Guy Blelloch and Siddhartha Chatterjee. Vcode:A data parallel intermediate language. In Proceedings of the 3rd Symposium on the Frontiers of Massively Parallel Computation, pages 471-480, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....structure. As an example, consider the implementation of Nesl [1] In the case of Nesl, the simplifier implements the flattening transformation [5, 20, 16] which transforms all nested into flat data parallelism. In CMU s implementation of the language [4] the library language is called VCODE [2] it implements a variety of operations on simple and segmented vectors, which are always uniformly distributed over the available processing elements. In fact, CMU s implementation does not generate an executable, but instead emits VCODE and interprets it at runtime by an interpreter linked to ....

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings Frontiers of Massively Parallel Computation, pages 471--480, October 1990.

.... are all or in part data parallel: there exist parallel extensions of FORTRAN, like High Performance Fortran [43] and PTRAN [3, 24] parallel extensions of C, like C [69] C [45] of Lisp, like CM List [39] and Paralation Lisp [60, 51] and applicative parallel programming languages, like NESL [8, 10, 11], Sisal [31, 62, 30] Crystal [21] Proteus [54, 34, 59] and Data parallel ML [32, 37, 38] None of these have been concerned with query constructs and their integration in such languages. In this paper we propose a new framework for parallel processing of collections. We define a highlevel ....

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Proc. of the 3rd Symposium on the Forntiers of Massively Parallel Computation, pages 471--480, University of Maryland, October 1990.

....which is associative but not commutative. fun append [ l = l j append (h: t) l = h: append t l) The function append takes two arguments (lists) and returns a list. We can use this function as an argument to fold over a list of lists. fold append [ 1,2] 3,4] 5,6] 7,8] 9,10] 11,12] [13,14], 15,6] In the above program fold with append as its function argument is applied to a list that has 8 sublists. The expected result is [1,2,3,4,5,6,7,8,9,10,11,12, 13,14,15,16] However, if the sublists are not distributed in a speci c order then incorrect results will be returned. This order ....

....and returns a list. We can use this function as an argument to fold over a list of lists. fold append [ 1,2] 3,4] 5,6] 7,8] 9,10] 11,12] 13,14] 15,6] In the above program fold with append as its function argument is applied to a list that has 8 sublists. The expected result is [1,2,3,4,5,6,7,8,9,10,11,12, 13,14,15,16]. However, if the sublists are not distributed in a speci c order then incorrect results will be returned. This order is mentioned in the 40 discussion of the parallel implementation for the fold HOF in Section 5.5.2. A possible incorrect result could be [1,2,3,4,5,6,7,8,13,14,15,16,9,10,11,12] ....

[Article contains additional citation context not shown here]

G. Blelloch and S. Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings of the 3rd Symposium Frontiers of Massively Parallel Computation, pages 471-480. IEEE, October 1990.

.... we intend initially to reduce data parallelism to the set of parallel vector operations provided by the CVL library [BCSZ90] developed by Guy Blelloch and colleagues at Carnegie Mellon as a machineindependent library used in the interpretation of the data parallel intermediate code VCODE [BC90] Likewise, we intend to reduce process parallelism to the set of procedures provided with the threads facility of Mach [BRS 85] 3.3. Refinement to SIMD We now apply these strategies to our N body program to yield execution on an SIMD architecture. Although sequence generators that ....

G. Blelloch and S. Chatterjee, "VCODE: a data-parallel intermediate language," in Proc. Frontiers 90, IEEE, 1990.

.... example, we intend intially to reduce data parallelism to the set of parallel vector operations provided by the CVL library [Ble90] developed by Guy Blelloch and colleagues at CarnegieMellon as a machine independent library used in the interpretation of the data parallel intermediate code VCODE [BC90] Likewise, we intend to reduce process parallelism to the set of procedures provided with the threads facility of Mach [BRS 85] 7. Related work There are a wide variety of programming languages that are cited as being useful for prototyping sequential computations. These languages include ....

G. Blelloch and S. Chatterjee, "VCODE: a dataparallel intermediate language," in Proceedings Frontiers 90, IEEE, 1990.

....I O paradigms. Thus, one could implement Cormen s approach by concentrating on the permutation recognition front end and relying on TPIE to provide a hardware independent back end at run time. TPIE could also be used as a back end for I O programs based on portable parallel languages like VCODE [BlC], since, as we shall see in Section 4.3, TPIE can be used to simulate many PRAM algorithms in secondary memory. This would allow code written in parallel languages such as NESL [Ble]to be made I O efficient. 3 System Design In designing and implementing the TPIE prototype, we have five goals in ....

G. E. Blelloch and S. Chatterjee, "VCODE: A Data-Parallel Intermediate Language," Proc. 3rd Symp. Frontiers of Massively Parallel Computation (1990), 471--480.

....programming. We expect this research to be most directly applicable to skeleton based languages (e.g. Darlington et al. [10] or data parallel languages, such as Nesl (Blelloch [4] and Adl (e.g. Alexander et al. [2] For example, the interpretation of Nesl into Vcode (Blelloch and Chatterjee [5]) i.e. the interpretation of nested lists into vectors) can be viewed as a shape data decomposition. These languages structure their programs with secondorder combinators, such as map and fold (e.g. Skillicorn [27] Meijer et al. [23] Cole [8] which are particularly well adapted to shape ....

G.E. Blelloch and S. Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings of the 3rd Symposium on the Frontiers of Massively Parallel Computation, pages 471--480, October 1990.

....operations. The first techniques shown there are only applicable to a subset of programs defined through a complex notion of access restricted routines. Moreover they are only applied statically and depend on the scan vector instruction set, which later became the intermediate language VCODE [BC90] used as target by the NESL compiler. The segmented vectors of VCODE, despite their name, amount to lists of lists because their allocation to processors is left unspecified by NESL programs. The semantics of NESL VCODE thus allows dynamic optimizations based on the number of processors [BH96] but ....

G. Blelloch and S. Chatterjee. VCODE: a data-parallel intermediate language. In J. JaJa, editor, 3rd IEEE Symp. Frontiers of Massively Parallel Comp., 1990.

....semantics for higher order functional programs on arrays. The following observations should convince the reader that the ADScont ADSalg framework brings new precision to data parallel language description. The choice of language primitives is an important question in data parallel programming [2, 1]. Too many or too powerful primitives hinder portability, while too few and less expressive ones limit expressive power. Now a fixed set of primitives (i.e. constants) can be formally related to different implementations by varying the algorithms for those constants and or the index space (I; L) ....

G. Blelloch and S. Chatterjee. VCODE: a data-parallel intermediate language. In J. JaJa, editor, 3rd IEEE Symp. Frontiers of Massively Parallel Comp., 1990.

.... we intend initially to reduce data parallelism to the set of parallel vector operations provided by the CVL library [BCSZ90] developed by Guy Blelloch and colleagues at Carnegie Mellon as a machineindependent library used in the interpretation of the data parallel intermediate code VCODE [BC90] Likewise, we intend to reduce process parallelism to the set of procedures provided with the threads facility of Mach [BRS 85] 3.3. Refinement to SIMD We now apply these strategies to our N body program to yield execution on an SIMD architecture. Although sequence generators that ....

G. Blelloch and S. Chatterjee, "VCODE: a data-parallel intermediate language," in Proc. Frontiers 90, IEEE, 1990.

....work on high performance array programming is devoted to parallel programming. We expect this research to be most directly applicable to skeleton based languages (e.g. DGTY95] or data parallel languages, such as Nesl [Ble92] and Adl [BAW95] For example, the interpretation of Nesl into Vcode [BC90] (i.e. the interpretation of nested lists into vectors) can be viewed as a shape data decomposition. These languages structure their programs with second order combinators, such as map and fold (e.g. Ski90, MFP91, Col93] which are particularly well adapted to shape analysis. Related to shape ....

G.E. Blelloch and S. Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings of the 3rd Symposium on the Frontiers of Massively Parallel Computation, pages 471--480, October 1990.

....MOA expressions (now strongly typed) to asynchronous machine dependent code for a network of Transputers [9] As a programming effort, this was a modest project. But the questions it raised and the answers it suggested are still extremely relevant, and are in agreement with those of other groups [2, 1]. The main conclusions of the MOA project are of two kinds. Firstly, this kind of language is very easy to use and the implementation provides excellent relative speedups with respect to the size of the network. Secondly its expressive power is limited by the set of available array primitives, yet ....

G. Blelloch and S. Chatterjee, VCODE: a data-parallel intermediate language, in Third IEEE Symposium on the Frontiers of Massively Parallel Computation, J. JaJa, ed., 1990.

....no spatial recursion. A language of this type must have a fixed set of predefined functions for data movement (map, reduce, scan etc. An advantage of DPML is to remove this restriction while retaining bounded asynchrony. See also the parallel version of FP by Walinsky and Banerjee [26] VCode [4] has complex instructions relative to the DPCAM. It s handling of parallelism through segmented instructions is different from ours. Languages like 8 1 2 [13] Alpha [20] and Crystal [7] place restrictions on programs for statically verified synchrony. For example Crystal prohibits secondorder ....

G. Blelloch and S. Chatterjee, VCODE: a data-parallel intermediate language, in Third IEEE Symposium on the Frontiers of Massively Parallel Computation, J. JaJa, ed., 1990.

....that row. A parallel function that sums the elements of a vector can then be applied in parallel to sum each row of this sparse matrix. This ability to operate efficiently on irregular data structures is one of NESL s main strengths. NESL is compiled into VCODE, a stack based intermediate language [5]. NESL s nested data structures are flattened out into segmented vectors [3] allowing a single function call to operate on the entire data structure at once. The resulting VCODE is then interpreted, with the interpreter using the CVL library to achieve portability and efficiency. Since VCODE ....

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings of Frontiers of Massively Parallel Computation, pages 471--480, October 1990.

....are orthogonal to the point of this paper and are not discussed here. The standard Nesl system consists of three layers, as shown in Figure 1 (see [7] for full details) The interactive front end of the system compiles Nesl programs into a machine independent intermediate language called Vcode [5]. The front end then invokes a Vcode interpreter, which in turn calls the low level Cvl vector library [6] The primary advantage of using an interpreted intermediate language for Nesl is that it allows users to switch transparently between running their programs on workstations (for development) ....

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Symposium on The Frontiers of Massively Parallel Computation, pages 471--480, October 1990.

.... of intermediate representation has certainly been around for a long time [3] More recently, there have been many attempts to define intermediate representations which address problems specific to compilation for parallel or distributed memory machines, or compiling data parallel languages [4]. The intermediate representation of the SUIF compiler appears to nicely solve the scalar parallel optimization dichotomy [5] It is not our intention here to provide an exhaustive survey of the field, but merely to assert that the intermediate languages discussed in this paper share some of their ....

Blelloch, G.E. and Chatterjee, S. Vcode: a data parallel intermediate language. In Proceedings of Third symposium on the Frontiers of Massively parallel computation. 1990, pages 471-80

No context found.

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings Frontiers of Massively Parallel Computation, pages47,s October 1990.

No context found.

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A data-parallel intermediate language. In Proceedings of the Symposium on the Frontiers of Massively Parallel Computation, pages 471--480. IEEE, October 1990.

No context found.

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A Data-Parallel Intermediate Language. In Proceedings of the Third Symposium on the Frontiers of Massively Parallel Computation, pages 471--480, College Park, MD, October 1990. Also available as technical report CMU-CS-90-146.

No context found.

Blelloch, G. E., and Chatterjee, S. VCODE: A data-parallel intermediate language. In Proceedings Frontiers of Massively Parallel Computation (Oct. 1990), pp. 471--480. 24

....make use of the best uniprocessor compiler technology available on the machine. 2 Language and Execution Models This section introduces the language and execution models used in the paper. After a brief overview of data parallel languages in Section 2. 1, we introduce the source language, VCODE [10], as our model of data parallelism in Section 2.2. Finally, in Section 2.3, we introduce our execution model, which corresponds to how we expect to execute the translated programs on MIMD machines. 2.1 Data parallel languages For the sake of completeness, this subsection contains a review of the ....

BLELLOCH, G. E., AND CHATTERJEE,S.VCODE: A data-parallel intermediate language. In Proceedings of the 3rd Symposium on the Frontiers of Massively Parallel Computation (College Park, MD, Oct. 1990), pp. 471--480.

....can be used as long as each primitive has an efficient parallel implementation. This excludes operations such as nonassociative scans as primitives. The set we have chosen is based on the scan vector model of computation [5] and is part of an experimental data parallel language called VCODE [6]. It is the ability to handle scans and permutations that sets our work apart from other work in this area. The code can be equivalently represented in single assignment form, which we also use (with a LISP like syntax) for ease of understanding. Using graphs as the internal representation of ....

....graph. One or the other of the subgraphs is executed, depending on the value of a scalar Boolean control input. The two subgraphs must produce the same number of results of matching types. It would be easy to add other structured looping constructs to the language. The complete language model [6] supports additional data types, segmented versions of the operations [5] and function definition call as encapsulation mechanisms. For the present, we ignore segmentation and assume that all nonrecursive function calls are inlined, and defer a discussion of these issues to Section 9. The ....

[Article contains additional citation context not shown here]

Guy E. Blelloch and Siddhartha Chatterjee. VCODE: A Data-Parallel Intermediate Language. In Proceedings of the Third Symposium on the Frontiers of Massively Parallel Computation, pages 471--480, College Park, MD, October 1990.

No context found.

Guy Blelloch and Siddhartha Chatterjee. Vcode:A data parallel intermediate language. In Proceedings of the 3rd Symposium on the Frontiers of Massively Parallel Computation, pages 471-480, 1990.

No context found.

G. Blelloch and S. Chatterjee. VCODE: a data-parallel intermediate language. In J. JaJa, editor, 3rd IEEE Symp. Frontiers of Massively Parallel Comp., 1990.

No context found.

G. E. Blelloch, S. Chatterjee, VCODE: A data-parallel intermediate language, Proceedings Frontiers of Massively Parallel Computation, pages 471-480, October 1990.

First 50 documents

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