检索结果-内蒙古大学图书馆

Allo: A programming Model for Composable Accelerator design

proceedings OF THE acm ON programming languageS-PacmPL 2024年第PLDI期8卷 593-620页

作者： Chen, Hongzheng Zhang, Niansong Xiang, Shaojie Zeng, Zhichen Dai, Mengjia Zhang, Zhiru Cornell Univ Ithaca NY 14853 USA Univ Sci & Technol China Hefei Peoples R China

Special-purpose hardware accelerators are increasingly pivotal for sustaining performance improvements in emerging applications, especially as the benefits of technology scaling continue to diminish. However, designers currently lack effective tools and methodologies to construct complex, high-performance accelerator architectures in a productive manner. Existing high-level synthesis (HLS) tools often require intrusive source-level changes to attain satisfactory quality of results. Despite the introduction of several new accelerator design languages (ADLs) aiming to enhance or replace HLS, their advantages are more evident in relatively simple applications with a single kernel. Existing ADLs prove less effective for realistic hierarchical designs with multiple kernels, even if the design hierarchy is flattened. In this paper, we introduce Allo, a composable programming model for efficient spatial accelerator design. Allo decouples hardware customizations, including compute, memory, communication, and data type from algorithm specification, and encapsulates them as a set of customization primitives. Allo preserves the hierarchical structure of an input program by combining customizations from different functions in a bottom-up, type-safe manner. This approach facilitates holistic optimizations that span across function boundaries. We conduct comprehensive experiments on commonly-used HLS benchmarks and several realistic deep learning models. Our evaluation shows that Allo can outperform state-of-the-art HLS tools and ADLs on all test cases in the PolyBench. For the GPT2 model, the inference latency of the Allo generated accelerator is 1.7x faster than the NVIDIA A100 GPU with 5.4x higher energy efficiency, demonstrating the capability of Allo to handle large-scale designs.

关键词： Hardware accelerators schedule language accelerator design language compiler optimization

来源：评论

学校读者我要写书评

暂无评论

programming Support for Autonomizing Software 2019

Programming Support for Autonomizing Software

引用

40th acm sigplan conference on programming language design and implementation (PLDI) part of acm's Federated Computing Research conference (FCRC)

作者： Lee, Wen-Chuan Liu, Peng Liu, Yingqi Ma, Shiqing Zhang, Xiangyu Purdue Univ W Lafayette IN 47907 USA IBM Res San Jose CA USA

ISBN: (纸本)9781450367127

Most traditional software systems are not built with the artificial intelligence support (AI) in mind. Among them, some may require human interventions to operate, e.g., the manual specification of the parameters in the data processing programs, or otherwise, would behave poorly. We propose a novel framework called Autonomizer to autonomize these systems by installing the AI into the traditional programs. Autonomizer is general so it can be applied to many real-world applications. We provide the primitives and the runtime support, where the primitives abstract common tasks of autonomization and the runtime support realizes them transparently. With the support of Autonomizer, the users can gain the AI support with little engineering efforts. Like many other AI applications, the challenge lies in the feature selection, which we address by proposing multiple automated strategies based on the program analysis. Our experiment results on nine real-world applications show that the autonomization only requires adding a few lines to the source code. Besides, for the data-processing programs, Autonomizer improves the output quality by 161% on average over the default settings. For the interactive programs such as game/driving, Autonomizer achieves higher success rate with lower training time than existing autonomized programs.

关键词： AI Deep Learning Software Autonomization Dynamic Program Analysis

来源：评论

学校读者我要写书评

暂无评论

Evaluating Iterative Optimization Across 1000 Data Sets 10

Evaluating Iterative Optimization Across 1000 Data Sets

引用

acm sigplan conference on programming language design and implementation

作者： Chen, Yang Huang, Yuanjie Eeckhout, Lieven Fursin, Grigori Peng, Liang Temam, Olivier Wu, Chengyong Chinese Acad Sci ICT LCSA Key Lab Comp Syst & Architecture Beijing Peoples R China Chinese Acad Sci Grad Sch Beijing Peoples R China INRIA Saclay France Univ Ghent Ghent Belgium

ISBN: (纸本)9781450300193

While iterative optimization has become a popular compiler optimization approach, it is based on a premise which has never been truly evaluated: that it is possible to learn the best compiler optimizations across data sets. Up to now, most iterative optimization studies find the best optimizations through repeated runs on the same data set. Only a handful of studies have attempted to exercise iterative optimization on a few tens of data sets. In this paper, we truly put iterative compilation to the test for the first time by evaluating its effectiveness across a large number of data sets. We therefore compose KDataSets, a data set suite with 1000 data sets for 32 programs, which we release to the public. We characterize the diversity of KDataSets, and subsequently use it to evaluate iterative optimization. We demonstrate that it is possible to derive a robust iterative optimization strategy across data sets: for all 32 programs, we find that there exists at least one combination of compiler optimizations that achieves 86% or more of the best possible speedup across all data sets using Intel's ICC (83% for GNU's GCC). This optimal combination is program-specific and yields speedups up to 1.71 on ICC and 2.23 on GCC over the highest optimization level (-fast and -03, respectively). This finding makes the task of optimizing programs across data sets much easier than previously anticipated, and it paves the way for the practical and reliable usage of iterative optimization. Finally, we derive pre-shipping and post-shipping optimization strategies for software vendors.

关键词： design Experimentation Measurement Performance

来源：评论

学校读者我要写书评

暂无评论

Parametric analysis for adaptive computation offloading

Parametric analysis for adaptive computation offloading

引用

conference on programming language design and implementation

作者： Wang, C Li, ZY Purdue Univ Dept Comp Sci W Lafayette IN 47907 USA

Many programs can be invoked under different execution options;input parameters and data files. Such different execution contexts may lead to strikingly different execution instances. The optimal code generation may be sensitive to the execution instances. In this paper, we show how to use parametric program analysis to deal with this issue for the optimization problem of computation offloading. Computation offloading has been shown to be an effective way to improve performance and energy saving on mobile devices. Optimal program partitioning for computation offloading depends on the tradeoff between the computation workload and the communication cost. The computation workload and communication requirement may change with different execution instances. Optimal decisions on program partitioning must be made at run time when sufficient information about workload and communication requirement becomes available. Our cost analysis obtains program computation workload and communication cost expressed as functions of run-time parameters, and our parametric partitioning algorithm finds the optimal program partitioning corresponding to different ranges of run-time parameters. At run time, the transformed program self-schedules its tasks on either the mobile device or the server, based on the optimal program partitioning that corresponds to the current values of run-time parameters. Experimental results on an HP IPAQ handheld device show that different run-time parameters can lead to quite different program partitioning decisions.

关键词： algorithms design experimentation measurement performance program partitioning program analysis program transformation computation offloading program profiling handheld devices adaptive optimization distributed system

来源：评论

学校读者我要写书评

暂无评论

Mirrors: design principles for meta-level facilities of object-oriented programming languages

Mirrors: Design principles for meta-level facilities of obje...

引用

19th conference on Object-Oriented programming, Systems, languages, and Applications

作者： Bracha, G Ungar, D Sun Microsyst Inc Santa Clara CA 95054 USA Sun Microsyst Inc Mountain View CA 94043 USA

ISBN: (纸本)1581138318

We identify three design principles for reflection and metaprogramming facilities in object oriented programming languages. Encapsulation: meta-level facilities must encapsulate their implementation. Stratification: meta-level facilities must be separated from base-level functionality. Ontological correspondence: the ontology of meta-level facilities should correspond to the ontology of the language they manipulate. Traditional/mainstream reflective architectures do not follow these precepts. In contrast, reflective APIs built around the concept of mirrors are characterized by adherence to these three principles. Consequently, mirror-based architectures have significant advantages with respect to distribution, deployment and general purpose metaprogramming.

关键词： design languages reflection metaprogramming mirrors Java self smalltalk

来源：评论

学校读者我要写书评

暂无评论

Fast RMWs for TSO: Semantics and implementation

Fast RMWs for TSO: Semantics and Implementation

引用

34th acm sigplan conference on programming language design and implementation (PLDI)

作者： Rajaram, Bhargava Nagarajan, Vijay Sarkar, Susmit Elver, Marco Univ Edinburgh Edinburgh EH8 9YL Midlothian Scotland Univ St Andrews St Andrews KY16 9AJ Fife Scotland

ISBN: (纸本)9781450320146

Read-Modify-Write (RMW) instructions are widely used as the building blocks of a variety of higher level synchronization constructs, including locks, barriers, and lock-free data structures. Unfortunately, they are expensive in architectures such as x86 and SPARC which enforce (variants of) Total-Store-Order (TSO). A key reason is that RMWs in these architectures are ordered like a memory barrier, incurring the cost of a write-buffer drain in the critical path. Such strong ordering semantics are dictated by the requirements of the strict atomicity definition (type-1) that existing TSO RMWs use. Programmers often do not need such strong semantics. Besides, weakening the atomicity definition of TSO RMWs, would also weaken their ordering - thereby leading to more efficient hardware implementations. In this paper we argue for TSO RMWs to use weaker atomicity definitions - we consider two weaker definitions: type-2 and type-3, with different relaxed ordering differences. We formally specify how such weaker RMWs would be ordered, and show that type-2 RMWs, in particular, can seamlessly replace existing type-1 RMWs in common synchronization idioms - except in situations where a type-1 RMW is used as a memory barrier. Recent work has shown that the new C/C++ 11 concurrency model can be realized by generating conventional (type-1) RMWs for C/C++ 11 SC-atomic-writes and/or SC-atomic-reads. We formally prove that this is equally valid using the proposed type-2 RMWs;type-3 RMWs, on the other hand, could be used for SC-atomic-reads (and optionally SC-atomic-writes). We further propose efficient microarchitectural implementations for type-2 (type-3) RMWs - simulation results show that our implementation reduces the cost of an RMW by up to 58.9% (64.3%), which translates into an overall performance improvement of up to 9.0% (9.2%) on a set of parallel programs, including those from the SPLASH-2, PARSEC, and STAMP benchmarks.

关键词： design Performance Experimentation Read-Modify-Write (RMW) Total-Store-Order (TSO) Atomics

来源：评论

学校读者我要写书评

暂无评论

Compiling a subset of apl into a typed intermediate language∗ 1

Compiling a subset of apl into a typed intermediate language...

引用

1st acm sigplan International Workshop on Libraries, languages and Compilers for Array programming, ARRAY 2014 - Part of PLDI 2014

作者： Elsman, Martin Dybdal, Martin Department of Computer Science University of Copenhagen Denmark

ISBN: (纸本)9781450329378

We present a compiler and a typed intermediate language for a subset of APL. The intermediate language treats all numeric data as multi-dimensional arrays and the type system makes explicit the ranks of arrays. Primitive operators are polymorphic in rank and in the type of the underlying data operated on. The frontend of the APL compiler deals with much of the gory details of the APL language, including infix resolution, resolution of identity items for reduce operations, resolution of default element values for empty arrays, scalar extensions, and resolution of certain kinds of overloading. We demonstrate the usefulness of the intermediate language by showing that it can be compiled efficiently, using known techniques, such as delayed arrays, into a C-like language. We also demonstrate that the language is sufficiently expressive that some primitive operators, such as APL's inner product operator, which works on arrays of arbitrary dimensions, can be compiled using more primitive operators.. Copyright © 2014 acm 978-1-4503-2937-8/14/06. 15.00.

关键词： Calculations

来源：评论

学校读者我要写书评

暂无评论

The undergraduate capstone software design experience

The undergraduate capstone software design experience

引用

1997 conference on TRI-Ada, TRI-Ada 1997

作者： Blair, Jean R. S. Ressler, Eugene K. Wagner, Thomas D. Department of Electrical Engineering and Computer Science United States Military Academy West PointNY10996 United States

ISBN: (纸本)0897919815

Ada 95 is being used as the implementation language for a senior level compiler design course at the United States Military Academy. This paper describes experiences and lessons learned as well as the scenario based approach for a large object oriented compiler project in that course. The term-long compiler project gives the students experience as a part of a (fictitious) design- implementation team building a production compiler. Ada 95 is compared to Pascal and C++ for the same course and positive and negative aspects of the three languages are described. © 1997 acm.

关键词： Software design

来源：评论

学校读者我要写书评

暂无评论

Typed Lua: An optional type system for Lua

Typed Lua: An optional type system for Lua

引用

8th Workshop on Dynamic languages and Applications, Dyla 2014, with PLDI 2014

作者： Maidl, André Murbach Mascarenhas, Fabio Ierusalimschy, Roberto PUC-Rio Rio de Janeiro Brazil UFRJ Rio de Janeiro Brazil

ISBN: (纸本)9781450329163

Dynamically typed languages trade flexibility and ease of use for safety, while statically typed languages prioritize the early detection of bugs, and provide a better framework for structure large programs. The idea of optional typing is to combine the two approaches in the same language: the programmer can begin development with dynamic types, and migrate to static types as the program matures. The challenge is designing a type system that feels natural to the programmer that is used to programming in a dynamic language. This paper presents the initial design of Typed Lua, an optionally-typed extension of the Lua scripting language. Lua is an imperative scripting language with first class functions and lightweight metaprogramming mechanisms. The design of Typed Lua's type system has a novel combination of features that preserves some of the idioms that Lua programmers are used to, while bringing static type safety to them. We show how the major features of the type system type these idioms with some examples, and discuss some of the design issues we faced. © 2014 acm.

关键词： gradual typing Lua programming language optional typing type systems

来源：评论

学校读者我要写书评

暂无评论

Garbage Collection for Monitoring Parametric Properties 11

Garbage Collection for Monitoring Parametric Properties

引用

32nd acm sigplan conference on programming language design and implementation (PLDI 11)

作者： Jin, Dongyun Meredith, Patrick O'Neil Griffith, Dennis Rosu, Grigore Univ Illinois Dept Comp Sci Urbana IL USA

ISBN: (纸本)9781450306638

Parametric properties are behavioral properties over program events that depend on one or more parameters. Parameters are bound to concrete data or objects at runtime, which makes parametric properties particularly suitable for stating multi-object relationships or protocols. Monitoring parametric properties independently of the employed formalism involves slicing traces with respect to parameter instances and sending these slices to appropriate non-parametric monitor instances. The number of such instances is theoretically unbounded and tends to be enormous in practice, to an extent that how to efficiently manage monitor instances has become one of the most challenging problems in runtime verification. The previous formalism-independent approach was only able to do The obvious, namely to garbage collect monitor instances when all bound parameter objects were garbage collected. This led to pathological behaviors where unnecessary monitor instances were kept for the entire length of a program. This paper proposes a new approach to garbage collecting monitor instances. Unnecessary monitor instances are collected lazily to avoid creating undue overhead. This lazy collection, along with some careful engineering, has resulted in RV, the most efficient parametric monitoring system to date. Our evaluation shows that the average overhead of RV in the DaCapo benchmark is 15%, which is two times lower than that of Java MOP and orders of magnitude lower than that of Tracematches.

关键词： runtime verification runtime monitoring testing debugging aspect-oriented programming garbage collection

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：