On the side-effects of code abstraction

Sutter, Bjorn De; Vandierendonck, Hans; Bus, Bruno De; Bosschere, Koen De

doi:10.1145/780731.780766

Cited by 5 publications

(5 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With respect to the instruction cache misses, the results presented in the top chart of Figure 20 are in line with the results presented by Debray et al [2000] and De Sutter et al [2003]. Because of the whole-program optimizations, the working set size decreased, and therefore fewer instruction cache misses were observed when no duplicate code elimination was applied that introduced overhead.…”

Section: Instruction Countssupporting

confidence: 82%

See 1 more Smart Citation

Link-time binary rewriting techniques for program compaction

Sutter

Bus

Bosschere

2005

ACM Trans. Program. Lang. Syst.

Self Cite

View full text Add to dashboard Cite

Small program size is an important requirement for embedded systems with limited amounts of memory. We describe how link-time compaction through binary rewriting can achieve code size reductions of up to 62% for statically bound languages such as C, C++, and Fortran, without compromising on performance. We demonstrate how the limited amount of information about a program at link time can be exploited to overcome overhead resulting from separate compilation. This is done with scalable, cost-effective, whole-program analyses, optimizations, and duplicate code and data elimination techniques. The discussed techniques are evaluated and their cost-effectiveness is quantified with SQUEEZE++, a prototype link-time compactor.

show abstract

Section: Instruction Countssupporting

confidence: 82%

“…Reducing the size of the whole program indeed does not imply that the working set size is reduced. In practice, the working set size even increases [De Sutter et al 2003]. This follows from the 90/10 rule that states that 90% of the executed instructions corresponds to 10% of the static instructions.…”

Section: Cost Of Duplicate Code Eliminationmentioning

confidence: 99%

Link-time binary rewriting techniques for program compaction

Sutter

Bus

Bosschere

2005

ACM Trans. Program. Lang. Syst.

Self Cite

View full text Add to dashboard Cite

show abstract

“…~time (f ) < threshold (1) where time09 denotes the percentage of execution time spent in the function. Functions are selected using a greedy algorithm with a ratio…”

Section: A Compilation Iterationmentioning

confidence: 99%

“…This trade-off issue has already been pointed out in [3,23] but no algorithm has been proposed to automatically compute a solution for maximizing the compression rate while minimizing the performance penalty. The selection of the code to compress in the previous works is left to the user [9,15,23] or is fixed [1,3].…”

Section: Introductionmentioning

confidence: 99%

A software-only compression system for trading-offs between performance and code size

Heydemann¹,

Bodin²,

Charles³

2005

Proceedings of the 2005 Workshop on Software and Compilers for Embedded Systems - SCOPES '05

View full text Add to dashboard Cite

The design of an embedded system is often heavily constrained by its performance objective and budget envelope. Software code compression may reduce the instruction memory space and then the overall cost of the system. However, it may also induce performance degradation. Previous studies proposed selective code compression using profile information in order to reduce the performance penalty. In this paper, we go one step further. We propose a software-only compression system, called SCS, that automatically finds trade-offs between code size and performance. Through an iterative approach, SCS automatically determines which fimctions to be compressed given a performance constraint and~or a code size constraint in order to guarantee a min~al performance and a maximal code size for an application. Experimentations illustrate that even with a non-optimal software decompression approach, SCS achieves a high compression rate with a minimal performance degradation.

show abstract

“…As various found code fragments can overlap, only one of the marked code fragments is extracted per mining step. Two different extraction methods are used: if a fragments ends with an unconditional return statement or a branch instruction, a cross jump or tail merge is used [16,17]. Otherwise a subroutine call is necessary.…”

mentioning

confidence: 99%