Design-Space Exploration of Stream Programs through Semantic-Preserving Transformations

Design-Space Exploration of Stream Programs through Semantic-Preserving Transformations. de Oliveira Castro, P., Louise, S., & Barthou, D. .
abstract bibtex

Stream languages explicitly describe fork-join parallelism and pipelines, offering a powerful programming model for many-core Multi-Processor Systems on Chip (MPSoC). In an embedded resource-constrained system, adapting stream programs to fit memory requirements is particularly important. In this paper we present a design-space exploration technique to reduce the minimal memory required when running stream programs on MPSoC; this allows to target memory constrained systems and in some cases obtain better performance. Using a set of semantically preserving transformations, we explore a large number of equivalent program variants; we select the variant that minimizes a buffer evaluation metric. To cope efficiently with large program instances we propose and evaluate an heuristic for this method. We demonstrate the interest of our method on a panel of ten significant benchmarks. As an illustration, we measure the minimal memory required using a multi-core modulo scheduling. Our approach lowers considerably the minimal memory required for seven of the ten benchmarks.

@unpublished{DEOLIVEIRACASTRO:2009:HAL-00447376:1,
  HAL_ID = {hal-00447376},
  pdf = {http://hal.archives-ouvertes.fr/hal-00447376/PDF/stream_transformations_hal.pdf},
  title = { {D}esign-{S}pace {E}xploration of {S}tream {P}rograms through {S}emantic-{P}reserving {T}ransformations},
  author={de Oliveira Castro, Pablo and Louise, St\'ephane and Barthou, Denis},
  abstract = {{S}tream languages explicitly describe fork-join parallelism and pipelines, offering a powerful programming model for many-core {M}ulti-{P}rocessor {S}ystems on {C}hip ({MPS}o{C}). {I}n an embedded resource-constrained system, adapting stream programs to fit memory requirements is particularly important. {I}n this paper we present a design-space exploration technique to reduce the minimal memory required when running stream programs on {MPS}o{C}; this allows to target memory constrained systems and in some cases obtain better performance. {U}sing a set of semantically preserving transformations, we explore a large number of equivalent program variants; we select the variant that minimizes a buffer evaluation metric. {T}o cope efficiently with large program instances we propose and evaluate an heuristic for this method. {W}e demonstrate the interest of our method on a panel of ten significant benchmarks. {A}s an illustration, we measure the minimal memory required using a multi-core modulo scheduling. {O}ur approach lowers considerably the minimal memory required for seven of the ten benchmarks.},
}

Downloads: 0

{"_id":"RwzqRiKoai3rNLhqX","bibbaseid":"deoliveiracastro-louise-barthou-designspaceexplorationofstreamprogramsthroughsemanticpreservingtransformations","downloads":0,"creationDate":"2016-10-17T01:52:58.128Z","title":"Design-Space Exploration of Stream Programs through Semantic-Preserving Transformations","author_short":["de Oliveira Castro, P.","Louise, S.","Barthou, D."],"year":null,"bibtype":"unpublished","biburl":"https://raw.githubusercontent.com/pablooliveira/bibjekyll/master/example/more.bib","bibdata":{"bibtype":"unpublished","type":"unpublished","hal_id":"hal-00447376","pdf":"http://hal.archives-ouvertes.fr/hal-00447376/PDF/stream_transformations_hal.pdf","title":"Design-Space Exploration of Stream Programs through Semantic-Preserving Transformations","author":[{"propositions":["de"],"lastnames":["Oliveira","Castro"],"firstnames":["Pablo"],"suffixes":[]},{"propositions":[],"lastnames":["Louise"],"firstnames":["Stéphane"],"suffixes":[]},{"propositions":[],"lastnames":["Barthou"],"firstnames":["Denis"],"suffixes":[]}],"abstract":"Stream languages explicitly describe fork-join parallelism and pipelines, offering a powerful programming model for many-core Multi-Processor Systems on Chip (MPSoC). In an embedded resource-constrained system, adapting stream programs to fit memory requirements is particularly important. In this paper we present a design-space exploration technique to reduce the minimal memory required when running stream programs on MPSoC; this allows to target memory constrained systems and in some cases obtain better performance. Using a set of semantically preserving transformations, we explore a large number of equivalent program variants; we select the variant that minimizes a buffer evaluation metric. To cope efficiently with large program instances we propose and evaluate an heuristic for this method. We demonstrate the interest of our method on a panel of ten significant benchmarks. As an illustration, we measure the minimal memory required using a multi-core modulo scheduling. Our approach lowers considerably the minimal memory required for seven of the ten benchmarks.","bibtex":"@unpublished{DEOLIVEIRACASTRO:2009:HAL-00447376:1,\n HAL_ID = {hal-00447376},\n pdf = {http://hal.archives-ouvertes.fr/hal-00447376/PDF/stream_transformations_hal.pdf},\n title = { {D}esign-{S}pace {E}xploration of {S}tream {P}rograms through {S}emantic-{P}reserving {T}ransformations},\n author={de Oliveira Castro, Pablo and Louise, St\\'ephane and Barthou, Denis},\n abstract = {{S}tream languages explicitly describe fork-join parallelism and pipelines, offering a powerful programming model for many-core {M}ulti-{P}rocessor {S}ystems on {C}hip ({MPS}o{C}). {I}n an embedded resource-constrained system, adapting stream programs to fit memory requirements is particularly important. {I}n this paper we present a design-space exploration technique to reduce the minimal memory required when running stream programs on {MPS}o{C}; this allows to target memory constrained systems and in some cases obtain better performance. {U}sing a set of semantically preserving transformations, we explore a large number of equivalent program variants; we select the variant that minimizes a buffer evaluation metric. {T}o cope efficiently with large program instances we propose and evaluate an heuristic for this method. {W}e demonstrate the interest of our method on a panel of ten significant benchmarks. {A}s an illustration, we measure the minimal memory required using a multi-core modulo scheduling. {O}ur approach lowers considerably the minimal memory required for seven of the ten benchmarks.},\n}\n\n","author_short":["de Oliveira Castro, P.","Louise, S.","Barthou, D."],"key":"DEOLIVEIRACASTRO:2009:HAL-00447376:1","id":"DEOLIVEIRACASTRO:2009:HAL-00447376:1","bibbaseid":"deoliveiracastro-louise-barthou-designspaceexplorationofstreamprogramsthroughsemanticpreservingtransformations","role":"author","urls":{},"downloads":0},"search_terms":["design","space","exploration","stream","programs","through","semantic","preserving","transformations","de oliveira castro","louise","barthou"],"keywords":[],"authorIDs":["58042efac2cb419a33000018"],"dataSources":["f5665z45ib69GeBRW"]}