Systematic application of data transfer and storage optimizing code transformations for power consumption and execution time reduction in ACROPOLIS: A pre-compiler for multimedia applications
A systematic methodology for the application of data transfer and storage optimizing code transformations to high-level descriptions of multimedia systems realized on instruction set processors is proposed. A detailed order for the application of different data transfer and storage optimizing transf...
Gespeichert in:
Veröffentlicht in: | Design automation for embedded systems 2003-03, Vol.8 (1), p.51-86 |
---|---|
Hauptverfasser: | , , , |
Format: | Artikel |
Sprache: | eng |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | A systematic methodology for the application of data transfer and storage optimizing code transformations to high-level descriptions of multimedia systems realized on instruction set processors is proposed. A detailed order for the application of different data transfer and storage optimizing transformations is proposed in the context of combined execution time and power optimizations. A use methodology including a number of support steps that allow the efficient application of the data transfer and storage oriented transformations is proposed as well. Application of the proposed transformation-based methodology moves the main part of the memory accesses from the large background memories (lying possibly off-chip) to smaller ones (on-chip) or even to foreground storage (registers). Data cache performance is improved thus reducing power consumption in the data memory hierarchy and related interconnects. Execution time and the power consumption due to instruction storage and transfers are reduced as well after the application of the proposed methodology. Experimental results from several real-life multimedia applications prove the effectiveness of the proposed methodology. The proposed approach has been applied in the context of realizations on custom hardware processors as well with promising results. |
---|---|
ISSN: | 0929-5585 |
DOI: | 10.1023/A:1022340119745 |