to the PCB. A first fan is disposed partially or fully beyond the edge of the PCB and is configured to direct air through the portion of the heat sink that extends beyond the edge of the PCB, along a first airflow path, and out of the graphics subsystem. A second fan is configured to direct air through the heat sink, along a second airflow path, towards the GPU.

Abstract: A time-to-digital converter (TDC) circuit includes control logic and a first self-referenced delay cell circuit coupled to the control logic. The first self-referenced delay cell circuit includes: a first bank of capacitors coupled to a first node between a first positive input and a first positive output, where the first bank of capacitors is selectively controlled by a first control signal from the control logic, the first control signal including a first up value corresponding to a first positive threshold; and a second bank of capacitors coupled to a second node between a first negative input and a first negative output, where the second bank of capacitors is selectively controlled by a second control signal from the control logic, the second control signal including a first down value corresponding to a first negative threshold.

Abstract: A client computer executes a client agent that determines the time between when an input is transmitted from a client computer to a remote server and when updated graphics are received from the remote server in response. The client agent interacts with a server agent that executes on the remote server. The client agent transmits an emulated keystroke to the server agent. In response, the server agent modifies a graphics object and composites the modified graphics object with a currently rendered frame. The client computer receives the frame and identifies the modified graphics object, indicating that the emulated keystroke was received by the remote server. The client agent then computes the time difference between when the emulated keystroke was transmitted to the remote server and when the modified graphics object was detected and/or displayed at the client computer. This time difference indicates the responsiveness of the remote desktop implementation.

Abstract: Novel solutions are provided for consistent Quality of Service in cloud gaming system that adaptively and dynamically compensate for poor network conditions by moderating rendered frame rates using frame rate capping to optimize for network latency savings (or surplus). In further embodiments, the encoding/sent frame rate to the client can also be managed in addition, or as an alternative to capping the rendered frame rates. The claimed embodiments not only maintain a constant Quality of Service (QoS) for the user, but may also be employed to leverage higher-performing networks to reduce operational costs.

Abstract: In various examples, cloud computing systems may store frames of video streams and metadata generated from the frames in separate data stores, with each type of data being indexed using shared timestamps. Thus, the frames of a video stream may be stored and/or processed and corresponding metadata of the frames may be stored and/or generated across any number of devices of the cloud computing system (e.g., edge and/or core devices) while being linked by the timestamps. A client device may provide a request or query to dynamically annotate the video stream using a particular subset of the metadata. In processing the request or query, the timestamps may be used to retrieve video data representing frames of the video stream and metadata extracted from those frames across the data stores. The retrieved metadata and video data may be used to annotate the frames for display on the client device.

Abstract: A style-based generative network architecture enables scale-specific control of synthesized output data, such as images. During training, the style-based generative neural network (generator neural network) includes a mapping network and a synthesis network. During prediction, the mapping network may be omitted, replicated, or evaluated several times. The synthesis network may be used to generate highly varied, high-quality output data with a wide variety of attributes. For example, when used to generate images of people's faces, the attributes that may vary are age, ethnicity, camera viewpoint, pose, face shape, eyeglasses, colors (eyes, hair, etc.), hair style, lighting, background, etc. Depending on the task, generated output data may include images, audio, video, three-dimensional (3D) objects, text, etc.

Abstract: A cooling system for a datacenter is disclosed. An evaporative cooling subsystem provides blown air for cooling the datacenter and a repurposable refrigerant cooling subsystem controls moisture of the blown air in a first configuration and independently cools the datacenter in a second configuration.

Abstract: Systems and methods are disclosed herein for a pedestrian crossing warning system that may use multi-modal technology to determine attributes of a person and provide a warning to the person in response to a calculated risk level to effect a reduction of the risk level. The system may utilize sensors to receive data indicative of a trajectory of a person external to the vehicle. Specific attributes of the person such as age or walking aids may be determined. Based on the trajectory data and the specific attributes, a risk level may be determined by the system using a machine learning model. The system may cause emission of a warning to the person in response to the risk level.

Abstract: In various examples, a signal processing pipeline is dynamically generated or instantiated for a signal processing request. To generate the pipeline, a graph topology—including nodes and edges—may be created to represent features, functionality, and characteristics of a signal processing system. The nodes, representing processing tasks, may be connected via edges having associated costs for performing, by a node, a processing task on an output of a prior or edge-connected node. For a given signal processing request, the nodes or processing tasks to be included may be selected and, using a graph routing algorithm and the costs between and among the determined nodes, a path through the nodes may be determined—thereby defining, at least in part, the signal processing pipeline.

Abstract: Verified deliveries are commonplace for various exchanges of goods, packages, and/or other items, but often require close proximity or contact between the exchanging parties or devices associated therewith—e.g., for digital or physical signature. To remedy this, system and methods described herein may leverage an ad hoc network established between a device of a provider and a device of a consumer for exchanging codes or tokens—that may be validated by an authentication service—to provide a verification process during an exchange between the parties. As a result, a safe distance may be maintained between the parties throughout the transaction—thereby avoiding exchange of germs while also increasing safety and security of both parties—and the verification process may be more reliable and secure.

Abstract: Systems and methods for detecting blockages in images are described. An example method may include receiving a plurality of images captured by a camera installed on an apparatus. The method may include identifying one or more candidate blocked regions in the plurality of images. Each of the candidate blocked regions may contain image data caused by blockages in the camera's field-of-view. The method may further include assigning scores to the one or more candidate blocked regions based on relationships among the one or more candidate blocked regions in the plurality of images. In response to a determination that one of the scores is above a predetermined blockage threshold, the method may include generating an alarm signal for the apparatus.

Abstract: Apparatuses, systems, and techniques are directed to automatic coalescing of GPU-initiated network communications. In one method, a communication engine receives, from a shared memory application executing on a first graphics processing unit (GPU), a first communication request assigned to or having a second GPU as a destination to be processed. The communication engine determines that the first communication request satisfies a coalescing criterion and stores the first communication request in association with a group of requests that have a common property. The communication engine coalesces the group of requests into a coalesced request and transports the coalesced request to the second GPU over a network.

Abstract: In a ray tracer, a cache for streaming workloads groups ray requests for coherent successive bounding volume hierarchy traversal operations by sending common data down an attached data path to all ray requests in the group at the same time or about the same time. Grouping the requests provides good performance with a smaller number of cache lines.

Abstract: Apparatuses, systems, and techniques are presented to reduce noise in audio. In at least one embodiment, one or more neural networks are used to determine a noise signal in one or more speech signals.

Abstract: An autonomous vehicle system removes ephemeral points from lidar samples. The system receives a plurality of light detection and ranging (lidar) samples captured by a lidar sensor. Along with the lidar samples, the system receives an aligned pose and an unwinding transform for each of the lidar samples. The system determines one or more occupied voxel cells in a three-dimensional (3D) space using the lidar samples, their aligned poses, and their unwinding transforms. The system identifies occupied voxel cells representative of noise associated with motion of an object relative to the lidar sensor. The system filters the occupied voxel cells by removing the cells representative of noise. The system inputs the filtered occupied voxel cells in a 3D map comprising voxel cells, e.g., during the map generation and/or a map update.