Fragment-Level Translation for Emulation Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Just-In-Time (JIT) translation in emulation slows down execution due to dynamic target system code changes, particularly in systems with a main CPU and attached digital signal processors (DSPs), leading to performance issues and increased translation loads.
Innovation Solution
Fragment-level translation splits the emulation task into instruction memory emulation, main translation, and secondary processor emulation, allowing for parallel processing and efficient resource usage by grouping instructions into fragments and translating them on a fragment-by-fragment basis, reducing unnecessary code generation and retranslation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If JIT translation is used to translate target system code to host system code, then code execution capability is achieved, but execution speed decreases due to translation overhead and dynamic code changes
Solution Approach 1:
The patent segments the translation process into fragment-level units rather than translating entire programs. Each fragment is a discrete code segment that can be translated independently, allowing the system to translate only the necessary portions of code that are actually executed, thereby reducing overall translation overhead and improving execution speed while maintaining adaptability.
Solution Approach 2:
The patent implements speculative forward translation where code fragments are translated in advance before they are needed during execution. This preliminary translation action eliminates the need for on-demand translation during critical execution paths, reducing translation overhead and improving execution speed.
2Adaptability or versatility
If dynamic target system code changes are handled by retranslation, then code updates are supported, but translation load increases and performance decreases
Solution Approach 1:
The patent applies local quality by tracking and managing only the specific fragments of code that have changed, rather than retreating entire programs. Each fragment maintains its own translation state and validity information, allowing the system to perform minimal retranslation only for affected local regions, thereby supporting code updates while maintaining high translation efficiency.
Solution Approach 2:
The patent implements feedback mechanisms through code change detection and fragment validity tracking. When code changes are detected, the system uses feedback information about which fragments are affected to determine the minimal set that needs retranslation, optimizing the balance between supporting code updates and maintaining translation efficiency.
3Reliability
If entire code segments are translated, then translation completeness is achieved, but unnecessary code generation increases resource usage
Solution Approach 1:
The patent applies partial action by translating only the necessary portions of code (fragments) that are actually executed, rather than translating entire code segments. This selective translation approach maintains translation completeness for executed code while significantly reducing unnecessary code generation and resource usage.
Data Source
AI summary
Emulation of a target system with a host system is disclosed. Two or more target system code instructions may be grouped into one or more fragments. A main translation function may be implemented by translating each fragment into a corresponding set of position-independent instructions executable by the host system. A target processor may be emulated by executing the corresponding set of position-independent executable instructions with the host system.