Power dissipation and thermal issues are increasingly significant in modern processors. As a result, it is crucial that power/performance tradeoffs be made more visible to chip architects and even compiler writers, in addition to circuit designers. Most existing power analysis tools achieve high accuracy by calculating power estimates for designs only after layout or floorplanning are complete. In addition to being available only late in the design process, such tools are often quite slow, which compounds the difficulty of running them for a large space of design possibilities.This paper presents Wattch, a framework for analyzing and optimizing microprocessor power dissipation at the architecture-level. Wattch is 1000X or more faster than existing layout-level power tools, and yet maintains accuracy within 10% of their estimates as verified using industry tools on leading-edge designs. This paper presents several validations of Wattch's accuracy. In addition, we present three examples that demonstrate how architects or compiler writers might use Wattch to evaluate power consumption in their design process.We see Wattch as a complement to existing lower-level tools; it allows architects to explore and cull the design space early on, using faster, higher-level tools. It also opens up the field of power-efficient computing to a wider range of researchers by providing a power evaluation methodology within the portable and familiar SimpleScalar framework.

Wattch: a framework for architectural-level power analysis and optimizations

Author(s): Asanovic, K; Bodik, R; Catanzaro, B; Gebis, J; Husbands, P; Keutzer, K; Patterson, D; Plishker, W; Shalf, J; Williams, SW | Abstract: The recent switch to parallel microprocessors is a milestone in the history of computing. Industry has laid out a roadmap for multicore designs that preserves the programming paradigm of the past via binary compatibility and cache coherence. Conventional wisdom is now to double the number of cores on a chip with each silicon generation. A multidisciplinary group of Berkeley researchers met nearly two years to discuss this change. Our view is that this evolutionary approach to parallel hardware and software may work from 2 or 8 processor systems, but is likely to face diminishing returns as 16 and 32 processor systems are realized, just as returns fell with greater instruction-level parallelism. We believe that much can be learned by examining the success of parallelism at the extremes of the computing spectrum, namely embedded computing and high performance computing. This led us to frame the parallel landscape with seven questions, and to recommend the following: • The overarching goal should be to make it easy to write programs that execute efficiently on highly parallel computing systems • The target should be 1000s of cores per chip, as these chips are built from processing elements that are the most efficient in MIPS (Million Instructions per Second) per watt, MIPS per area of silicon, and MIPS per development dollar. • Instead of traditional benchmarks, use 13 “Dwarfs” to design and evaluate parallel programming models and architectures. (A dwarf is an algorithmic method that captures a pattern of computation and communication.) • “Autotuners” should play a larger role than conventional compilers in translating parallel programs. • To maximize programmer productivity, future programming models must be more human-centric than the conventional focus on hardware or applications. • To be successful, programming models should be independent of the number of processors. • To maximize application efficiency, programming models should support a wide range of data types and successful models of parallelism: task-level parallelism, word-level parallelism, and bit-level parallelism. 1 The Landscape of Parallel Computing Research: A View From Berkeley • Architects should not include features that significantly affect performance or energy if programmers cannot accurately measure their impact via performance counters and energy counters. • Traditional operating systems will be deconstructed and operating system functionality will be orchestrated using libraries and virtual machines. • To explore the design space rapidly, use system emulators based on Field Programmable Gate Arrays (FPGAs) that are highly scalable and low cost. Since real world applications are naturally parallel and hardware is naturally parallel, what we need is a programming model, system software, and a supporting architecture that are naturally parallel. Researchers have the rare opportunity to re-invent these cornerstones of computing, provided they simplify the efficient programming of highly parallel systems.

/pdf/the-landscape-of-parallel-computing-research-a-view-from-1i5pswggnf.pdf

The Landscape of Parallel Computing Research: A View from Berkeley

With increasing clock frequencies and silicon integration, power aware computing has become a critical concern in the design of embedded processors and systems-on-chip. One of the more effective and widely used methods for power-aware computing is dynamic voltage scaling (DVS). In order to obtain the maximum power savings from DVS, it is essential to scale the supply voltage as low as possible while ensuring correct operation of the processor. The critical voltage is chosen such that under a worst-case scenario of process and environmental variations, the processor always operates correctly. However, this approach leads to a very conservative supply voltage since such a worst-case combination of different variabilities is very rare. In this paper, we propose a new approach to DVS, called Razor, based on dynamic detection and correction of circuit timing errors. The key idea of Razor is to tune the supply voltage by monitoring the error rate during circuit operation, thereby eliminating the need for voltage margins and exploiting the data dependence of circuit delay. A Razor flip-flop is introduced that double-samples pipeline stage values, once with a fast clock and again with a time-borrowing delayed clock. A metastability-tolerant comparator then validates latch values sampled with the fast clock. In the event of timing error, a modified pipeline mispeculation recovery mechanism restores correct program state. A prototype Razor pipeline was designed in a 0.18 /spl mu/m technology and was analyzed. Razor energy overhead during normal operation is limited to 3.1%. Analyses of a full-custom multiplier and a SPICE-level Kogge-Stone adder model reveal that substantial energy savings are possible for these devices (up to 64.2%) with little impact on performance due to error recovery (less than 3%).

Razor: a low-power pipeline based on circuit-level timing speculation

Scaling advanced CMOS technology to the next generation improves performance, increases transistor density, and reduces power consumption. Technology scaling typically has three main goals: 1) reduce gate delay by 30%, resulting in an increase in operating frequency of about 43%; 2) double transistor density; and 3) reduce energy per transition by about 65%, saving 50% of power (at a 43% increase in frequency). These are not ad hoc goals; rather, they follow scaling theory. This article looks closely at past trends in technology scaling and how well microprocessor technology and products have met these goals. It also projects the challenges that lie ahead if these trends continue. This analysis uses data from various Intel microprocessors; however, this study is equally applicable to other types of logic designs. Is process technology meeting the goals predicted by scaling theory? An analysis of microprocessor performance, transistor density, and power trends through successive technology generations helps identify potential limiters of scaling, performance, and integration.

Design challenges of technology scaling

The demand for an all-in-one phone with integrated personal information management and data access capabilities is beginning to accelerate. While personal digital assistants (PDAs) with built-in cellular, WiFi, and Voice-Over-IP technologies have the ability to serve these needs in a single package, the rate at which energy is consumed by PDA-based phones is very high. Thus, these devices can quickly drain their own batteries and become useless to their owner.In this paper, we introduce a technique to increase the battery lifetime of a PDA-based phone by reducing its idle power, the power a device consumes in a "standby" state. To reduce the idle power, we essentially shut down the device and its wireless network card when the device is not being used---the device is powered only when an incoming call is received. Using this technique, we can increase the battery lifetime by up to 115%.In this paper, we describe the design of our "wake-on-wireless" energy-saving strategy and the prototype device we implemented. To evaluate our technique, we compare it with alternative approaches. Our results show that our technique can provide a significant lifetime improvement over other technologies.

/pdf/wake-on-wireless-an-event-driven-energy-saving-strategy-for-2s0x8xxwjq.pdf

Wake on wireless: an event driven energy saving strategy for battery operated devices

In this paper we investigate possible ways to improve the energy efficiency of a general purpose microprocessor. We show that the energy of a processor depends on its performance, so we chose the energy-delay product to compare different processors. To improve the energy-delay product we explore methods of reducing energy consumption that do not lead to performance loss (i.e. wasted energy), and explore methods to reduce delay by exploiting instruction level parallelism. We found that careful design reduced the energy dissipation by almost 25%. Pipelining can give approximately a 2/spl times/ improvement in energy-delay product. Superscalar issue, however, does not improve the energy-delay product any further since the overhead required offsets the gains in performance. Further improvements will be hard to come by since a large fraction of the energy (50-80%) is dissipated in the clock network and the on-chip memories. Thus, the efficiency of processors will depend more on the technology being used and the algorithm chosen by the programmer than the micro-architecture.

Energy dissipation in general purpose microprocessors

This paper investigates the effect of lowering the supply and threshold voltages on the energy efficiency of CMOS circuits. Using a first-order model of the energy and delay of a CMOS circuit, we show that lowering the supply and threshold voltage is generally advantageous, especially when the transistors are velocity saturated and the nodes have a high activity factor, In fact, for modern submicron technologies, this simple analysis suggests optimal energy efficiency at supply voltages under 0.5 V. Other process and circuit parameters have almost no effect on this optimal operating point. If there is some uncertainty in the value of the threshold or supply voltage, however, the power advantage of this very low voltage operation diminishes. Therefore, unless active feedback is used to control the uncertainty, in the future the supply and threshold voltage will not decrease drastically, but rather will continue to scale down to maintain constant electric fields.

/pdf/supply-and-threshold-voltage-scaling-for-low-power-cmos-2zwloxdyu4.pdf

Supply and threshold voltage scaling for low power CMOS

Recently there has been a surge of interest in low-power devices and design techniques. While many papers have been published describing power-saving techniques for use in digital systems, trade-offs between the methods are rarely discussed. We address this issue by using an energy-delay metric to compare many of the proposed techniques. Using this metric also provides insight into some of the basic trade-offs in low-power design. The next section describes the energy-loss mechanisms that are present in CMOS circuits, which provides the parameters that must be changed to lower the power dissipation. With these factors in mind, the rest of the paper reviews the energy saving techniques that have been proposed. These proposals fall into one of three main strategies: trade speed for power, do not waste power, and find a lower power problem.

R. Gonzalez

Papers

Energy dissipation in general purpose microprocessors

Supply and threshold voltage scaling for low power CMOS

Low-power digital design