MulticoreWare and Imagination have enabled GPU compute on TI’s TDA4VM processor, adding around 50 GFLOPS of extra compute and demonstrating improvement in the performance of common workloads used for ADAS systems (ADAS).

The collaboration has achieved over 100x performance gains when running a stereo block matching (StereoBM) algorithm on the GPU rather than on the CPU on a high resolution (3200×2000) image.

14oct15Connected-Car-Picture1.jpg

This enables automotive customers to use IMG BXS GPUs with OpenCL to achieve maximum performance per watt on TDA4 SoCs for sensor processing workloads (camera, radar, and lidar) augmenting other compute accelerators existing in the platform.

“The automotive software revolution is driving the need for intelligent features on low-power SoCs,” says Multicore’s Vish Rajalingam, “our experience with OpenCL and Edge AI across all major AI accelerators, GPUs, and SoCs unlocks incredible possibilities for the future of mobility.”

MulticoreWare optimized the StereoBM algorithm to leverage GPU cores efficiently. The IMG BXS-4-64 GPU intrinsics and adaptive memory handling helped achieve optimal performance on higher resolution camera data.

“The graphics processor in a car is a flexible, programmable, and high-performance option for supporting a vehicle’s ADAS workloads,” says Imagination vp Jake Kochnowicz.