J.L. Baer and D.P. Bovet, Compilation of Arithmetic Expressions for Parallel Computations, Proceedings of IFIP Congress, North-Holland, Amsterdam, pp. 340-346, 1968.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....different from . Formally, the topmost cluster of a BDAG is obtained by the algorithm CLUSTER in Fig. 6. Given a cluster, a minimum delay tree can be built by combining the elements of the cluster in an appropriate way, trying the tallest subtrees to be closer to the root. Baer and Boven [19] proposed an algorithm to build such a tree. It is an iterative algorithm that maintains all elements of the cluster in a priority This distinction is also maintained in the package by properly keeping track of the complemented edges found in the paths. Fig. 6. Algorithm for minimum delay ....

J. Baer and D. Bovet, "Compilation of arithmetic expressions for parallel computations," in Proc. IFIP Congress North-Holland, The Netherlands, 1968, pp. 340--346.

....different from . Formally, the topmost cluster of a tree is obtained by the algorithm CLUSTER in Figure 3. Given a cluster, a minimum delay tree can be built by combining the elements of the cluster in an appropriate way, trying the tallest sub trees to be closer to the root. Baer and Boven [1] proposed an algorithm to build such a tree. It is an iterative algorithm that maintains all elements of the cluster in a priority queue ordered by the height of the 2 c d e f g h j k i j k c d e f g h b b a Figure 4. Application ....

J.L. Baer and D.P. Bovet. Compilation of arithmetic expressions for parallel computations. In Proc. IFIP Congress, pages 340--346, Amsterdam, 1968. North-Holland.

....a system designer can concentrate on more important higher level trade offs. This research was sponsored by the ESPRIT2260 ( SPRITE ) project of the EC. Transformations are often used for optimisation purposes. In parallel compilers, transformations are used to exploit parallelism in flow graphs [1, 2]. In high level synthesis, transformations are mainly used to optimise throughput [3, 4, 5] Recently, transformations are also used in power optimisation [6] where the steering is limited to a generic global optimisation technique on a subset of the possible transformations. Also using ....

....collectR. Under the assumption that v = defn(inp(u; 1) and w = defn(inp(u; 2) are valid, collectR(u) u 2 O j Iu j= 2 j U inp(u;1) j= 1 j U inp(u;2) j= 1 v; w 2 O type(v) type(w) right distr(v; u) j I v j= 2 j Iw j= 2 inp(v; 2) inp(w; 2) a b c d v w u a b c Y X Y X Y d [2] [2] 2] 2] 2] The elementary transformations are characterised by strict preconditions, such as the last precondition of collectR which does not allow the common signal c to be located at the first input of either operation. These strict preconditions make the elementary transformations ....

[Article contains additional citation context not shown here]

J.L. Bear, D.P. Bovet, "Compilation of Arithmetic Expressions for Parallel Computations," Information Processing 68, North-Holland Publishing Company, Amsterdam, the Netherlands, 1969, pp. 340-346.

.... scheduling [6] Techniques to eliminated dependences between unrolled iterations were implemented in the Bulldog, Multiflow trace and Cydra 5 compilers [7] 8] 9] Several height reduction techniques that reduce the length of dependence chains in arithmetic calculations have been proposed [10] [11] The transformations discussed in this paper utilize the concepts presented in these previous studies. The use of code scheduling schemes that can take advantage of eliminated dependences is necessary to get performance improvement from these transformations. Many previous studies have ....

....number of temporary registers required. For superscalar and VLIW processors, however, these methods often limit performance by restricting parallel computation of individual components of an arithmetic expression. Tree height reduction exposes ILP in the computation of an arithmetic expression [10] [11] Tree height reduction first constructs an expression tree to represent the arithmetic expression. The tree is then balanced to reduce the height. The height represents the number of cycles to compute the expression using a specific processor model. The compiler used in this study uses a ....

[Article contains additional citation context not shown here]

J. L. Baer and D. P. Bovet, "Compilation of arithmetic expressions for parallel computations," in Proceedings of IFIP Congress, pp. 34--46, 1968.

....the use of a meta transformation to guide transformation application as possibilities arise. Results observed on several image and video benchmarks demonstrate that transformation integration increases performance through better resource utilization. 1 Introduction Tree height reduction (THR) [1, 12] is a well known technique for reducing the critical path length and increasing the parallelism of expressions and or recurrences through the introduction of redundant computation. THR has been applied to the synthesis of DSP applications [8, 11, 13, 19] and will continue to play an important role ....

....are formulated for memory analysis. In [9] we present a technique for removing memory operations that are redundant over loop execution. Our technique uses memory anti aliasing theory so as to detect redundancy in a general manner. Tree Height Reduction Tree height reduction was first studied [1, 12] as a method for reducing critical dependency chains to increase parallelism and was later extended in [3] Various synthesis systems [8, 19] include support for THR, but do not specifically factor resource availability into the process, or do so in an exhaustive manner [11] In [13] an ....

J. L. Baer and D. P. Bovet. Compilation of Arithmetic Expressions for Parallel Computations. Proc. of IFIP Congress, pages 34--46, 1968.

No context found.

J.L. Baer and D.P. Bovet, Compilation of Arithmetic Expressions for Parallel Computations, Proceedings of IFIP Congress, North-Holland, Amsterdam, pp. 340-346, 1968.

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