Architects use cycle-by-cycle simulation to evaluate design choices and understand tradeoffs and interactions among design parameters. Efficiently exploring exponential-size design spaces with many interacting parameters remains an open problem: the sheer number of experiments renders detailed simulation intractable. We attack this problem via an automated approach that builds accurate, confident predictive design-space models. We simulate sampled points, using the results to teach our models the function describing relationships among design parameters. The models produce highly accurate performance estimates for other points in the space, can be queried to predict performance impacts of architectural changes, and are very fast compared to simulation, enabling efficient discovery of tradeoffs among parameters in different regions. We validate our approach via sensitivity studies on memory hierarchy and CPU design spaces: our models generally predict IPC with only 1-2% error and reduce required simulation by two orders of magnitude. We also show the efficacy of our technique for exploring chip multiprocessor (CMP) design spaces: when trained on a 1 % sample drawn from a CMP design space with 250K points and up to 55× performance swings among different system configurations, our models predict performance with only 4-5 % error on average. Our approach combines with techniques to reduce time per simulation, achieving net time savings of three-four orders of magnitude.

A key challenge in benchmarking is to predict the performance of an application of interest on a number of platforms in order to determine which platform yields the best performance. This paper proposes an approach for doing this. We measure a number of microarchitecture-independent characteristics from the application of interest, and relate these characteristics to the characteristics of the programs from a previously profiled benchmark suite. Based on the similarity of the application of interest with programs in the benchmark suite, we make a performance prediction of the application of interest. We propose and evaluate three approaches (normalization, principal components analysis and genetic algorithm) to transform the raw data set of microarchitecture-independent characteristics into a benchmark space in which the relative distance is a measure for the relative performance differences. We evaluate our approach using all of the SPEC CPU2000 benchmarks and real hardware performance numbers from the SPEC website. Our framework estimates per-benchmark machine ranks with a 0.89 average and a 0.80 worst case rank correlation coefficient.

Abstract — Modern architecture research relies heavily on detailed pipeline simulation. Simulating the full execution of an industry standard benchmark can take weeks to complete. Simulating the full execution of the whole benchmark suite for one architecture configuration can take months. To address this issue researchers have examined using targetted sampling based on phase behavior to significantly reduce the simulation time of each program in the benchmark suite. However, even with this sampling approach, simulating the full benchmark suite across a large range of architecture designs can take days to weeks to complete. The goal of this paper is to further reduce simulation time for architecture design space exploration. We reduce simulation time by finding similarity between benchmarks and program inputs at the level of samples (100M instructions of execution). This allows us to use a representative sample of execution from one benchmark to accurately represent a sample of execution of other benchmarks and inputs. The end result of our analyis is a small number of sample points of execution. These are selected across the whole benchmark suite in order to accurately represent the complete simulation of the whole benchmark suite for design space exploration. We show that this provides approximately the same accuracy as the SimPoint sampling approach while reducing the number of simulated instructions by a factor of 1.5. I.

Abstract—This paper proposes a methodology for measuring the similarity between programs based on their inherent microarchitecture-independent characteristics, and demonstrates two applications for it: 1) finding a representative subset of programs from benchmark suites and 2) studying the evolution of four generations of SPEC CPU benchmark suites. Using the proposed methodology, we find a representative subset of programs from three popular benchmark suites—SPEC CPU2000, MediaBench, and MiBench. We show that this subset of representative programs can be effectively used to estimate the average benchmark suite IPC, L1 data cache miss-rates, and speedup on 11 machines with different ISAs and microarchitectures—this enables one to save simulation time with little loss in accuracy. From our study of the similarity between the four generations of SPEC CPU benchmark suites, we find that, other than a dramatic increase in the dynamic instruction count and increasingly poor temporal data locality, the inherent program characteristics have more or less remained unchanged. Index Terms—Measurement techniques, modeling techniques, performance of systems, performance attributes. æ 1

Process variations will greatly impact the stability, leakage power consumption, and performance of future microprocessors. These variations are especially detrimental to 6T SRAM (6-transistor static memory) structures and will become critical with continued technology scaling. In this paper, we propose new on-chip memory architectures based on novel 3T1D DRAM (3-transistor, 1-diode dynamic memory) cells. We provide a detailed comparison between 6T and 3T1D designs in the context of a L1 data cache. The effects of physical device variation on a 3T1D cache can be lumped into variation of data retention times. This paper proposes a range of cache refresh and placement schemes that are sensitive to retention time, and we show that most of the retention time variations can be masked by the microarchitecture when using these schemes. We have performed detailed circuit and architectural simulations assuming different degrees of variability in advanced technology nodes, and we show that the resulting memory architecture can tolerate large process variations with little or even no impact on performance when compared to ideal 6T SRAM designs. Furthermore, these designs are robust to memory cell stability issues and can achieve large power savings. These advantages make the new memory architectures a promising choice for on-chip variation-tolerant cache structures required for next generation microprocessors.

Design variability due to die-to-die and within-die process variations has the potential to significantly reduce the maximum operating frequency and the effective yield of high-performance microprocessors in future process technology generations. This variability manifests itself by increasing the frequency variance and decreasing the mean frequency of fabricated chips. In this paper we develop a model for the impact of variability on the performance of multiported SRAM-based structures such as physical register files which are key architectural components that may encounter variability problems. We find that naively resizing or increasing the access latency of these performance critical datapath resources can have frequency benefits, but may incur a significant IPC loss that limits overall system performance. We propose an extension to latency adaptation called port switching which more efficiently exploits the technique to remedy the IPC loss. We find that even under a conservative, worst case study, 18 % mean frequency improvement with less than 5 % IPC loss is possible for the 65nm technology node. Finally, we contrast the impact of die-to-die and within-die variations on chip performance and demonstrate that the proposed technique can compensate the frequency loss mainly due to within-die variations.

Abstract — Composing a representative workload is a crucial step during the design process of a microprocessor. The workload should be composed in such a way that it is representative for the target domain of application and yet, the amount of redundancy in the workload should be minimized as much as possible in order not to overly increase the total simulation time. As a result, there is an important trade-off that needs to be made between workload representativeness and simulation accuracy versus simulation speed. Previous work used statistical data analysis techniques to identify representative benchmarks and corresponding inputs, also called a subset, from a large set of potential benchmarks and inputs. These methodologies measure a number of program characteristics on which Principal Components Analysis (PCA) is applied before identifying distinct program behaviors among the benchmarks using cluster analysis. In this paper we propose Independent Components Analysis (ICA) as a better alternative to PCA as it does not assume that the original data set has a Gaussian distribution, which allows ICA to better find the important axes in the workload space. Our experimental results using SPEC CPU2000 benchmarks show that ICA significantly outperforms PCA in that ICA achieves smaller benchmark subsets that are more accurate than those found by PCA. I.

Increasing demand for power-efficient, high-performance computing requires tuning applications and/or the underlying hardware to improve the mapping between workload heterogeneity and computational resources. To assess the potential benefits of hardware tuning, we propose a framework that leverages synergistic interactions between recent advances in (a) sampling, (b) predictive modeling, and (c) optimization heuristics. This framework enables qualitatively new capabilities in analyzing the performance and power characteristics of adaptive microarchitectures. For the first time, we are able to simultaneously consider high temporal and comprehensive spatial adaptivity. In particular, we optimize efficiency for many, short adaptive intervals and identify the best configuration of 15 parameters, which define a space of 240B points. With frequent sub-application reconfiguration and a fully reconfigurable hardware substrate, adaptive microarchitectures achieve bips 3 /w efficiency gains of up to 5.3x (median 2.4x) relative to their static counterparts already optimized for a given application. This 5.3x efficiency gain is derived from a 1.6x performance gain and 0.8x power reduction. Although several applications achieve a significant fraction of their potential efficiency with as few as three adaptive parameters, the three most significant parameters differ across applications. These differences motivate a hardware substrate capable of comprehensive adaptivity to meet these diverse application requirements.

Efficiently exploring exponential-size architectural design spaces with many interacting parameters remains an open problem: the sheer number of experiments required renders detailed simulation intractable. We attack this via an automated approach that builds accurate predictive models. We simulate sampled points, using results to teach our models the function describing relationships among design parameters. The models can be queried and are very fast, enabling efficient design tradeoff discovery. We validate our approach via two uniprocessor sensitivity studies, predicting IPC with only 1-2 % error. In an experimental study using the approach, training on 1 % of a 250Kpoint CMP design space allows our models to predict performance with only 4-5 % error. Our predictive modeling combines well with techniques that reduce the time taken by each simulation experiment, achieving net time savings of three-four orders of magnitude.

Due to the amount of time required to design a new processor, one set of benchmark programs may be used during the design phase while another may be the standard when the design is finally delivered. Using one benchmark suite to design a processor while using a different, presumably more current, suite to evaluate its ultimate performance may lead to sub-optimal design decisions if there are large differences between the characteristics of the two suites and their respective compilers. We call this change across time “drift”. To evaluate the impact of using yesterday’s benchmark and compiler technology to design tomorrow’s processors, we compare common benchmarks from the SPEC 95 and SPEC 2000 benchmark suites. Our results yield three key conclusions. First, we show that the amount of drift, for common programs in successive SPEC benchmark suites, is significant. In