Adaptive-Fidelity Multiplication Hardware for Accuracy and Energy Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern computing systems face challenges in efficiently executing large-scale composite computations due to their complexity, leading to intractable computational tasks and significant energy consumption, especially with increasing demand for computational resources.
Innovation Solution
A computational hardware block executes multiplication computations in multiple temporal phases with adaptive fidelity, adjusting the number of phases based on a fidelity control value, which can be set ex ante or during computation, to balance resource usage and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the number of temporal phases is increased to improve computational accuracy, then manufacturing precision is improved, but use of energy increases and device complexity increases
Solution Approach 1:
The system dynamically adjusts the number of temporal phases executed based on a fidelity control value, allowing the computational process to adapt its precision level. This dynamic adjustment enables the system to execute only the necessary number of phases required to achieve adequate accuracy, avoiding unnecessary energy consumption from excessive precision calculations.
Solution Approach 2:
The fidelity control value serves as a parameter that directly controls the number of temporal phases executed. By changing this parameter, the system can adjust the balance between computational accuracy and energy consumption, executing more phases when high precision is needed and fewer phases when lower precision suffices, thus optimizing energy usage.
2Manufacturing precision
If the number of temporal phases is increased to improve computational accuracy, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The system employs dynamic control through the fidelity control value to adjust the number of temporal phases executed. This dynamic approach allows the computational device to adapt its complexity level based on the specific computation requirements, avoiding the need for a fixed high-complexity architecture that would be required to handle all possible accuracy requirements.
Solution Approach 2:
By using the fidelity control value as a adjustable parameter, the system can modify the number of temporal phases executed without changing the underlying hardware architecture. This parameter-based control allows the same computational device to operate at different complexity levels, executing only the necessary number of phases for each specific computation task.
3Productivity
If computational resources are increased to handle larger composite computations, then productivity is improved, but use of energy increases
Solution Approach 1:
The system applies partial action by executing only the necessary number of temporal phases required to achieve adequate computational accuracy for each specific task. Rather than always executing the maximum number of phases or using full computational resources, the system performs just enough computation to meet the required fidelity level, thereby improving productivity without proportionally increasing energy consumption.
Solution Approach 2:
The dynamic adjustment of temporal phases based on the fidelity control value allows the system to optimize its resource utilization in real-time. For computations that require lower precision, fewer resources are allocated, while for computations requiring higher precision, more resources are dynamically allocated, thus improving overall productivity without linearly increasing energy consumption across all tasks.
Data Source
AI summary
Methods and systems relating to computational hardware are disclosed herein. One disclosed method for executing a multiplication computation using a computational hardware block includes storing a first operand and a second operand for the multiplication computation. The first operand includes a first set of bit strings. The second operand includes a second set of bit strings. The method also includes multiplying the first set of bit strings and the second set of bit strings in a set of temporal phases using the computational hardware block. Each temporal phase uses a different group of bit strings from the first set of bit strings and the second set of bit strings. A cardinality of the set of temporal phases is determined by a fidelity control value. The fidelity control value adaptively sets a fidelity of execution of the multiplication computation.