The Experts below are selected from a list of 4026 Experts worldwide ranked by ideXlab platform
Eric Rotenberg - One of the best experts on this subject based on the ideXlab platform.
-
Slipstream processors revisited exploiting branch sets
International Symposium on Computer Architecture, 2020Co-Authors: Vinesh Srinivasan, Rangeen Basu Roy Chowdhury, Eric RotenbergAbstract:Delinquent branches and loads remain key performance limiters in some applications. One approach to mitigate them is pre-execution. Broadly, there are two classes of pre-execution: one class repeatedly forks small helper threads, each targeting an individual dynamic instance of a delinquent branch or load; the other class begins with two redundant threads in a leader-follower arrangement, and speculatively reduces the leading thread. The objective of this paper is to design a new pre-execution microarchitecture that meets four criteria: (i) retains the simpler coordination of a leader-follower microarchitecture, (ii) is fully automated with just hardware, (iii) targets both branches and loads, (iv) and is effective. We review prior preexecution proposals and show that none of them meet all four criteria. We develop Slipstream 2.0 to meet all four criteria. The key innovation in the space of leader-follower architectures is to remove the forward control-flow slices of delinquent branches and loads, from the leading thread. This innovation overcomes key limitations in the only other hardware-only leader-follower prior works: Slipstream and Dual Core Execution (DCE). Slipstream removes backward slices of confident branches to pre-execute unconfident branches, which is ineffective in phases dominated by unconfident branches when branch pre-execution is most needed. DCE is very effective at tolerating cache-missed loads, unless their dependent branches are mispredicted. Removing forward control-flow slices of delinquent branches and delinquent loads enables two firsts, respectively: (1) leader-follower-style branch pre-execution without relying on confident instruction removal, and (2) tolerance of cache-missed loads that feed mispredicted branches. For SPEC 2006/2017 SimPoints wherein Slipstream 2.0 is auto-enabled, it achieves geomean speedups of 67%, 60%, and 12%, over baseline (one core), Slipstream, and DCE.
-
ISCA - Slipstream Processors Revisited: Exploiting Branch Sets
2020 ACM IEEE 47th Annual International Symposium on Computer Architecture (ISCA), 2020Co-Authors: Vinesh Srinivasan, Rangeen Basu Roy Chowdhury, Eric RotenbergAbstract:Delinquent branches and loads remain key performance limiters in some applications. One approach to mitigate them is pre-execution. Broadly, there are two classes of pre-execution: one class repeatedly forks small helper threads, each targeting an individual dynamic instance of a delinquent branch or load; the other class begins with two redundant threads in a leader-follower arrangement, and speculatively reduces the leading thread. The objective of this paper is to design a new pre-execution microarchitecture that meets four criteria: (i) retains the simpler coordination of a leader-follower microarchitecture, (ii) is fully automated with just hardware, (iii) targets both branches and loads, (iv) and is effective. We review prior preexecution proposals and show that none of them meet all four criteria. We develop Slipstream 2.0 to meet all four criteria. The key innovation in the space of leader-follower architectures is to remove the forward control-flow slices of delinquent branches and loads, from the leading thread. This innovation overcomes key limitations in the only other hardware-only leader-follower prior works: Slipstream and Dual Core Execution (DCE). Slipstream removes backward slices of confident branches to pre-execute unconfident branches, which is ineffective in phases dominated by unconfident branches when branch pre-execution is most needed. DCE is very effective at tolerating cache-missed loads, unless their dependent branches are mispredicted. Removing forward control-flow slices of delinquent branches and delinquent loads enables two firsts, respectively: (1) leader-follower-style branch pre-execution without relying on confident instruction removal, and (2) tolerance of cache-missed loads that feed mispredicted branches. For SPEC 2006/2017 SimPoints wherein Slipstream 2.0 is auto-enabled, it achieves geomean speedups of 67%, 60%, and 12%, over baseline (one core), Slipstream, and DCE.
-
Slipstream execution mode for cmp based multiprocessors
High-Performance Computer Architecture, 2003Co-Authors: K.z. Ibrahim, Gregory T. Byrd, Eric RotenbergAbstract:Scalability of applications on distributed shared-memory (DSM) multiprocessors is limited by communication overheads. At some point, using more processors to increase parallelism yields diminishing returns or even degrades performance. When increasing concurrency is futile, we propose an additional mode of execution, called Slipstream mode, that instead enlists extra processors to assist parallel tasks by reducing perceived overheads. We consider DSM multiprocessors built from dual-processor chip multiprocessor (CMP) nodes with shared L2 cache. A task is allocated on one processor of each CMP node. The other processor of each node executes a reduced version of the same task. The reduced version skips shared-memory stores and synchronization, running ahead of the true task. Even with the skipped operations, the reduced task makes accurate forward progress and generates an accurate reference stream, because branches and addresses depend primarily on private data. Slipstream execution mode yields two benefits. First, the reduced task prefetches data on behalf of the true task. Second, reduced tasks provide a detailed picture of future reference behavior, enabling a number of optimizations aimed at accelerating coherence events, e.g., self-invalidation. For multiprocessor systems with up to 16 CMP nodes, Slipstream mode outperforms running one or two conventional tasks per CMP in 7 out of 9 parallel scientific benchmarks. Slipstream mode is 12-19% faster with prefetching only and up to 29% faster with self-invalidation enabled.
-
Slipstream processors
2003Co-Authors: Zachary Robert Purser, Eric RotenbergAbstract:Processors execute a program's full dynamic instruction stream to arrive at its final output, yet there exist shorter instruction streams that produce the same overall effect. This thesis proposes creating a shorter but otherwise equivalent version of the original program by removing ineffectual computation and computation related to highly-predictable control flow. The shortened program is run concurrently with and slightly ahead of a full copy of the program on a chip multiprocessor (CMP) or simultaneous multithreading (SMT) processor. The leading program passes all of its control-flow and data-flow outcomes to the trailing program for checking. This redundant program arrangement provides two key benefits. (1) Improved single program performance. The leading program is sped up because it retires fewer instructions. Although the number of retired instructions is not reduced in the trailing program, it fetches and executes instructions more efficiently by virtue of having near-oracle branch and value predictions from the leading program. Thus, the trailing program is also sped up in the wake or “Slipstream” of the leading program, at the same time validating the speculative leading program and redirecting it as needed. Slipstream execution using two processors of a CMP substrate outperforms conventional non-redundant execution using only one of the processors. Likewise, given a sufficiently reduced leading program, Slipstream execution using two contexts of an SMT substrate outperforms conventional non-redundant execution using only one of the contexts. (2) Fault tolerance. The shorter program is a subset of the full program and this partial redundancy is exploited for detecting and recovering from transient hardware faults. This does not require any additional hardware support, since the same mechanisms used to detect and recover from misspeculation in the leading program apply equally well to transient fault detection and recovery. In fact, there is no way to distinguish between misspeculation and faults. The broader rationale for Slipstream is extending, not replacing, the capabilities of CMP/SMT processors, providing additional modes of execution. This thesis demonstrates the feasibility and benefits of the Slipstream execution model.
-
HPCA - Slipstream execution mode for CMP-based multiprocessors
The Ninth International Symposium on High-Performance Computer Architecture 2003. HPCA-9 2003. Proceedings., 1Co-Authors: K.z. Ibrahim, Gregory T. Byrd, Eric RotenbergAbstract:Scalability of applications on distributed shared-memory (DSM) multiprocessors is limited by communication overheads. At some point, using more processors to increase parallelism yields diminishing returns or even degrades performance. When increasing concurrency is futile, we propose an additional mode of execution, called Slipstream mode, that instead enlists extra processors to assist parallel tasks by reducing perceived overheads. We consider DSM multiprocessors built from dual-processor chip multiprocessor (CMP) nodes with shared L2 cache. A task is allocated on one processor of each CMP node. The other processor of each node executes a reduced version of the same task. The reduced version skips shared-memory stores and synchronization, running ahead of the true task. Even with the skipped operations, the reduced task makes accurate forward progress and generates an accurate reference stream, because branches and addresses depend primarily on private data. Slipstream execution mode yields two benefits. First, the reduced task prefetches data on behalf of the true task. Second, reduced tasks provide a detailed picture of future reference behavior, enabling a number of optimizations aimed at accelerating coherence events, e.g., self-invalidation. For multiprocessor systems with up to 16 CMP nodes, Slipstream mode outperforms running one or two conventional tasks per CMP in 7 out of 9 parallel scientific benchmarks. Slipstream mode is 12-19% faster with prefetching only and up to 29% faster with self-invalidation enabled.
John Sheridan - One of the best experts on this subject based on the ideXlab platform.
-
The impact of rails on high-speed train Slipstream and wake
Journal of Wind Engineering and Industrial Aerodynamics, 2020Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Slipstream is the induced movement of air as a high-speed train (HST) passes. Previous studies have shown that the development of Slipstream is highly geometry dependent, including both the geometries of the trains and nearby objects. Much effort has been channelled into reducing Slipstream through optimisation of the train geometry; however, the impact of the rails on train aerodynamics remains largely unexplored. This study analyses the effect of rails on HST Slipstream characteristics by systematically comparing the wakes for two geometric configurations incorporating: No Rails (NR) and With Rails (WR). The train model remains identical in both configurations, with the only difference being whether rails are included. This study highlights the potential effects of rails on HST Slipstream characteristics, and reveals the underlying mechanism of how the rails shape the wake flow structures. By examining both mean and time-dependent flow features, the simulations show that the rails can significantly alter Slipstream characteristics, especially the downstream evolution of the wake, effectively by reducing the lateral movement of the mean streamwise vortex structures, despite the relatively small length scale of the rail cross-section. Perhaps surprisingly, the Slipstream measured at the standard distance from the train centerplane, is found be significantly reduced.
-
the effect of bogies on high speed train Slipstream and wake
Journal of Fluids and Structures, 2018Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Slipstream, which is the induced air movement generated by a high-speed train (HST) as it passes, is both a safety hazard to commuters and trackside workers, and can cause damage to infrastructure along track lines. Bogies have been shown to exert a strong influence on HST aerodynamics by altering the underbody flow, and their effects have been extensively studied previously from the perspectives of ballast flight and drag reduction. In contrast, the effect of bogies on Slipstream characteristics, and especially on the structure of the wake, is less well understood. This study explicitly investigates the effect of bogies on HST Slipstream characteristics based on two generic train configurations: a Simplified Train Model (M1) with the bogies covered by a flat surface, and a Full-featured Train Model (M2) with simplified bogie-sets. The bogie effects are revealed through a systematic comparison of flow structures, Slipstream characteristics and aerodynamic forces between these configurations. Remarkably, this study discovers that the generation of the strong spanwise oscillation of the wake, observed especially in the presence of bogies, can be interpreted as due to seeding and amplification of a natural instability of the time-mean pair of counter-rotating vortices behind the tail, rather than through direct side-to-side vortex shedding from the bogie geometry. This paper also documents how the altered wake flow affects Slipstream characteristics through statistical and gust analyses, and the effect of the bogies on aerodynamic loading by comparing the train surface pressure distributions between the configurations.
-
The effect of the ground condition on high-speed train Slipstream
Journal of Wind Engineering and Industrial Aerodynamics, 2018Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Understanding the induced movement of air as a high-speed train passes (Slipstream) is important for commuter and track-side worker safety. Slipstream is affected by the movement of the train relative to the ground, but this is difficult to include in wind-tunnel tests. Using simulations based on the Improved Delayed Detached Eddy Simulation model, this study investigates the effect of relative ground motion on Slipstream for three different ground/wheel configurations: a stationary ground with stationary wheels, a moving ground with stationary wheels, and a moving ground with rotating wheels. By examining the interaction between the train-induced flow structure and ground boundary layer, this study identifies two ways that the ground boundary layer changes Slipstream: through directly altering the high Slipstream velocity region due to the ground boundary-layer development, and through indirect widening of the wake by deformation of the trailing vortices. The altered aerodynamic loading on a train due to relative ground motion is visualised through the surface pressure distribution, allowing the resultant impact on drag and lift to be assessed. For wheel rotation, it is concluded that its effect is mainly restricted to be within the bogie regions, with limited influence on the wake behind the train.
-
the performance of different turbulence models urans sas and des for predicting high speed train Slipstream
Journal of Wind Engineering and Industrial Aerodynamics, 2017Co-Authors: Shibo Wang, James Robin Bell, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract The air movement induced by a high-speed train (HST) as it passes, the Slipstream, is a safety hazard to commuters and trackside workers, and can cause damage to infrastructure along track lines. Because of its importance, many numerical studies have been undertaken to investigate this phenomenon. However, to the authors' knowledge, a systematic comparison of the accuracy of different turbulence models applied to the prediction of Slipstream has not yet been conducted. This study investigates and evaluates the performance of three widely used turbulence models: URANS, SAS and DES, to predict the Slipstream of a full-featured generic train model, and the results are compared with wind-tunnel experimental data to determine the fidelity of the models. Specifically, this research aims to determine the suitability of different turbulence modelling approaches, involving significantly different computational resources, for modelling different aspects of Slipstream.
-
A wind-tunnel methodology for assessing the Slipstream of high-speed trains
Journal of Wind Engineering and Industrial Aerodynamics, 2017Co-Authors: James Robin Bell, David Burton, Astrid H Herbst, Mark C Thompson, John SheridanAbstract:Abstract A 1/10th-scale wind-tunnel technique for assessing the Slipstream of high-speed trains (HST) is assessed through comparison to full-scale field and 1/25th-scale moving-model experimental results of an Inter-City Express 3 (ICE3) – a HST in operation throughout Europe and Asia – with the view of applying the wind-tunnel methodology for checking regulatory compliance in the design phase of a HST. The effect of the experimental limitations inherent in a wind-tunnel Slipstream methodology: the presence of a stationary floor, reduced length of the model, limited test-section size and ground-fixed frame-of-reference are also investigated. Subsequently, recommendations for the use of wind-tunnel and moving model methodologies for assessing the Slipstream of prototype HSTs with an applied, industrial aerodynamics focus, are made.
Mark C Thompson - One of the best experts on this subject based on the ideXlab platform.
-
The impact of rails on high-speed train Slipstream and wake
Journal of Wind Engineering and Industrial Aerodynamics, 2020Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Slipstream is the induced movement of air as a high-speed train (HST) passes. Previous studies have shown that the development of Slipstream is highly geometry dependent, including both the geometries of the trains and nearby objects. Much effort has been channelled into reducing Slipstream through optimisation of the train geometry; however, the impact of the rails on train aerodynamics remains largely unexplored. This study analyses the effect of rails on HST Slipstream characteristics by systematically comparing the wakes for two geometric configurations incorporating: No Rails (NR) and With Rails (WR). The train model remains identical in both configurations, with the only difference being whether rails are included. This study highlights the potential effects of rails on HST Slipstream characteristics, and reveals the underlying mechanism of how the rails shape the wake flow structures. By examining both mean and time-dependent flow features, the simulations show that the rails can significantly alter Slipstream characteristics, especially the downstream evolution of the wake, effectively by reducing the lateral movement of the mean streamwise vortex structures, despite the relatively small length scale of the rail cross-section. Perhaps surprisingly, the Slipstream measured at the standard distance from the train centerplane, is found be significantly reduced.
-
the effect of bogies on high speed train Slipstream and wake
Journal of Fluids and Structures, 2018Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Slipstream, which is the induced air movement generated by a high-speed train (HST) as it passes, is both a safety hazard to commuters and trackside workers, and can cause damage to infrastructure along track lines. Bogies have been shown to exert a strong influence on HST aerodynamics by altering the underbody flow, and their effects have been extensively studied previously from the perspectives of ballast flight and drag reduction. In contrast, the effect of bogies on Slipstream characteristics, and especially on the structure of the wake, is less well understood. This study explicitly investigates the effect of bogies on HST Slipstream characteristics based on two generic train configurations: a Simplified Train Model (M1) with the bogies covered by a flat surface, and a Full-featured Train Model (M2) with simplified bogie-sets. The bogie effects are revealed through a systematic comparison of flow structures, Slipstream characteristics and aerodynamic forces between these configurations. Remarkably, this study discovers that the generation of the strong spanwise oscillation of the wake, observed especially in the presence of bogies, can be interpreted as due to seeding and amplification of a natural instability of the time-mean pair of counter-rotating vortices behind the tail, rather than through direct side-to-side vortex shedding from the bogie geometry. This paper also documents how the altered wake flow affects Slipstream characteristics through statistical and gust analyses, and the effect of the bogies on aerodynamic loading by comparing the train surface pressure distributions between the configurations.
-
The effect of the ground condition on high-speed train Slipstream
Journal of Wind Engineering and Industrial Aerodynamics, 2018Co-Authors: Shibo Wang, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract Understanding the induced movement of air as a high-speed train passes (Slipstream) is important for commuter and track-side worker safety. Slipstream is affected by the movement of the train relative to the ground, but this is difficult to include in wind-tunnel tests. Using simulations based on the Improved Delayed Detached Eddy Simulation model, this study investigates the effect of relative ground motion on Slipstream for three different ground/wheel configurations: a stationary ground with stationary wheels, a moving ground with stationary wheels, and a moving ground with rotating wheels. By examining the interaction between the train-induced flow structure and ground boundary layer, this study identifies two ways that the ground boundary layer changes Slipstream: through directly altering the high Slipstream velocity region due to the ground boundary-layer development, and through indirect widening of the wake by deformation of the trailing vortices. The altered aerodynamic loading on a train due to relative ground motion is visualised through the surface pressure distribution, allowing the resultant impact on drag and lift to be assessed. For wheel rotation, it is concluded that its effect is mainly restricted to be within the bogie regions, with limited influence on the wake behind the train.
-
the performance of different turbulence models urans sas and des for predicting high speed train Slipstream
Journal of Wind Engineering and Industrial Aerodynamics, 2017Co-Authors: Shibo Wang, James Robin Bell, David Burton, Astrid H Herbst, John Sheridan, Mark C ThompsonAbstract:Abstract The air movement induced by a high-speed train (HST) as it passes, the Slipstream, is a safety hazard to commuters and trackside workers, and can cause damage to infrastructure along track lines. Because of its importance, many numerical studies have been undertaken to investigate this phenomenon. However, to the authors' knowledge, a systematic comparison of the accuracy of different turbulence models applied to the prediction of Slipstream has not yet been conducted. This study investigates and evaluates the performance of three widely used turbulence models: URANS, SAS and DES, to predict the Slipstream of a full-featured generic train model, and the results are compared with wind-tunnel experimental data to determine the fidelity of the models. Specifically, this research aims to determine the suitability of different turbulence modelling approaches, involving significantly different computational resources, for modelling different aspects of Slipstream.
-
A wind-tunnel methodology for assessing the Slipstream of high-speed trains
Journal of Wind Engineering and Industrial Aerodynamics, 2017Co-Authors: James Robin Bell, David Burton, Astrid H Herbst, Mark C Thompson, John SheridanAbstract:Abstract A 1/10th-scale wind-tunnel technique for assessing the Slipstream of high-speed trains (HST) is assessed through comparison to full-scale field and 1/25th-scale moving-model experimental results of an Inter-City Express 3 (ICE3) – a HST in operation throughout Europe and Asia – with the view of applying the wind-tunnel methodology for checking regulatory compliance in the design phase of a HST. The effect of the experimental limitations inherent in a wind-tunnel Slipstream methodology: the presence of a stationary floor, reduced length of the model, limited test-section size and ground-fixed frame-of-reference are also investigated. Subsequently, recommendations for the use of wind-tunnel and moving model methodologies for assessing the Slipstream of prototype HSTs with an applied, industrial aerodynamics focus, are made.
Gregory T. Byrd - One of the best experts on this subject based on the ideXlab platform.
-
extending openmp to support Slipstream execution mode
International Parallel and Distributed Processing Symposium, 2003Co-Authors: K.z. Ibrahim, Gregory T. ByrdAbstract:OpenMP has emerged as a widely accepted standard for writing shared memory programs. Hardware-specific extensions such as data placement are usually needed to improve the scalability of applications based on this standard. This paper investigates the implementation of an OpenMP compiler that supports Slipstream execution mode, a new optimization mechanism for CMP-based distributed shared memory multiprocessors. Slipstream mode uses additional processors to reduce communication overhead, rather than to increase parallelism. We discuss how each OpenMP construct can be implemented to take advantage of Slipstream mode, and we present a minor extension that allows runtime or compile-time control of Slipstream execution. We also investigate the interaction between Slipstream mechanisms and OpenMP scheduling. Our implementation supports both static and dynamic scheduling in Slipstream mode. We extended the Omni OpenMP compiler to generate binaries that support Slipstream mode, and we show the performance of Slipstream-enabled codes using OpenMP codes from the NAS Parallel Benchmark suite, running on the SimOS simulator. Our extension to OpenMP allowed the benchmarks to achieve an average performance improvement of 14% with static scheduling. For dynamic scheduling the performance improvement is 12% on average.
-
Slipstream execution mode for cmp based multiprocessors
High-Performance Computer Architecture, 2003Co-Authors: K.z. Ibrahim, Gregory T. Byrd, Eric RotenbergAbstract:Scalability of applications on distributed shared-memory (DSM) multiprocessors is limited by communication overheads. At some point, using more processors to increase parallelism yields diminishing returns or even degrades performance. When increasing concurrency is futile, we propose an additional mode of execution, called Slipstream mode, that instead enlists extra processors to assist parallel tasks by reducing perceived overheads. We consider DSM multiprocessors built from dual-processor chip multiprocessor (CMP) nodes with shared L2 cache. A task is allocated on one processor of each CMP node. The other processor of each node executes a reduced version of the same task. The reduced version skips shared-memory stores and synchronization, running ahead of the true task. Even with the skipped operations, the reduced task makes accurate forward progress and generates an accurate reference stream, because branches and addresses depend primarily on private data. Slipstream execution mode yields two benefits. First, the reduced task prefetches data on behalf of the true task. Second, reduced tasks provide a detailed picture of future reference behavior, enabling a number of optimizations aimed at accelerating coherence events, e.g., self-invalidation. For multiprocessor systems with up to 16 CMP nodes, Slipstream mode outperforms running one or two conventional tasks per CMP in 7 out of 9 parallel scientific benchmarks. Slipstream mode is 12-19% faster with prefetching only and up to 29% faster with self-invalidation enabled.
-
Slipstream execution mode for cmp-based shared memory systems
2003Co-Authors: K.z. Ibrahim, Gregory T. ByrdAbstract:Scalability of applications on distributed shared-memory (DSM) multiprocessors is limited by communication and synchronization overheads. At some point, using more processors to increase parallelism yields diminishing returns or even degrades performance. When increasing concurrency is futile, we propose an additional mode of execution, called Slipstream mode , that instead enlists extra processors to assist parallel tasks by reducing perceived overheads. We consider DSM multiprocessors built from dual-processor chip multiprocessor (CMP) nodes (e.g., IBM Power-4 CMP) with shared L2 cache. A parallel task is allocated on one processor of each CMP node. The other processor of each node executes a reduced version of the same task. The reduced version skips shared-memory stores and synchronization, allowing it to run ahead of the true task. Even with the skipped operations, the reduced task makes accurate forward progress and generates an accurate reference stream, because branches and addresses depend primarily on private data. Slipstream execution mode yields multiple benefits. First, the reduced task prefetches data on behalf of the true task. Second, reduced tasks provide a detailed picture of future reference behavior, enabling a number of optimizations aimed at accelerating coherence events. We investigate a well-known optimization, self-invalidation. We also investigate providing confidence mechanism for speculation after barrier synchronization. We investigate the implementation of an OpenMP compiler that supports Slipstream execution mode. We discuss how each OpenMP construct can be implemented to take advantage of Slipstream mode, and we present a minor extension that allows runtime or compile-time control of Slipstream execution. We also investigate the interaction between Slipstream mechanisms and OpenMP scheduling. Our implementation supports both static and dynamic scheduling in Slipstream mode. For multiprocessor systems with up to 16 CMP nodes, Slipstream mode is 12–19% faster with prefetching only. With self-invalidation also enabled, performance is improved by as much as 29%. We extended Slipstream mode to provide a confidence mechanism for barrier speculation. This mechanism identifies dependencies and tries to avoid dependency violations that lead to misspeculations (and subsequently rollbacks). Rollbacks are reduced by up to 95% and the improvement in performance is up to 13%. Slipstream execution mode enables a wide range of optimizations based on an accurate future image of the program behavior. It does not require custom auxiliary hardware tables used by history-based predictors.
-
IPDPS - Extending OpenMP to support Slipstream execution mode
Proceedings International Parallel and Distributed Processing Symposium, 1Co-Authors: K.z. Ibrahim, Gregory T. ByrdAbstract:OpenMP has emerged as a widely accepted standard for writing shared memory programs. Hardware-specific extensions such as data placement are usually needed to improve the scalability of applications based on this standard. This paper investigates the implementation of an OpenMP compiler that supports Slipstream execution mode, a new optimization mechanism for CMP-based distributed shared memory multiprocessors. Slipstream mode uses additional processors to reduce communication overhead, rather than to increase parallelism. We discuss how each OpenMP construct can be implemented to take advantage of Slipstream mode, and we present a minor extension that allows runtime or compile-time control of Slipstream execution. We also investigate the interaction between Slipstream mechanisms and OpenMP scheduling. Our implementation supports both static and dynamic scheduling in Slipstream mode. We extended the Omni OpenMP compiler to generate binaries that support Slipstream mode, and we show the performance of Slipstream-enabled codes using OpenMP codes from the NAS Parallel Benchmark suite, running on the SimOS simulator. Our extension to OpenMP allowed the benchmarks to achieve an average performance improvement of 14% with static scheduling. For dynamic scheduling the performance improvement is 12% on average.
-
HPCA - Slipstream execution mode for CMP-based multiprocessors
The Ninth International Symposium on High-Performance Computer Architecture 2003. HPCA-9 2003. Proceedings., 1Co-Authors: K.z. Ibrahim, Gregory T. Byrd, Eric RotenbergAbstract:Scalability of applications on distributed shared-memory (DSM) multiprocessors is limited by communication overheads. At some point, using more processors to increase parallelism yields diminishing returns or even degrades performance. When increasing concurrency is futile, we propose an additional mode of execution, called Slipstream mode, that instead enlists extra processors to assist parallel tasks by reducing perceived overheads. We consider DSM multiprocessors built from dual-processor chip multiprocessor (CMP) nodes with shared L2 cache. A task is allocated on one processor of each CMP node. The other processor of each node executes a reduced version of the same task. The reduced version skips shared-memory stores and synchronization, running ahead of the true task. Even with the skipped operations, the reduced task makes accurate forward progress and generates an accurate reference stream, because branches and addresses depend primarily on private data. Slipstream execution mode yields two benefits. First, the reduced task prefetches data on behalf of the true task. Second, reduced tasks provide a detailed picture of future reference behavior, enabling a number of optimizations aimed at accelerating coherence events, e.g., self-invalidation. For multiprocessor systems with up to 16 CMP nodes, Slipstream mode outperforms running one or two conventional tasks per CMP in 7 out of 9 parallel scientific benchmarks. Slipstream mode is 12-19% faster with prefetching only and up to 29% faster with self-invalidation enabled.
Meyer Nahon - One of the best experts on this subject based on the ideXlab platform.
-
Development and Validation of a Propeller Slipstream Model for Unmanned Aerial Vehicles
Journal of Aircraft, 2015Co-Authors: Waqas Khan, Meyer NahonAbstract:Recent interest in high-angle-of-attack flight, aerobatic maneuvering, vertical/short takeoff and landing, etc., of small unmanned aerial vehicles necessitates more detailed modeling of the complex aerodynamics associated with these flight regimes. This includes modeling the effect of the propeller Slipstream, also called prop wash, which is the main source of airflow that helps maintain lift and control during near-zero forward-speed flight like that encountered during vertical/short takeoff and landing, as well as during high-angle-of-attack flight/aerobatic maneuvering like hovering. Propeller Slipstream models based on conventional theories, such as the momentum theory, have been used extensively in the literature to predict the induced air velocity within the Slipstream. However, because these conventional theories consider only acceleration of air within the Slipstream and not diffusion, their applicability in regions far downstream of the propeller where diffusion is dominant, is questionable. This...
-
Improvement and validation of a propeller Slipstream model for small unmanned aerial vehicles
2014 International Conference on Unmanned Aircraft Systems ICUAS 2014 - Conference Proceedings, 2014Co-Authors: Waqas Khan, Meyer NahonAbstract:Airflow from the propeller, called the propeller Slipstream or propwash, plays an important role in the aerodynamics of small unmanned aerial vehicles (UAVs). In fact, flights at low forward speeds or extreme angle of attack (AoA) maneuvers of small fixed-wing UAVs are possible only because the propeller Slipstream provides the airflow necessary to maintain lift and control under these conditions. Almost all related works in the literature consider propeller Slipstream effect on the UAV's aerodynamics by means of simple theories such as the momentum theory, classical lifting line theory etc. However, these theories take into account only the acceleration of air within the Slipstream while failing to account for its diffusion, thereby limiting their applicability to the region near the propeller where acceleration is dominant; far downstream of the propeller, these theories predict unreasonably high induced velocities since diffusion of the Slipstream is not accounted for. A propeller Slipstream model that considers both acceleration and diffusion within the Slipstream has already been presented in a previous work by the authors. The main objective of the current work is to present improvements made to the model in light of detailed experimental measurements of the induced velocity downstream of the propeller. Thereafter, validity of the propeller Slipstream model is also demonstrated via additional experiments. The model is shown to be accurate up to an axial distance of 5 propeller diameters from the propeller plane, with a root mean square error of 0.45 m/s at 1750 rpm and 1.21 m/s at 6425 rpm.
-
Propeller Slipstream model for small unmanned aerial vehicles
AIAA Modeling and Simulation Technologies (MST) Conference, 2013Co-Authors: Waqas Khan, Ryan James Caverly, Meyer NahonAbstract:McGill University, Montreal, Quebec, Canada Propeller Slipstream, or propwash, can significantly affect the aerodynamic characteristics of propeller driven aircraft by providing additional airflow over their aerodynamic and control surfaces. It is therefore essential to have a good knowledge of the induced velocity within the propeller Slipstream to determine the aerodynamic forces and moments on Slipstream-immersed components. Existing Slipstream models based on simple momentum and lifting line theory have limited application since they consider only the acceleration of air within the Slipstream and do not take into account the diffusion phenomenon. As such, they yield good results near the propeller where acceleration is dominant but fail to predict induced velocity accurately far behind the propeller where diffusion dominates. This paper presents a Slipstream model that takes into account both the acceleration and diffusion phenomena via simple analytical and semi-empirical equations to predict induced velocity accurately up to ~ 8 - 10 propeller diameters downstream of the propeller plane.
