UC Riverside engineers awarded grant to develop new generation of GPUs for autonomous systems

June 29, 2020 | AUVSI News

The National Science Foundation has awarded three University of California, Riverside (UC Riverside) engineers a $1.2 million grant to develop a new generation of energy-efficient, energy-elastic, and real-time-aware Graphics processing units (GPUs) that can be used in resource-constrained environments such as emerging embedded and autonomous systems, including UAS and autonomous vehicles.

Daniel Wong, Hyoseung Kim, and Nael Abu-Ghazaleh are the recipients of the grant. All three are Marlan and Rosemary Bourns College of Engineering faculty members, and are professors of electrical and computer engineering.

GPUs commonly provide the computational power needed to enable autonomous systems. They are widely used in supercomputing and cloud computing to significantly speed up applications, such as image processing, deep learning, and other computationally intensive workloads. That added speed comes at a cost, though, as GPUs used in parallel computing consume large amounts of energy, which limits their use in self-contained, often battery-operated environments such as vehicles and drones.

Effective GPUs for autonomous systems need to be energy efficient and capable of executing workloads in real time. An example of this is that for an autonomous vehicle to safely navigate on the road, it has to be able to process a variety of sensor information, such as camera and lidar, and make a decision within milliseconds to prevent the vehicle from crashing.

But modern embedded GPUs have several limitations when used in autonomous system settings. GPUs tend to be energy inefficient, which leads to insufficient computational power and limited autonomous system capability. To successfully perform real-time operations and meet the workload deadlines required to provide correct and safe operations, GPU hardware and software need to be timing aware, so the UC Riverside project will look to solve these issues by providing solutions that cover both software and hardware in order to enable real-time embedded GPUs for autonomous systems.

“Current GPUs consume almost the same amount of power when actively processing a workload and when idle, wasting energy,” Wong says.

The UC Riverside team will design “energy-elastic” hardware, which lets the GPU consume power based on the amount of work it has to do.

“If it is doing little work, it will consume less power; if it needs to do more work, it will consume more power,” Wong explains.

GPU hardware is made up a lot of schedulers, which are unaware of the timing requirement of workloads. As a result, if multiple workloads are running on the GPU, it’s possible that some workloads may miss the deadline due to competition for hardware resources. The UCR group will create timing-aware hardware and software, which allows the various hardware schedulers to prioritize workloads to make sure that deadlines are met.

The researchers will design real-time scheduling software that coordinates with hardware schedulers. Having hardware that keeps the software updated on the workload’s progress will allow the software to make better scheduling decisions and improve real-time operations.

“Today’s software and hardware do not coordinate to enforce workload deadlines,” Kim says. “This project would enable multiple workloads to run safely together and increase the capabilities provided by embedded GPUs in autonomous systems.”