Abstract: Instruction set architecture extensions to configure priority ordering of divergent target branch instructions on SIMT computing platforms to enable tools such as compilers (e.g., under influence of execution profilers) or human software developers to configure branch direction prioritization explicitly in code. Extensions for simple (two-way) branch instructions as well as multi-target (more than two branch target instructions) are disclosed.

Abstract: Various embodiments include a modulo operation generator associated with a cache memory in a computer-based system. The modulo operation generator generates a first sum by performing an addition and/or a subtraction function on an input address. A first portion of the first sum is applied to a lookup table that generates a correction value. The correction value is then added to a second portion of the first sum to generate a second sum. The second sum is adjusted, as needed, to be less than the divisor. The adjusted second sum forms a residue value that identifies a cache memory slice in which the input data value corresponding to the input address is stored. By generating the residue value in this manner, the cache memory efficiently distributes input data values among the slices in a cache memory even when the number of slices is not a power of two.

Abstract: Apparatuses, systems, and techniques to optimize kernel selection for performing a computation. In at least one embodiment, a neural network is trained and utilized to generate a list of kernels so that an (e.g., optimal) kernel may be identified. The neural network receives characteristics of the input matrices and determines relevancy scores for a list of possible kernels. Based on an ordered listing of kernels by relevant score, a kernel is selected from the list and utilized to perform the computation and provide the result.

Abstract: A latent code defined in an input space is processed by the mapping neural network to produce an intermediate latent code defined in an intermediate latent space. The intermediate latent code may be used as appearance vector that is processed by the synthesis neural network to generate an image. The appearance vector is a compressed encoding of data, such as video frames including a person's face, audio, and other data. Captured images may be converted into appearance vectors at a local device and transmitted to a remote device using much less bandwidth compared with transmitting the captured images. A synthesis neural network at the remote device reconstructs the images for display.

Abstract: In general, an application executes on a compute unit, such as a central processing unit (CPU) or graphics processing unit (GPU), to perform some function(s). In some circumstances, improved performance of an application, such as a graphics application, may be provided by executing the application across multiple compute units. However, when using multiple compute units in this manner, synchronization must be provided between the compute units. Synchronization, including the sharing of the data, is typically accomplished through memory. While a shared memory may cause bottlenecks, employing local memory for each compute unit may itself require synchronization (coherence) which can be costly in terms of resources, delay, etc. The present disclosure provides read-write page replication for multiple compute units that avoids the traditional challenges associated with coherence.

Abstract: Multiple snapshots of a scene are captured within an executing application (e.g., a video game). When each snapshot is captured, associated color values per pixel and a distance or depth value z per pixel are stored. The depth information from the snapshots is accessed, and a point cloud representing the depth information is constructed. A mesh structure is constructed from the point cloud. The light field(s) on the surface(s) of the mesh structure are calculated. A surface light field is represented as a texture. A renderer uses the surface light field with geometry information to reproduce the scene captured in the snapshots. The reproduced scene can be manipulated and viewed from different perspectives.

Abstract: When communicating through shared memory, a producer thread generates a value that is written to a location in a shared memory. The value is read from the shared memory by a consumer thread. The challenge is to ensure that the consumer thread reads the location only after the value is written and is thereby synchronized. When a memory location is written by a producer thread, a flag that is simultaneously stored in the memory location along with the value is toggled. The consumer thread tracks information to determine whether the flag stored in the location indicates whether the producer has written the value to the location. The flag is read and written simultaneously with reading and writing the location in memory, thereby eliminating the need for a memory fence. After all of the consumer threads read the value, the location may be reused to write additional value(s) and simultaneously toggle the flag.

Abstract: In various embodiments, a current measurement circuit measures an input current within an integrated circuit. The current measurement circuit includes an integration capacitor, an operational amplifier, a comparison capacitor, an inverter, and multiple switches. The current measurement circuit is coupled to a clocking circuit that, during operation, generates a two-phase clock having a frequency that is proportional to the input current. At least a portion of the switches are turned on during a first phase of the two-phase clock and are turned off during a second phase of the two-phase clock.

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: According to an aspect of an embodiment, operations may comprise (a) accessing a portion of a high definition (HD) map comprising a point cloud of a region through which a vehicle is driving, (b) identifying a base LIDAR from a plurality of LIDARs mounted on the vehicle, (c) for each of the LIDARs: receiving a LIDAR scan comprising a point cloud of the region, and determining a pose for the LIDAR, (d) for each LIDAR other than the base LIDAR, determining a transform for the LIDAR with respect to the base LIDAR, (e) repeating (c) to generate a plurality of samples, (f) for each of the samples, repeating (d) to determine a plurality of transforms for each LIDAR with respect to the base LIDAR, and (g) calibrating each of the LIDARs other than the base LIDAR by determining an aggregate transform for the LIDAR.

Abstract: A game summary may be produced using an event log of in-game events and corresponding game content based on game data associated with a gameplay session(s). The event log may include metadata that indicates times of in-game events and associations between in-game events and game content items that capture the in-game events. A user may interact with in-game events with temporal context, allowing for more informed selections and a better understanding of the gameplay session. Using the event log, a game summary may provide such features as a timeline to convey relative timing of in-game events, a list of in-game events, a map of a virtual environment of the game that is temporally annotated based on in-game events, game status information, and statistical and performance information. A game summary may show trends over time and/or game sessions and convey information for selected sets of players, such as teams.

Abstract: In accordance with the disclosure, an inductor may be formed over a semiconductor substrate of one or both dies in a face-to-face die arrangement while reducing the parasitic capacitance between the inductor and the adjacent die. In disclosed embodiments, a semiconductor device may include a void (e.g., an air gap) between the inductor and the adjacent die to reduce the parasitic capacitance between the inductor and the adjacent die. The void may be formed in the die that includes the inductor and/or the adjacent die. In some respects, the void may be etched in interface layers (e.g., comprising bump pads and dielectric material) between the semiconductor dies, and may extend along the length of the inductor.

Abstract: A semiconductor assembly is described that includes a substrate having top and bottom sides. An integrated circuit die coupled to the substrate includes first and second distinct sets of ground pads. In some embodiments, the first and second sets of ground pads are configured to have distinct ground return paths to a host system. In further embodiments, one of the ground return paths may include a metal plate coupled between ground contacts on the top side of the substrate and ground contacts on a printed circuit board of the host system.

Abstract: Appearance driven automatic three-dimensional (3D) modeling enables optimization of a 3D model comprising the shape and appearance of a particular 3D scene or object. Triangle meshes and shading models may be jointly optimized to match the appearance of a reference 3D model based on reference images of the reference 3D model. Compared with the reference 3D model, the optimized 3D model is a lower resolution 3D model that can be rendered in less time. More specifically, the optimized 3D model may include fewer geometric primitives compared with the reference 3D model. In contrast with the conventional inverse rendering or analysis-by-synthesis modeling tools, the shape and appearance representations of the 3D model are automatically generated that, when rendered, match the reference images.

Abstract: In various examples, high beam control for vehicles may be automated using a deep neural network (DNN) that processes sensor data received from vehicle sensors. The DNN may process the sensor data to output pixel-level semantic segmentation masks in order to differentiate actionable objects (e.g., vehicles with front or back lights lit, bicyclists, or pedestrians) from other objects (e.g., parked vehicles). Resulting segmentation masks output by the DNN(s), when combined with one or more post processing steps, may be used to generate masks for automated high beam on/off activation and/or dimming or shading—thereby providing additional illumination of an environment for the driver while controlling downstream effects of high beam glare for active vehicles.

Abstract: In various examples, a trigger signal may be received that is indicative of a vehicle maneuver to be performed by a vehicle. A recommended vehicle trajectory for the vehicle maneuver may be determined in response to the trigger signal being received. To determine the recommended vehicle trajectory, sensor data may be received that represents a field of view of at least one sensor of the vehicle. A value of a control input and the sensor data may then be applied to a machine learning model(s) and the machine learning model(s) may compute output data that includes vehicle control data that represents the recommended vehicle trajectory for the vehicle through at least a portion of the vehicle maneuver. The vehicle control data may then be sent to a control component of the vehicle to cause the vehicle to be controlled according to the vehicle control data.

Abstract: In various examples, a generative model is used to synthesize datasets for use in training a downstream machine learning model to perform an associated task. The synthesized datasets may be generated by sampling a scene graph from a scene grammar—such as a probabilistic grammar—and applying the scene graph to the generative model to compute updated scene graphs more representative of object attribute distributions of real-world datasets. The downstream machine learning model may be validated against a real-world validation dataset, and the performance of the model on the real-world validation dataset may be used as an additional factor in further training or fine-tuning the generative model for generating the synthesized datasets specific to the task of the downstream machine learning model.

Abstract: In various embodiments, a computing system includes, for example, a plurality of processing units that share access to a system cache. A cache management application receives, for example, resource savings information for each processing unit. The resource savings information indicates, for example, amounts of a resource (e.g., power) that are saved when different units of the system cache are allocated to a processing unit. The cache management application determines, for example, the number of units of system cache to allocate to each processing unit based on the received resource savings information.

Abstract: A latent code defined in an input space is processed by the mapping neural network to produce an intermediate latent code defined in an intermediate latent space. The intermediate latent code may be used as appearance vector that is processed by the synthesis neural network to generate an image. The appearance vector is a compressed encoding of data, such as video frames including a person's face, audio, and other data. Captured images may be converted into appearance vectors at a local device and transmitted to a remote device using much less bandwidth compared with transmitting the captured images. A synthesis neural network at the remote device reconstructs the images for display.

Abstract: A real-time neural radiance caching technique for path-traced global illumination is implemented using a neural network for caching scattered radiance components of global illumination. The neural (network) radiance cache handles fully dynamic scenes, and makes no assumptions about the camera, lighting, geometry, and materials. In contrast with conventional caching, the data-driven approach sidesteps many difficulties of caching algorithms, such as locating, interpolating, and updating cache points. The neural radiance cache is trained via online learning during rendering. Advantages of the neural radiance cache are noise reduction and real-time performance. Importantly, the runtime overhead and memory footprint of the neural radiance cache are stable and independent of scene complexity.