Gm Global Technolgoy Operations, Inc.

Roman L Millett

Shuqing Zeng

Wei Tong

Shige Wang

A signal processing system includes a central processing unit (CPU) in communication with an accelerator, and an instruction scheduler in communication with the accelerator. A first memory device including a first instruction set is configured to operate the accelerator, a second instruction set is configured to operate the CPU, and a second memory device is configured to receive a datafile. The accelerator includes a plurality of processing engines (PEs) and an instruction scheduler, the instruction set includes a plurality of operators, and the instruction scheduler is configured to implement the operators in the accelerator employing the PEs. The CPU employs the operators implemented in the accelerator to analyze the datafile to extract a feature therefrom.

Method and apparatus for a neural network

In one example implementation according to aspects of the present disclosure, a computer-implemented method includes capturing a plurality of images at a camera associated with a vehicle and storing image data associated with the plurality of images to a memory. The method further includes dispatching vehicle perception tasks to a plurality of processing elements of an accelerator in communication with the memory. The method further includes performing, by at least one of the plurality of processing elements, the vehicle perception tasks for the vehicle perception using a neural network, wherein performing the vehicle perception tasks comprises performing an activation bypass for values below a first threshold, and performing weight pruning of synapses and neurons of the neural network based at least in part on a second threshold. The method further includes controlling the vehicle based at least in part on a result of performing the vehicle perception tasks.

Activation zero-bypass and weight pruning in neural networks for vehicle perception systems

Examples of techniques for using fixed-point quantization in deep neural networks are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes capturing a plurality of images at a camera associated with a vehicle and storing image data associated with the plurality of images to a memory. The method further includes dispatching vehicle perception tasks to a plurality of processing elements of an accelerator in communication with the memory. The method further includes performing, by at least one of the plurality of processing elements, the vehicle perception tasks for the vehicle perception using a neural network, wherein performing the vehicle perception tasks comprises quantizing a fixed-point value based on an activation input and a synapse weight. The method further includes controlling the vehicle based at least in part on a result of performing the vehicle perception tasks.

Fixed-point quantization in neural networks for vehicle perception systems

A method and system including a central processing unit (CPU), an accelerator, a communication bus and a system memory device for dynamically processing an image file are described. The accelerator includes a local memory buffer, a data transfer scheduler, and a plurality of processing engines. The data transfer scheduler is arranged to manage data transfer between the system memory device and the local memory buffer, wherein the data transfer includes data associated with the image file. The local memory buffer is configured as a circular line buffer, and the data transfer scheduler includes a ping-pong buffer for transferring output data from the one of the processing engines to the system memory device. The local memory buffer is configured to execute cross-layer usage of data associated with the image file.

Method and apparatus for memory access management for data processing

An adaptive parallel imaging processing system in a vehicle is provided. The system may include, but is not limited to, a plurality of processors and a resource management system including, but not limited to, an execution monitor, the execution monitor configured to calculate an average utilization of each of the plurality of processors over a moving window, and a service scheduler controlling a request queue for each of the plurality of processors, the service scheduler scheduling image processing tasks in the respective request queue for the each of the plurality of processors based upon the average utilization of each of the plurality of processors, the capabilities of each of the plurality of processors, and a priority associated with each image processing task, wherein an autonomous vehicle control system is configured to generate the instructions to control the at least one vehicle system based upon the processed image processing tasks.

Growing community of inventors

Clinton Township, MI, United States of America

Roman L Millett