A. V. Aho, M. Ganapathi, and S. W. K. Tjiang, "Code generation using tree matching and dynamic programming," ACM Trans. Programming Languages and Systems, vol. 11, no. 4, pp. 491--516, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....to possible storage locations. To reduce the number of grammar rules, common subpatterns can be factored out; the rule describing a subpattern is then connected to the remaining rules via additional nonterminals. Examples of code selector generators using regular tree grammars include Twig [38], Burg [39] Iburg [40] and the Graham Glanville code generators [41] Several recent compilers for embedded processors have adopted Iburg for the code selection phase, such as CBC [36] Record [42] and the Spam project compiler [43] In the CBC compiler the regular tree grammar, that serves ....

....is executed. 37] also proves that these strong normal forms are optimal for the canonical problem introduced above. Dynamic programming is also used in more recent code selection approaches for processors with a heterogeneous register structure, such as the code selectors generated by Twig [38], Beg [53] and Iburg [40] Again, dynamic programming is used for cost calculation. However, this time a separate cost is calculated per register class,at each of the nodes in the subject tree. To keep the problem tractable, a number of simplifications are made. First of all, it is assumed that ....

A. V. Aho et al., "Code generation using tree matching and dynamic programming," ACM Trans. Prog. Lang. and Syst., vol. 11, no. 4, pp. 491--516, Oct. 1989.

....stringf ( s s d , p x.name, n 0 : n) which sets q x.name to p x. name concatenated with n or n. stringf returns a pointer to a string formatted as specified by its print style arguments. For example, in struct node struct node link; int count; a; f( int b[10] b[4] = a.count; suppose a and b point to the symbols for a and b, respectively. a x. name is set to a by defsymbol, and b x.name is set to 40 by local. address (q1, a, 4) is called with q1 representing the address of a. count, and q1 x. name is set to a 4 . Likewise, address (q2, ....

....and b, respectively. a x. name is set to a by defsymbol, and b x.name is set to 40 by local. address (q1, a, 4) is called with q1 representing the address of a. count, and q1 x. name is set to a 4 . Likewise, address (q2, b, 16) sets q2 x. name to 40 16 , which is the address of b[4]. 5.2 Compound Statements Source language blocks bracket the lifetime of locals. gencode announces the beginning and end of a block by calling blockbeg (e) and blockend (e) respectively. The argument e points to a target specific Env structure, which holds the data necessary to reuse registers ....

A. V. Aho, M. Ganapathi, and S. W. K. Tjiang, `Code generation using tree matching and dynamic programming,' ACM Trans. on Programming Languages and Systems, 11, 491--516 (1989).

....DFG representation of code from Fig. 17. Fig. 19. DFG from Fig. 18 covered by instruction patterns. Fig. 20. Using MAC for DFG covering. Fig. 21. Decomposition of a DFG into DFTs. Araujo and Malik [63] showed how the powerful standard technique of tree pattern matching with dynamic programming [59] widely used in compilers for general purpose processors can be effectively applied also to PDSPs with irregular data paths. Tree pattern matching with dynamic programming solves the code generation problem by parsing a given DFT with respect to an instruction set specification given as a tree ....

....derivation of a given DFT according to the grammar rules. This requires only two passes (bottom up and top down) over the nodes of the input DFT, so that the runtime is linear in the number of DFT nodes. Code generators based on this paradigm can be automatically generated with tools like twig [59] and iburg [62] The key idea in the approach by Araujo and Malik is the use of register specific instruction patterns or grammar rules. Instead of separating detailed register allocation from code selection as in classical compiler construction, the instruction patterns contain implicit ....

A. V. Aho, M. Ganapathi, and S. W. K Tjiang, "Code generation using tree matching and dynamic programming," ACM Trans. Programming Languages and Systems, vol. 11, no. 4, pp. 491--516, 1989.

....left according to the assigned set of functions. If more than one right side matches, the algorithm forks and follows all alternatives. The selection of the best solution by evaluation of a cost function, and the use of dynamic programming similar to the algorithms by Aho, Ganapathni, and Tjiang [1] is not feasible at this stage, as no local cost function can be applied. The compensation of yet unknown side effects can also not be taken into consideration at this stage. The estimated duration of the complete transaction can be used as a cost function. The evaluation of the cost function must ....

Alfred V. Aho, Mahadevan Ganapathi, and Steven W. K. Tjiang. Code generation using tree matching and dynamic programming. TOPLAS, 11(4):491--516, 1989.

....to hold additional information and to be used as conditions for matching in rewrite rules. Another difference between HATS and our approach is that HATS may sometimes use problem specific transformations, which our transformational system does not currently support. Two other systems, Twig [1] and iburg [11] produce code generators that modify the parse tree. Unlike APTS, which makes many passes over the parse tree, these code generator generators work by making only two passes over the parse tree. The first pass finds a set of minimal cost patterns that cover the tree. The second ....

Aho, A. V., M. Ganapathi, and S. W. K. Tjiang: 1989, `Code Generation Using Tree Matching and Dynamic Programming'. ACM Transactions on Programming Languages and Systems 11(4), 491--516.

.... is given in [Aho90] The major impetus to widespread use of pattern matching was due to a code generation scheme suggested by Hoffman and McDonnel [Hof82] The tree matching method has been used in a number of successful compiler projects [Aho90] Through use of the tree processing language Twig [Aho89] it was also applied in logic synthesis for technology mapping [Keu87] Template matching has attracted a great deal of interest in behavioral synthesis [Not91] Other pattern matching approaches in logic synthesis were predating the application of tree matching methods [Dar81, DeG85] Data ....

A.V. Aho, M. Ganapathi, S.W.K. Tjiang: "Code Generation Using Tree Matching and Dynamic Programming", ACM Trans. on Prog. Languages and Systems, Vol. 11, No. 4, pp. 491-516, 1989.

....for RISCs, during this phase an in nite number of virtual registers are assumed, which are later folded to the available amount of physical registers by the register allocator. A. Code selection The code selector uses the widespread technique of tree pattern matching with dynamic programming [23] for mapping data ow trees (DFTs) into assembly code. The basic idea in this approach is to represent the target machine instruction set in the form of a cost attributed tree grammar, and parsing each given DFT with respect to that grammar. As a result, an optimum derivation for the given cost ....

A.V. Aho, M. Ganapathi, S.W.K Tjiang, \Code Generation Using Tree Matching and Dynamic Programming," ACM Trans. on Programming Languages and Systems, vol. 11, No. 4, 1989.

....in general. This would still be a widely useful system since a broad class of problems can be described in terms of questions about paths through graphs including shortest paths and regular expression evaluation [3] solutions of dataflow equations for compiler optimizations [21] code generation [2], constraints for graphics systems [17, 23] and most problems that can be naturally posed as dynamic programs [7] On the other hand, if the performance or memory costs of using the declarative library are too large, then it will never be used. This paper examines methods for efficiently ....

Alfred V. Aho, Mahadevan Ganapathi, and Steven W.K. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, October 1989.

.... pattern signed means[int32 int8 int16 ] pattern int means[signed unsigned] pattern double means[ieee64] pattern float means[ieee32] pattern real means [ieee64 float] pattern dataformat means[octet word] pattern longint means [int32 uint32] pattern two(type t) means[2] assembles[ 2 ] pattern four(type t) means[4] assembles[ 4 ] pattern eight(type t) means[8] assembles[ 8 ] pattern scale means[two four eight] Define the address forms used in lea instructions these differ from the address forms used in other instructions as the semantics includes no ....

Aho, A.V., Ganapathi, M, TJiang S.W.K., Code Generation Using Tree Matching and Dynamic Programming, ACM Trans, Programming Languages and Systems 11, no.4, 1989, pp.491-516.

.... fficiently,anditsmanipulationiseasy(ProebstingandFischer1992) SinceitsdevelopmentbyHoffman et al. 1982) treepatternmatchingwithdynamicprogramming(DP)hasbeenacceptedasapractical techniqueforcodegeneratorsandcodegeneratorgenerators(CGGs) Pelegri Liopart1988,Ahoetal. 1989,Balachandranetal.1990,Fraseretal.1991,Proebsting1992,ProebstingandFischer1992, Fraseretal.1992,FraserandHanson1995,Gough1996) CodegeneratorswhichusestreepatternmatchingcanbeautomaticallygeneratedfrommachinespecificationbyaCGG. ....

....Fraseretal.1992,FraserandHanson1995,Gough1996) CodegeneratorswhichusestreepatternmatchingcanbeautomaticallygeneratedfrommachinespecificationbyaCGG. Theneedforautomaticgenerationofcodegeneratorshasbeencausedbytheincreasing varietyofprocessors (Ahoetal.1989).CGGsseemtobepromisingvehiclesforestimatingtherelationshipbetweenaparticulararchitectureandtheefficiencyoftargetcode, intermsofthesizeofcodeandits executiontime.AlthoughCGGsallowcodegeneratorstobedevelopedeasily,anautomaticallygenerated ....

Aho, A.V., Ganapathi, M. and Tjiang, S.W.K. (1989) Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 1, 159--75.

....the intermediate representation tree into some fixed goal tree, simultaneously outputting a sequence of machine instructions that implement the input program. These code generating transformations often include a cost metric to be minimized (to decide between several equivalent transformations) [Aho et al. 1989], Fraser et al. 1992] Language based editors are programs that understand the program being edited and can assist the programmer by checking syntax, performing intelligent formatting, and offering other language specific services. In such systems, tree transformations are useful for ....

Aho, A. V., Ganapathi, M., and Tjiang, S. W. K. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, October 1989.

....tree. The first pass is bottom up and finds a set of patterns that cover the tree with minimum cost. The second pass is top down and executes the semantic actions associated with minimumcost patterns at the nodes they matched. Code generator generators based on this model include BEG [5] Twig [2], and burg [8] BEG matchers are hard coded and mirror the tree patterns in the same way that recursivedescent parsers mirror their input grammars. They use dynamic programming at compile time to identify a minimum cost cover. Twig matchers use a table driven variant of string matching [1] that, ....

A. V. Aho, M. Ganapathi, and S. W. K. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491-- 516, Oct. 1989.

....whose nodes are labeled and in which the left to right order among siblings is significant. 1 Such trees have many applications in vision, molecular biology, programming compilation and natural language processing, including the representation of images [15] patterns [11] intermediate code [1], grammar parses [3] dictionary definitions [2,13] and secondary structures of RNA [17] They are frequently used in other disciplines as well. Many of the above applications involve comparing ordered trees. For example, in natural language processing, computational linguists store dictionary ....

....cut (cf. Section 1) is defined as calculating treedist cut(P; D) min S2Subtrees(D) ftreedist(P; Cut(D; S) g. Our goal is an algorithmic framework that can handle tree matching with and without cut, where the pattern tree P may contain both path VLDC s and umbrella VLDC s. D[9] D[3] D[2] D[1] D[7] D[5] D[6] D[8] D[4] D[2] D[3] D[1] D[9] D 0 D Fig. 6. Cutting at node D[8] 2.5 Notation We use l(i) to represent the postorder number of the leftmost leaf descendant of the subtree rooted at T [i] When T [i] is a leaf, l(i) i. The postorder number for the parent of T [i] is ....

[Article contains additional citation context not shown here]

A. V. Aho, M. Ganapathi, and S. W. K. Tjiang. Code generation using tree matching and dynamic programming. ACM Trans. Programming Languages and Systems, 11(4):491--516, Oct. 1989.

....in models and structures embodied in specialized tools with which to construct these other parts. For example, Lexical analysis generators: LEX. Parser generators: YACC[48] Front end analyses: Attributed grammars [72] Instruction selection: Abstract Interpretation[27] twig[1], Graham Glanville codegenerator generators[38, 45, 44] attributed parsing[32, 33] These tools provide a layer of abstractions to hide implementation details from the compiler writer, and to provide powerful algorithms that can be tailored to the needs of compiler writers. With the tools, ....

A. V. Aho, M. Ganapathi, and S. W. Tjiang. "Code generation using tree matching and dynamic programming". ACM Trans. on Programming Lang. and Systems 11, 4 (October 1989), 491-516.

No context found.

A. V. Aho, M. Ganapathi, and S. W. K. Tjiang, "Code generation using tree matching and dynamic programming," ACM Trans. Programming Languages and Systems, vol. 11, no. 4, pp. 491--516, 1989.

No context found.

Aho, A.V., Ganapathi, M, TJiang S.W.K., Code Generation Using Tree Matching and Dynamic Programming, ACM Trans, Programming Languages and Systems 11, no.4, 1989, pp.491-516.

No context found.

Alfred V. Aho, Mahadevan Ganapathi, and Steven W. K. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, October 1989.

No context found.

A.V. Aho M. Ganpathi and Tjiang. Code generation using tree matching and dynamic programming, October 1989.

No context found.

A.V. Aho, M. Ganapathi, S.W.K Tjiang. Code Generation Using Tree Matching and Dynamic Programming. ACM Trans. on Programming Languages and Systems, vol. 11, no. 4, 1989, pp. 491-516.

No context found.

Aho A.V. et al., Code generation using tree matching and dynamic programming. ACM transactions on programming languages and systems, vol. 11 (1989), no. 4, p. 491-516.

No context found.

Alfred V. Aho, Mahadevan Ganapathi, and Steven W. K. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, October 1989.

No context found.

Alfred V. Aho, Mahadevan Ganapathi, and Steven W. K. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, October 1989.

No context found.

Alfred V. Aho, Mahadevan Ganapathi and Steven W. K. Tjiang, "Code Generation Using Tree Matching and Dynamic Programming", ACM Transactions on Programming Languages and Systems, Vol. 11, No. 4, October 1989

No context found.

AHO,A.V.,GANAPATHI, M., AND TJIANG, S. W. K. 1989. Code generation using tree matching and dynamic programming. ACM Trans. Program. Lang. Syst. 11, 4 (Oct.), 491--516.

No context found.

A. Aho, M. Ganapathi, and S. Tjiang. Code generation using tree matching and dynamic programming. ACM Transactions on Programming Languages and Systems, 11(4):491--516, 1989.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC