English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Compiler writers have avoided using dynamic programming to guide instruction selection becaus e of fears of exponential growth in the size of the search space . We show how the use of predicates to guar d the matching of patterns reduces the complexity, both in time and in space, to linear with a small coefficient . The benefits of dynamic programming are thus achieved at relatively low cost . We present a language and a translation tool that partially automate the construction of the Instruction Selection Component (ISC) of a compiler . The specification language is nonprocedural and supports three separate descriptions : the inter mediate language (IL) emitted by our compiler front-ends, the instruction set (ISP) of our target machine , and a set of patterns that map between IL tree shapes and sequences of instructions . This language supports the computation of both inherited and synthesized attributes . The patterns allow semantic predicates and multi-dimensional costing . The generated ISC uses bottom-up tree pattern matching and dynamic programming to generat e very good local code on a per-statement basis . Section 2 gives an overview of the specification language, including some of the ways in which w e express similarity among related concepts . Section 3 describes the translator tool . Section 4 discusse s instruction selection at compiler run-time . Section 5 illustrates the deferral of computation via additiona l nonterminals and synthesized attributes, and how some of the deferred computation can be eliminate d depending on context . Section 6 lists some techniques for preventing exponential explosion of the searc h space during dynamic programming . Section 7 presents data that demonstrate that dynamic programming yields competitive compile-time performance in a family of (commercially-available) optimizing compilers . Section 8 compares our results with previous work . Appendix A goes into more detail on dynami c programming .

Dynamic programming and industrial-strength instruction selection: code generation by tiring, but not exhaustive, search