Imagination Technologies picks MulticoreWare for GPU algorithm optimisation

Author: EIS Release Date: Nov 28, 2022

Imagination Technologies has picked software services company MulticoreWare, which specialises in computer vision, sensor data and AI processing, as a software development partner to optimise algorithms on Imagination GPUs.

 

Unisoc T710

“MulticoreWare is utilising our IP for general-purpose GPU applications. By working together, we can offer our customers an optimised PowerVR deployment,” said Imagination product director Gilberto Rodriguez.

 

As a demonstration, the companies implemented a stereo block matching algorithm on Unisoc’s T710 SoC development platform, with MulticoreWare using OpenCL to implementing the StereoBM algorithm and deliver “greater than ~50x performance gain compared to the CPUs on the same platform”, claimed Imagination. “MulticoreWare analysed the CPU performance to identify bottlenecks. The goal was to achieve maximum GPU parallelism, which was enabled by efficient implementation of internal register usage and the configuration of an appropriate global workgroup size that is adaptive to the image resolution”.

 

Imagination’s GM9446 GPU memory layout was used to calculate the adaptive workgroup size. CPU time in one configuration was 54.25ms, it said, whereas the modified implementation was 780μs.

 

“We are excited to be partnering with Imagination to in implementing algorithm optimisations GPUs, with a planned roadmap to include RISC-V software acceleration in future,” said MulticoreWare v-p Vish Rajalingam.