Continuous shape shifting: Enabling loop co-optimization via near-free dynamic code rewriting

Jain, Animesh; Laurenzano, Michael A.; Tang, Lingjia; Mars, Jason

doi:10.1109/micro.2016.7783726

Cited by 11 publications

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Purely static approaches work at compilation time. ey might be applied by the compiler, either automatically, i.e., without user intervention (Cong and Yuan 2012;Jain et al 2016;Luk et al 2009;Poesia et al 2017;Rossbach et al 2013;Sreelatha et al 2018;Tang et al 2013), or not. In the la er case, users can use annotations (Mendonc ¸a et al 2017), domain speci c programming languages (Luk et al 2009;Rossbach et al 2013) or library calls (Augonnet et al 2011) to indicate where each program part should run.…”

Section: A General Overview On Program Scheduling In Heterogeneous Systemsmentioning

confidence: 99%

Mapping Computations in Heterogeneous Multicore Systems with Statistical Regression on Program Inputs

Silva

Leão

Petrucci

et al. 2021

ACM Trans. Embed. Comput. Syst.

View full text Add to dashboard Cite

A hardware configuration is a set of processors and their frequency levels in a multicore heterogeneous system. This article presents a compiler-based technique to match functions with hardware configurations. Such a technique consists of using multivariate linear regression to associate function arguments with particular hardware configurations. By showing that this classification space tends to be convex in practice, this article demonstrates that linear regression is not only an efficient tool to map computations to heterogeneous hardware, but also an effective one. To demonstrate the viability of multivariate linear regression as a way to perform adaptive compilation for heterogeneous architectures, we have implemented our ideas onto the Soot Java bytecode analyzer. Code that we produce can predict the best configuration for a large class of Java and Scala benchmarks running on an Odroid XU4 big.LITTLE board; hence, outperforming prior techniques such as ARM’s GTS and CHOAMP, a recently released static program scheduler.

show abstract

Section: A General Overview On Program Scheduling In Heterogeneous Systemsmentioning

confidence: 99%

Mapping Computations in Heterogeneous Multicore Systems with Statistical Regression on Program Inputs

Silva

Leão

Petrucci

et al. 2021

ACM Trans. Embed. Comput. Syst.

View full text Add to dashboard Cite

show abstract

“…Purely static approaches work at compilation time. They might be applied by the compiler, either automatically, i.e., without user intervention [8,12,16,24,26,29], or not. In the latter case, developers can use annotations [19], domain specific programming languages [16,26] or library calls [1] to indicate where each program part should run.…”

Section: Related Workmentioning

confidence: 99%

Compiler-assisted adaptive program scheduling in big.LITTLE systems

Novaes

Petrucci²,

Gamatié

et al. 2019

Proceedings of the 24th Symposium on Principles and Practice of Parallel Programming

View full text Add to dashboard Cite

Energy-aware architectures provide applications with a mix of low (LITTLE) and high (big) frequency cores. Choosing the best hardware configuration for a program running on such an architecture is difficult, because program parts benefit differently from the same hardware configuration. State-ofthe-art techniques to solve this problem adapt the program's execution to dynamic characteristics of the runtime environment, such as energy consumption and throughput. We claim that these purely dynamic techniques can be improved if they are aware of the program's syntactic structure. To support this claim, we show how to use the compiler to partition source code into program phases: regions whose syntactic characteristics lead to similar runtime behavior. We use reinforcement learning to map pairs formed by a program phase and a hardware state to the configuration that best fit this setup. To demonstrate the effectiveness of our ideas, we have implemented the Astro system. Astro uses Q-learning to associate syntactic features of programs with hardware configurations. As a proof of concept, we provide evidence that Astro outperforms GTS, the ARM-based Linux scheduler tailored for heterogeneous architectures, on the parallel benchmarks from Rodinia and Parsec.

show abstract

“…In addition to prior work investigating the utilization and efficiency of datacenter systems [39][40][41][42][43][44][45][46][47][48][49][50][51][52], there has been growing interest in building large scale datacenter systems for Deep Neural Network workloads. Various accelerators, such as GPUs, ASICs, and FPGAs, have been proposed for datacenters to better handle DNN computation [9,[53][54][55].…”

Section: Related Workmentioning

confidence: 99%

Neurosurgeon

et al. 2017

Self Cite

View full text Add to dashboard Cite

The computation for today's intelligent personal assistants such as Apple Siri, Google Now, and Microsoft Cortana, is performed in the cloud. This cloud-only approach requires significant amounts of data to be sent to the cloud over the wireless network and puts significant computational pressure on the datacenter. However, as the computational resources in mobile devices become more powerful and energy efficient, questions arise as to whether this cloud-only processing is desirable moving forward, and what are the implications of pushing some or all of this compute to the mobile devices on the edge. In this paper, we examine the status quo approach of cloud-only processing and investigate computation partitioning strategies that effectively leverage both the cycles in the cloud and on the mobile device to achieve low latency, low energy consumption, and high datacenter throughput for this class of intelligent applications. Our study uses 8 intelligent applications spanning computer vision, speech, and natural language domains, all employing state-of-the-art Deep Neural Networks (DNNs) as the core machine learning technique. We find that given the characteristics of DNN algorithms, a fine-grained, layer-level computation partitioning strategy based on the data and computation variations of each layer within a DNN has significant latency and energy advantages over the status quo approach. Using this insight, we design Neurosurgeon, a lightweight scheduler to automatically partition DNN computation between mobile devices and datacenters at the granularity of neural network layers. Neurosurgeon does not require per-application profiling. It adapts to various DNN architectures, hardware platforms, wireless networks, and server load levels, intelligently partitioning computation for

show abstract

Continuous shape shifting: Enabling loop co-optimization via near-free dynamic code rewriting

Cited by 11 publications

References 30 publications

Mapping Computations in Heterogeneous Multicore Systems with Statistical Regression on Program Inputs

Mapping Computations in Heterogeneous Multicore Systems with Statistical Regression on Program Inputs

Compiler-assisted adaptive program scheduling in big.LITTLE systems

Neurosurgeon

Contact Info

Product

Resources

About