What is it about?
Processor overclocking is an important technique that enables the faster code execution in certain operating conditions. We propose a method for low power processors which considers the "delay" of each operation to optimize the execution performance. In our approach, the processor behavior adapts to each individual operation instead of relying on a pre-defined static functionality. This increases the performance of a system by a large margin.
Featured Image
Photo by Adi Goldstein on Unsplash
Why is it important?
We focus on the individual characteristics of the processor instead of relying in a "general purpose" methodology which would do the trick, but it would under-perform in certain designs. By taking into account the microarchitecture of the processor we also ensure that out methodology can be adapted to server many designs. In our approach we utilize a more "optimistic" approach on how the code should execute on a given system by using a novel way to view the clock frequency of a circuit. Thus, of asking " what is the worst case of the system?" we try to simply ask "what the system can do?"/
Perspectives
This work involved a lot of hard work and dedication to complete since the engineering problems that hindered us from finalizing the architecture were huge. Personally, I am really excited that the system we designed managed to obtain such a performance increase and we are all very pleased that our initial idea held some water. I think that this publication on this prestigious journal can be considered as a reward to our commitment and effort.
Athanasios Tziouvaras
University of Thessaly
Read the Original
This page is a summary of: Low-power Near-data Instruction Execution Leveraging Opcode-based Timing Analysis, ACM Transactions on Architecture and Code Optimization, June 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3504005.
You can read the full text:
Contributors
The following have contributed to this page
