This paper examines a set of commercially representative embedded programs and compares them to an existing benchmark suite, SPEC2000. A new version of SimpleScalar that has been adapted to the ARM instruction set is used to characterize the performance of the benchmarks using configurations similar to current and next generation embedded processors. Several characteristics distinguish the representative embedded programs from the existing SPEC benchmarks including instruction distribution, memory behavior, and available parallelism. The embedded benchmarks, called MiBench, are freely available to all researchers.

MiBench: A free, commercially representative embedded benchmark suite

Significant advances have been made in compilation technology for capitalizing on instruction-level parallelism (ILP). The vast majority of ILP compilation research has been conducted in the context of general-purpose computing, and more specifically the SPEC benchmark suite. At the same time, a number of microprocessor architectures have emerged which have VLIW and SIMD structures that are well matched to the needs of the ILP compilers. Most of these processors are targeted at embedded applications such as multimedia and communications, rather than general-purpose systems. Conventional wisdom, and a history of hand optimization of inner-loops, suggests that ILP compilation techniques are well suited to these applications. Unfortunately, there currently exists a gap between the compiler community and embedded applications developers. This paper presents MediaBench, a benchmark suite that has been designed to fill this gap. This suite has been constructed through a three-step process: intuition and market driven initial selection, experimental measurement to establish uniqueness, and integration with system synthesis algorithms to establish usefulness.

MediaBench: a tool for evaluating and synthesizing multimedia and communications systems

Benchmarking has become one of the most important methods for quantitative performance evaluation of processor and computer system designs. Benchmarking of modern multiprocessors such as chip multiprocessors is challenging because of their application domain, scalability and parallelism requirements. In my thesis, I have developed a methodology to design effective benchmark suites and demonstrated its effectiveness by developing and deploying a benchmark suite for evaluating multiprocessors. 
More specifically, this thesis includes several contributions. First, the thesis shows that a new benchmark suite for multiprocessors is needed because the behavior of modern parallel programs is significantly different from those represented by SPLASH-2, the most popular parallel benchmark suite developed over ten years ago. Second, the thesis quantitatively describes the requirements and characteristics of a set of multithreaded programs and their underlying technology trends. Third, the thesis presents a systematic approach to scale and select benchmark inputs with the goal of optimizing benchmarking accuracy subject to constrained execution or simulation time. Finally, the thesis describes a parallel benchmark suite called PARSEC for evaluating modern shared-memory multiprocessors. Since its initial release, PARSEC has been adopted by many architecture groups in both research and industry.

Benchmarking modern multiprocessors

Power dissipation is increasingly important in CPUs ranging from those intended for mobile use, all the way up to high-performance processors for high-end servers. While the bulk of the power dissipated is dynamic switching power, leakage power is also beginning to be a concern. Chipmakers expect that in future chip generations, leakage's proportion of total chip power will increase significantly.This paper examines methods for reducing leakage power within the cache memories of the CPU. Because caches comprise much of a CPU chip's area and transistor counts, they are reasonable targets for attacking leakage. We discuss policies and implementations for reducing cache leakage by invalidating and “turning off” cache lines when they hold data not likely to be reused. In particular, our approach is targeted at the generational nature of cache line usage. That is, cache lines typically have a flurry of frequent use when first brought into the cache, and then have a period of “dead time” before they are evicted. By devising effective, low-power ways of deducing dead time, our results show that in many cases we can reduce LI cache leakage energy by 4x in SPEC2000 applications without impacting performance. Because our decay-based techniques have notions of competitive on-line algorithms at their roots, their energy usage can be theoretically bounded at within a factor of two of the optimal oracle-based policy. We also examine adaptive decay-based policies that make energy-minimizing policy choices on a per-application basis by choosing appropriate decay intervals individually for each cache line. Our proposed adaptive policies effectively reduce LI cache leakage energy by 5x for the SPEC2000 with only negligible degradations in performance.

/pdf/cache-decay-exploiting-generational-behavior-to-reduce-cache-3wl24vywlv.pdf

Cache decay: exploiting generational behavior to reduce cache leakage power

Three-dimensional integration enables stacking memory directly on top of a microprocessor, thereby significantly reducing wire delay between the two. Previous studies have examined the performance benefits of such an approach, but all of these works only consider commodity 2D DRAM organizations. In this work, we explore more aggressive 3D DRAM organizations that make better use of the additional die-to-die bandwidth provided by 3D stacking, as well as the additional transistor count. Our simulation results show that with a few simple changes to the 3D-DRAM organization, we can achieve a 1.75x speedup over previously proposed 3D-DRAM approaches on our memory-intensive multi-programmed workloads on a quad-core processor. The significant increase in memory system performance makes the L2 miss handling architecture (MHA) a new bottleneck, which we address by combining a novel data structure called the Vector Bloom Filter with dynamic MSHR capacity tuning. Our scalable L2 MHA yields an additional 17.8% performance improvement over our 3D-stacked memory architecture.

/pdf/3d-stacked-memory-architectures-for-multi-core-processors-1ufiq9rsba.pdf

3D-Stacked Memory Architectures for Multi-core Processors

We present a framework for rapidly exploring the design space of low power application-specific programmable processors (ASPP), in particular mediaprocessors. We focus on a category of processors that are programmable yet optimized to reduce power consumption for a specific set of applications. The key components of the framework presented in this paper are a retargetable instruction level parallelism (ILP) compiler, processor simulators, a set of complete media applications written in a high level language and an architectural component selection algorithm. The fundamental idea behind the framework is that with the aid of a retargetable ILP compiler and simulators it is possible to arrange architectural parameters (e.g., the issue width, the size of cache memory units, the number of execution units, etc.) to meet low power design goals under area constraints.

http://www.cecs.uci.edu/~papers/compendium94-03/papers/1999/dac99/pdffiles/19_1.pdf

Power efficient mediaprocessors: design space exploration

This paper presents a system level approach for the synthesis of hard real-time multitask application specific systems. The algorithm takes into account task precedence constraints among multiple hard real-time tasks and targets a multiprocessor system consisting of a set of heterogeneous off-the-shelf processors. The optimization goal is to select a minimal cost multi-subset of processors while satisfying all the required timing and precedence constraints. There are three design phases: resource allocation, assignment, and scheduling. Since the resource allocation is a search for a minimal cost multi-subset of processors, we adopted an A* search based technique for the first synthesis phase. A variation of the force-directed optimization technique is used to assign a task to an allocated processor. The final scheduling of a hard-real time task is done by the task level scheduler which is based on Earliest Deadline First (EDF) scheduling policy. Our task level scheduler incorporates force-directed scheduling methodology to address the situations where EDF is not optimal. The experimental results on a variety of examples show that the approach is highly effective and efficient.

/pdf/synthesis-of-hard-real-time-application-specific-systems-1c2u1lr8ou.pdf

Synthesis of Hard Real-Time Application Specific Systems

We developed a new hierarchical modular approach for synthesis of area-minimal core-based data-intensive systems. The optimization approach employs a novel global least-constraining most-constrained heuristic to minimize the instruction cache misses for a given application, instruction cache size and organization. Based on this performance optimization technique, we constructed a strategy to search for a minimal-area processor core, and an instruction and data cache which satisfy the performance characteristics of a set of target applications. The synthesis platform integrates the existing modeling, profiling, and simulation tools with the developed system-level synthesis tools. The effectiveness of the approach is demonstrated on a variety of modern real-life multimedia and communication applications.

/pdf/application-driven-synthesis-of-core-based-systems-25hi2szbcf.pdf

Application-driven synthesis of core-based systems

Due to the increasing popularity of multimedia and communications applications, requirements for application-specific systems typically include design flexibility and data management ability. Since the development of such systems is a market-driven task, reducing the time to market and manufacturing cost, while still satisfying application performance requirements, is an important system synthesis requirement. We have developed a new approach for area optimization of core-based systems. The approach uses basic block relocation in order to reduce the number of cache misses and, thus, enable hardware savings during system synthesis. Given a processor model, a cache model, and a set of nonpreemptive tasks with timing constraints, the goal of the synthesis framework is to select a system configuration (processor, I-cache, and D-cache) of minimal area that satisfies the performance constraints. The system synthesis framework has two key components. The first component is a code optimization engine that relocates basic blocks within a given assembly program in order to reduce the number of cache misses. The second component is a search mechanism that leverages the improvements in code performance obtained by the first component to select the most area-efficient system configuration. In order to bridge the gap between the profiling and modeling tools, we have constructed a new performance evaluation platform. It integrates the existing modeling, profiling, and simulation tools with the developed system-level synthesis tools. The effectiveness of the synthesis approach is demonstrated on a variety of modern real-life multimedia and communication applications.

/pdf/application-driven-synthesis-of-memory-intensive-systems-on-4dboafx83i.pdf

Chunho Lee

Papers

MediaBench: a tool for evaluating and synthesizing multimedia and communications systems

Power efficient mediaprocessors: design space exploration

Synthesis of Hard Real-Time Application Specific Systems

Application-driven synthesis of core-based systems

Application-driven synthesis of memory-intensive systems-on-chip