Dictionary-based code compression techniques in VLIW architecturesVLIW 아키텍쳐에서의 사전 참조 방식의 코드 압축 기법

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 330
  • Download : 0
Reducing code size is crucial in embedded systems as well as in high-performance systems to overcome the communication bottleneck between memory and CPU, especially with VLIW (Very Long Instruction Word) processors that require a high-bandwidth instruction prefetching. This paper presents two new approaches for dictionary-based code compression in VLIW processor-based systems. The first approach uses opcode-isomorphism and operand-isomorphism among instruction words. After instruction words are divided into two groups, one for an opcode group and the other for an operand group, the proposed compression algorithm is applied to each group to obtain maximal code compression. Frequently-used instruction words are extracted from the original code to be mapped into two dictionaries, an it opcode dictionary and an it operand dictionary. According to the SPEC95 benchmarks, the proposed technique has achieved an average code compression ratio of 61.56%, 61.3%, and 57.98% in 4-issue, 8-issue, and 12-issue VLIW architectures, respectively and our scheme provide a good trade-off between code compression ratio and the decoding logic delay. Moreover, opcode-isomorphic or operand-isomorphic instruction words achieve the power reduction by 39.93%, 39.40%, and 44.31% in 4-issue, 8-issue, and 12-issue VLIW architectures, respectively. As the second approach, we propose a code compression method based on the dictionary-based code compression, which uses the mobility of instructions to enhance the code compression ratio compared to the first isomorphism method. In the compilation phase, instructions in a basic block can be rescheduled using their mobility. This rescheduling helps to find more identical or opcode-isomorphic/operand-isomorphic instruction words. To implement this scheme, we have used the two-phase algorithm, making global compaction information and compacting the instructions. This scheme improves the code compression ratio by about 10% compared to that of the pre...
Advisors
Kyung, Chong-Minresearcher경종민researcher
Description
한국과학기술원 : 전기및전자공학전공,
Publisher
한국과학기술원
Issue Date
2000
Identifier
157632/325007 / 000955108
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 전기및전자공학전공, 2000.2, [ ix, 80 p. ]

URI
http://hdl.handle.net/10203/35838
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=157632&flag=dissertation
Appears in Collection
EE-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0