Ismail Akturk

Improving application behavior on heterogeneous manycore systems through kernel mapping

Many-core accelerators are being more frequently deployed to improve the system processing capabilities. In such systems, application mapping must be enhanced to maximize utilization of the underlying architecture. Especially, in graphics processing units (GPUs), mapping kernels that are part of multi-kernel applications has a great impact on overall performance, since kernels may exhibit different characteristics on different CPUs and GPUs. While some kernels run faster on GPUs, others may perform better in CPUs. Thus, heterogeneous execution may yield better performance than executing the application only on a CPU or only on a GPU. In this paper, we investigate on two approaches: a novel profiling-based adaptive kernel mapping algorithm to assign each kernel of an application to the proper device, and a Mixed-Integer Programming (MIP) implementation to determine optimal mapping. We utilize profiling information for kernels on different devices and generate a map that identifies which kernel should run where in order to improve the overall performance of an application. Initial experiments show that our approach can efficiently map kernels on CPUs and GPUs, and outperforms CPU-only and GPU-only approaches.

Accordion: Toward soft Near-Threshold Voltage Computing

While more cores can find place in the unit chip area every technology generation, excessive growth in power density prevents simultaneous utilization of all. Due to the lower operating voltage, Near-Threshold Voltage Computing (NTC) promises to fit more cores in a given power envelope. Yet NTC prospects for energy efficiency disappear without mitigating (i) the performance degradation due to the lower operating frequency; (ii) the intensified vulnerability to parametric variation. To compensate for the first barrier, we need to raise the degree of parallelism - the number of cores engaged in computation. NTC-prompted power savings dominate the power cost of increasing the core count. Hence, limited parallelism in the application domain constitutes the critical barrier to engaging more cores in computation. To avoid the second barrier, the system should tolerate variation-induced errors. Unfortunately, engaging more cores in computation exacerbates vulnerability to variation further. To overcome NTC barriers, we introduce Accordion, a novel, light-weight framework, which exploits weak scaling along with inherent fault tolerance of emerging R(ecognition), M(ining), S(ynthesis) applications. The key observation is that the problem size not only dictates the number of cores engaged in computation, but also the application output quality. Consequently, Accordion designates the problem size as the main knob to trade off the degree of parallelism (i.e. the number of cores engaged in computation), with the degree of vulnerability to variation (i.e. the corruption in application output quality due to variation-induced errors). Parametric variation renders ample reliability differences between the cores. Since RMS applications can tolerate faults emanating from data-intensive program phases as opposed to control, variation-afflicted Accordion hardware executes fault-tolerant data-intensive phases on error-prone cores, and reserves reliable cores for control.

Application-Specific Heterogeneous Network-on-Chip Design

As a result of increasing communication demands, application-specific and scalable Network-onChips (NoCs) have emerged to connect processing cores and subsystems in Multiprocessor Systemon-Chips.A challenge in application-specific NoC design is to find the right balance among different tradeoffs, such as communication latency, power consumption and chip area. We propose a novel approach that generates latency-aware heterogeneous NoC topology. Experimental results show that our approach improves the total communication latency up to 27% with modest power consumption.

Effective Kernel Mapping for OpenCL Applications in Heterogeneous Platforms

Many core accelerators are being deployed in many systems to improve the processing capabilities. In such systems, application mapping need to be enhanced to maximize the utilization of the underlying architecture. Especially in GPUs mapping becomes critical for multi-kernel applications as kernels may exhibit different characteristics. While some of the kernels run faster on GPU, others may refer to stay in CPU due to the high data transfer overhead. Thus, heterogeneous execution may yield to improved performance compared to executing the application only on CPU or only on GPU. In this paper, we propose a novel profiling-based kernel mapping algorithm to assign each kernel of an application to the proper device to improve the overall performance of an application. We use profiling information of kernels on different devices and generate a map that identifies which kernel should run on where to improve the overall performance of an application. Initial experiments show that our approach can effectively map kernels on CPU and GPU, and outperforms to a CPU-only and GPU-only approach.

Decoupled Control and Data Processing for Approximate Near-Threshold Voltage Computing

Near-threshold voltage computing can significantly improve energy efficiency; however, both operating speed and resilience to parametric variation reduce as the operating voltage reaches the threshold voltage. To prevent degradation in throughput performance, more cores should contribute to computation. The corresponding expansion in the chip area, however, is likely to further exacerbate the already intensified vulnerability to variation. Variation itself results in slower cores and ample speed differences among the cores. Worse, variation-induced slowdown might prevent operation at the designated clock speed, leading to timing errors. In this article, the authors exploit the intrinsic error tolerance of emerging Recognition, Mining, and Synthesis applications to mitigate variation. RMS applications can tolerate errors emanating from data-intensive program phases as opposed to control. Accordingly, the authors reserve reliable cores for control, and they execute error-tolerant data-intensive phases on error-prone cores. They also provide a design space exploration for decoupled control and data processing to mitigate variation at near-threshold voltages.

ILP-Based Communication Reduction for Heterogeneous 3D Network-on-Chips

Network-on-Chip (NoC) architectures and three-dimensional integrated circuits (3D ICs) have been introduced as attractive options for overcoming the barriers in interconnect scaling while increasing the number of cores. Combining these two approaches is expected to yield better performance and higher scalability. This paper explores the possibility of combining these two techniques in a heterogeneity aware fashion. We explore how heterogeneous processors can be mapped onto the given 3D chip area to minimize the data access costs. Our initial results indicate that the proposed approach generates promising results within tolerable solution times.

On Approximate Speculative Lock Elision

Each synchronization point represents a point of serialization, and thereby can easily hurt parallel scalability. As demonstrated by recent studies, approximating, i.e., relaxing synchronization by eliminating a subset of synchronization points spatio-temporally can help improve parallel scalability, as long as approximation incurred violations of basic execution semantics remain predictable and controllable. Even if the divergence from fully-synchronized execution renders lower computation accuracy rather than catastrophic program termination, for approximation to be viable, the accuracy loss must be bounded. In this paper, we assess the viability of approximate synchronization using Speculative Lock Elision (SLE), which was adopted by hardware transactional memory implementations from industry, as a baseline for comparison. Specifically, we investigate the efficacy of exploiting semantic and temporal characteristics of critical sections in preventing excessive loss in computation accuracy, and devise a light-weight, proof-of-concept Approximate Speculative Lock Elision (ASLE) implementation, which exploits existing hardware support for SLE.

Toward Dynamic Precision Scaling

This article makes the case for dynamic precision scaling to improve power efficiency by tailoring arithmetic precision adaptively to temporal changes in algorithmic noise tolerance throughout the execution.

AMNESIAC: Amnesic Automatic Computer

Due to imbalances in technology scaling, the energy consumption of data storage and communication by far exceeds the energy consumption of actual data production, i.e., computation. As a consequence, recomputing data can become more energy efficient than storing and retrieving precomputed data. At the same time, recomputation can relax the pressure on the memory hierarchy and the communication bandwidth. This study hence assesses the energy efficiency prospects of trading computation for communication. We introduce an illustrative proof-of-concept design, identify practical limitations, and provide design guidelines.

Accuracy Bugs: A New Class of Concurrency Bugs to Exploit Algorithmic Noise Tolerance

Parallel programming introduces notoriously difficult bugs, usually referred to as concurrency bugs. This article investigates the potential for deviating from the conventional wisdom of writing concurrency bug--free, parallel programs. It explores the benefit of accepting buggy but approximately correct parallel programs by leveraging the inherent tolerance of emerging parallel applications to inaccuracy in computations. Under algorithmic noise tolerance, a new class of concurrency bugs, accuracy bugs, degrade the accuracy of computation (often at acceptable levels) rather than causing catastrophic termination. This study demonstrates how embracing accuracy bugs affects the application output quality and performance and analyzes the impact on execution semantics.