ExaSAT: An exascale co-design tool for performance modeling

Unat, Didem; Chan, Cy; Zhang, Weiqun; Williams, Samuel; Bachan, John; Bell, John B.; Shalf, John

doi:10.1177/1094342014568690

Cited by 33 publications

(21 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…smooth, restrict, etc.) are modeled using the ExaSAT performance modeling framework [24], which predicts that the performance of the multigrid kernels will be determined by the hardware's on-node memory bandwidth and cache sizes. By tuning these parameters to represent an exascale class node architecture, we can estimate the compute time for each kernel during simulation.…”

Section: Resultsmentioning

confidence: 99%

“…The AMR-specific aspects of the API currently allow description of skeletonized algorithms, meaning that they generate the computation and communication event traces necessary to drive the network simulator, without doing any actual floatingpoint computation. The time spent in on-node compute kernels is represented by performance model profiles generated by the ExaSAT performance modeling framework [24].…”

Section: A Descriptionmentioning

confidence: 99%

See 1 more Smart Citation

Topology-Aware Performance Optimization and Modeling of Adaptive Mesh Refinement Codes for Exascale

P¹,

D²,

P³

et al. 2016

2016 First International Workshop on Communication Optimizations in HPC (COMHPC)

View full text Add to dashboard Cite

Abstract-We introduce a topology-aware performance optimization and modeling workflow for AMR simulation that includes two new modeling tools, ProgrAMR and Mota Mapper, which interface with the BoxLib AMR framework and the SSTmacro network simulator. ProgrAMR allows us to generate and model the execution of task dependency graphs from high-level specifications of AMR-based applications, which we demonstrate by analyzing two example AMR-based multigrid solvers with varying degrees of asynchrony. Mota Mapper generates multiobjective, network topology-aware box mappings, which we apply to optimize the data layout for the example multigrid solvers. While the sensitivity of these solvers to layout and execution strategy appears to be modest for balanced scenarios, the impact of better mapping algorithms can be significant when performance is highly constrained by network hop latency. Furthermore, we show that network latency in the multigrid bottom solve is the main contributing factor preventing good scaling on exascale-class machines.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: A Descriptionmentioning

confidence: 99%

Topology-Aware Performance Optimization and Modeling of Adaptive Mesh Refinement Codes for Exascale

P¹,

D²,

P³

et al. 2016

2016 First International Workshop on Communication Optimizations in HPC (COMHPC)

View full text Add to dashboard Cite

show abstract

“…LSE [68] is a fully concurrentstructural modeling framework designed to maximize reusability of components. There are also many other works in the field of HPC for automatically performing performance modeling [3,4,67]. Most of these languages and systems serve a different purpose of expressing the mapping between performance/power models and specific detailed application/architecture and are not well suited for high-level analytical design space exploration.…”

Section: Systems and Languages Supporting Analytical Modelingmentioning

confidence: 99%

Language Support for Navigating Architecture Design in Closed Form

Cui

Tzimpragos

Tao

et al. 2019

J. Emerg. Technol. Comput. Syst.

View full text Add to dashboard Cite

As computer architecture continues to expand beyond software-agnostic microarchitecture to specialized and heterogeneous logic or even radically different emerging computing models (e.g., quantum cores, DNA storage units), detailed cycle-level simulation is no longer presupposed. Exploring designs under such complex interacting relationships (e.g., performance, energy, thermal, frequency) calls for a more integrative but higher-level approach. We propose Charm, a modeling language supporting closed-form high-level architecture modeling. Charm enables mathematical representations of mutually dependent architectural relationships to be specified, composed, checked, evaluated, reused, and shared. The language is interpreted through a combination of automatic symbolic evaluation, scalable graph transformation, and efficient compiler techniques, generating executable DAGs and optimized analysis procedures. Charm also exploits the advancements in satisfiability modulo theory solvers to automatically search the design space to help architects explore multiple design knobs simultaneously (e.g., different CNN tiling configurations). Through two case studies, we demonstrate that Charm allows one to define high-level architecture models in a clean and concise format, maximize reusability and shareability, capture unreasonable assumptions, and significantly ease design space exploration at a high level.

show abstract

“…The numbers of species (N sp ) in the chemical kinetic models of interest range between O(10) and O(100) and the number of reactions (N reac ) scales linearly (i.e., N reac ≈ 5N sp [2]). The number of independent ODE systems (N ode ) range from O(10 4 )-O(10 6 ) per device [30] in 3-D combustion simulations. Two SIMD parallel processing strategies can be designed based on these expected values.…”

Section: Parallel Integrator Implementationsmentioning

confidence: 99%

Accelerating finite-rate chemical kinetics with coprocessors: Comparing vectorization methods on GPUs, MICs, and CPUs

Stone¹,

Alferman

Niemeyer

2018

Computer Physics Communications

View full text Add to dashboard Cite

Accurate and efficient methods for solving stiff ordinary differential equations (ODEs) are a critical component of turbulent combustion simulations with finite-rate chemistry. The ODEs governing the chemical kinetics at each mesh point are decoupled by operator-splitting allowing each to be solved concurrently. An efficient ODE solver must then take into account the available thread and instruction-level parallelism of the underlying hardware, especially on many-core coprocessors, as well as the numerical efficiency. A stiff Rosenbrock and a nonstiff Runge-Kutta ODE solver are both implemented using the single instruction, multiple thread (SIMT) and single instruction, multiple data (SIMD) paradigms within OpenCL. Both methods solve multiple ODEs concurrently within the same instruction stream. The performance of these parallel implementations was measured on three chemical kinetic models of increasing size across several multicore and many-core platforms. Two separate benchmarks were conducted to clearly determine any performance advantage offered by either method. The first measured the run-time of evaluating the righthand-side source terms in parallel and the second benchmark integrated a series of constant-pressure, homogeneous reactors using the Rosenbrock and Runge-Kutta solvers. The right-hand-side evaluations with SIMD parallelism on the host multicore Xeon CPU and many-core Xeon Phi co-processor performed approximately three times faster than the baseline multithreaded C ++ code. The SIMT parallel model on the host and Phi was 13 % to 35 % slower than the baseline while the SIMT model on the NVIDIA Kepler GPU provided approximately the same performance as the SIMD model on the Phi. The runtimes for both ODE solvers decreased significantly with the SIMD implementations on the host CPU (2.5-2.7 ×) and Xeon Phi coprocessor (4.7-4.9 ×) compared to the baseline parallel code. The SIMT implementations on the GPU ran 1.4-1.6 times faster than the baseline multithreaded CPU code; however, this was significantly slower than the SIMD versions on the host CPU or the Xeon Phi. The performance difference between the three platforms was attributed to thread divergence caused by the adaptive step-sizes within the ODE integrators. 1 arXiv:1608.05794v2 [physics.comp-ph] 29 Aug 2017Analysis showed that the wider vector width of the GPU incurs a higher level of divergence than the narrower Sandy Bridge or Xeon Phi. The significant performance improvement provided by the SIMD parallel strategy motivates further research into more ODE solver methods that are both SIMD-friendly and computationally efficient.

show abstract

ExaSAT: An exascale co-design tool for performance modeling

Cited by 33 publications

References 42 publications

Topology-Aware Performance Optimization and Modeling of Adaptive Mesh Refinement Codes for Exascale

Topology-Aware Performance Optimization and Modeling of Adaptive Mesh Refinement Codes for Exascale

Language Support for Navigating Architecture Design in Closed Form

Accelerating finite-rate chemical kinetics with coprocessors: Comparing vectorization methods on GPUs, MICs, and CPUs

Contact Info

Product

Resources

About