Hung-Wei Tseng, a UC Riverside associate professor of electrical and computer engineering, has proposed a groundbreaking paradigm shift in computer architecture. The innovative method aims to double the processing power of your smartphone, tablet, personal computer, or server using the existing hardware already in these devices.
In his recent paper, Tseng discussed this new approach to computer architecture that could significantly boost processing power without requiring additional hardware.
In today’s computer devices, we often find graphics processing units (GPUs), hardware accelerators for artificial intelligence (AI) and machine learning (ML), or digital signal processing units. However, these components work independently and hence need to transfer information from one processing unit to another, which often results in creating a bottleneck.
In their paper, Tseng and UCR computer science graduate student Kuan-Chieh Hsu have introduced a new concept called “simultaneous and heterogeneous multithreading” or SHMT. It is a programming and execution model that enables opportunities for “real” parallel processing using heterogeneous processing units.
The proposed SHMT framework on an embedded system platform uses a multi-core ARM processor, an NVIDIA GPU, and a Tensor Processing Unit hardware accelerator simultaneously. This model also has an abstraction and a runtime system that helps with parallel execution. Additionally, SHMT needs to additionally address the heterogeneity in data precision across different processing units to ensure accurate results.
The system achieved a 1.96 times speedup and a 51% reduction in energy consumption. “You don’t have to add new processors because you already have them,” Tseng said.
The simultaneous use of different processing components could have several benefits, such as reducing computer hardware costs and carbon emissions from running data processing centers. Additionally, this approach could also reduce the need for freshwater used to keep servers cool.
The paper by Tseng highlights that further research is needed to address questions related to system implementation, hardware support, code optimization, and which applications would benefit the most from this approach, among other issues. However, the concept of SHMT represents an exciting step toward more efficient and sustainable computing.
Journal reference:
- Kuan-Chieh Hsu and Hung-Wei Tseng. Simultaneous and Heterogenous Multithreading. DOI: 10.1145/3613424.3614285