The University of California, Riverside (UCR) has introduced a groundbreaking research concept known as simultaneous and heterogeneous multithreading (SHMT), which has the potential to double the speed of devices without hardware upgrades.
This innovative approach utilizes existing processors in modern gadgets like phones and computers, such as graphics processing units (GPUs) and tensor processing units (TPUs), to execute tasks more efficiently.
By running multiple subtasks simultaneously across different processors, SHMT aims to reduce bottlenecks and improve speed and energy efficiency. The study by UCR’s team demonstrated a 1.95 times faster sample code execution while reducing energy consumption by 51 percent. However, challenges remain in integrating SHMT into everyday devices, including optimizing computing jobs for different processors and ensuring quality assurance across heterogeneous architectures.