ALEA: Fine-Grain Energy Profiling with Basic Block Sampling

Energy efficiency is an essential requirement for all contemporary computing systems. We thus need tools to measure the energy consumption of computing systems and to understand how workloads affect it. Significant recent research effort has targeted direct power measurements on production computing systems using on-board sensors or external instruments. These direct methods have in turn guided studies of software techniques to reduce energy consumption via workload allocation and scaling. Unfortunately, direct energy measurements are hampered by the low power sampling frequency of power sensors. The coarse granularity of power sensing limits our understanding of how power is allocated in systems and our ability to optimize energy efficiency via workload allocation.
We present ALEA, a tool to measure power and energy consumption at the granularity of basic blocks, using a probabilistic approach. ALEA provides fine-grained energy profiling via sta- tistical sampling, which overcomes the limitations of power sens- ing instruments. Compared to state-of-the-art energy measurement tools, ALEA provides finer granularity without sacrificing accuracy. ALEA achieves low overhead energy measurements with mean error rates between 1.4% and 3.5% in 14 sequential and paral- lel benchmarks tested on both Intel and ARM platforms. The sampling method caps execution time overhead at approximately 1%. ALEA is thus suitable for online energy monitoring and optimization. Finally, ALEA is a user-space tool with a portable, machine-independent sampling method. We demonstrate two use cases of ALEA, where we reduce the energy consumption of a k-means computational kernel by 37% and an ocean modelling code by 33%, compared to high-performance execution baselines, by varying the power optimization strategy between basic blocks.

采用基本块规约的控制流检测性能优化方法

控制流检测是抵御单粒子事件的有效手段之一.目前的主流方法是采用嵌入式签名技术, 但是该技术引入的检测指令过多, 导致程序效率低下. 本文使用基本块规约的技术, 在原基本块的基础上, 选择合适的约束量重新划分基本块, 减少引入的检测指令. 与8个常见算法的性能比较表明, 该方法在软错误检测覆盖率基本不变的前提下,能有效提高目标程序效率.

Toward Effective Scalar Hardware for Highly Vectorizable Applications

The performance of a program will ultimately be limited by its serial (scalar) portion, as pointed out by Amdahl′s Law. Reported studies thus far of instruction-level parallelism have mixed data-parallel program portions with scalar program portions, often leading to contradictory and controversial results. We report an instruction-level behavioral characterization of scalar code containing minimal data-parallelism, extracted from highly vectorized programs of the PERFECT benchmark suite running on a Cray Y-MP system. We classify scalar basic blocks according to their instruction mix, characterize the data dependencies seen in each class, and, as a first step, measure the maximum intrablock instruction-level parallelism available. We observe skewed rather than balanced instruction distributions in scalar code and in individual basic block classes of scalar code; nonuniform distribution of parallelism across instruction classes; and, as expected, limited available intrablock parallelism. We identify frequently occurring data-dependence patterns and discuss new instructions to reduce latency. Toward effective scalar hardware, we study latency-pipelining trade-offs and restricted multiple instruction issue mechanisms.

基于基本块的控制流检错方法优化研究

系统的高可靠性是研究航空航天领域的一个重要指标. 由于太空环境的特殊性, 辐射和高能粒子会造成计算机系统的出现瞬时性错误, 这种错误被称作软错误, 它对航空航天器件造成了很大的影响, 严重降低系统的可靠性. 检测和防护这种软错误是航空航天系统中的重要研究方向之一. 软错误的检测和防护包括硬件防护与检错, 软硬件混合检错以及纯软件检错等. 随着商用器件的广泛使用, 与之相配合的各种软错误软件检错方法开始得到深入的研究, 在各种软件检错方法中, 控制流检测是抵御单粒子事件的有效手段之一.目前的主流方法是采用嵌入式签名技术, 但是该技术引入的检测指令过多, 导致程序效率低下. 本文从总结控制流检测技术的共同点出发, 分析该技术导致效率低下的原因：由于基本块定义的约束导致程序中基本块过多, 进而在代码注入过程中引入过多的判断及跳转指令, 导致程序效率低下. 本文针对这种情况, 提出了一种基于源代码分析的基本块规约的方法. 该方法通过修改基本块定义的约束, 使在新的基本块定义下每个基本块能够容纳更多的指令, 减少检测指令的注入, 提高效率；并且在新的基本块定义下, 原来的控制流检错方法仍可以不加修改的直接应用于新的基本块定义上. 该方法能在不修改benchmark源代码以及控制流检测方法的基础上, 选择合适的约束量重新划分基本块, 减少引入的检测指令. 本文中使用该方法以ECCA, CFCSS和RSCFC三个控制流检错方法作为验证对象, 使用这3种控制流检错方法, 在不同的约束量作用下, 对8个常见算法的benchmark进行了软错误覆盖率测试和效率测试. 多次实验数据表明, 该方法在提高检错算法效率的同时, 能够保持软错误检错的覆盖率基本不变. 在对控制流检错算法进行优化的同时, 本文还完成了相应的控制流分析工具, 基于模拟器的错误注入和代码片段执行时间检测工具等. 有效的对优化算法进行了评估和测试.

On the fly type specialization without type analysis

Les langages de programmation typés dynamiquement tels que JavaScript et Python repoussent la vérification de typage jusqu’au moment de l’exécution. Afin d’optimiser la performance de ces langages, les implémentations de machines virtuelles pour langages dynamiques doivent tenter d’éliminer les tests de typage dynamiques redondants. Cela se fait habituellement en utilisant une analyse d’inférence de types. Cependant, les analyses de ce genre sont souvent coûteuses et impliquent des compromis entre le temps de compilation et la précision des résultats obtenus. Ceci a conduit à la conception d’architectures de VM de plus en plus complexes. Nous proposons le versionnement paresseux de blocs de base, une technique de compilation à la volée simple qui élimine efficacement les tests de typage dynamiques redondants sur les chemins d’exécution critiques. Cette nouvelle approche génère paresseusement des versions spécialisées des blocs de base tout en propageant de l’information de typage contextualisée. Notre technique ne nécessite pas l’utilisation d’analyses de programme coûteuses, n’est pas contrainte par les limitations de précision des analyses d’inférence de types traditionnelles et évite la complexité des techniques d’optimisation spéculatives. Trois extensions sont apportées au versionnement de blocs de base afin de lui donner des capacités d’optimisation interprocédurale. Une première extension lui donne la possibilité de joindre des informations de typage aux propriétés des objets et aux variables globales. Puis, la spécialisation de points d’entrée lui permet de passer de l’information de typage des fonctions appellantes aux fonctions appellées. Finalement, la spécialisation des continuations d’appels permet de transmettre le type des valeurs de retour des fonctions appellées aux appellants sans coût dynamique. Nous démontrons empiriquement que ces extensions permettent au versionnement de blocs de base d’éliminer plus de tests de typage dynamiques que toute analyse d’inférence de typage statique.

On the fly type specialization without type analysis

Les langages de programmation typés dynamiquement tels que JavaScript et Python repoussent la vérification de typage jusqu’au moment de l’exécution. Afin d’optimiser la performance de ces langages, les implémentations de machines virtuelles pour langages dynamiques doivent tenter d’éliminer les tests de typage dynamiques redondants. Cela se fait habituellement en utilisant une analyse d’inférence de types. Cependant, les analyses de ce genre sont souvent coûteuses et impliquent des compromis entre le temps de compilation et la précision des résultats obtenus. Ceci a conduit à la conception d’architectures de VM de plus en plus complexes. Nous proposons le versionnement paresseux de blocs de base, une technique de compilation à la volée simple qui élimine efficacement les tests de typage dynamiques redondants sur les chemins d’exécution critiques. Cette nouvelle approche génère paresseusement des versions spécialisées des blocs de base tout en propageant de l’information de typage contextualisée. Notre technique ne nécessite pas l’utilisation d’analyses de programme coûteuses, n’est pas contrainte par les limitations de précision des analyses d’inférence de types traditionnelles et évite la complexité des techniques d’optimisation spéculatives. Trois extensions sont apportées au versionnement de blocs de base afin de lui donner des capacités d’optimisation interprocédurale. Une première extension lui donne la possibilité de joindre des informations de typage aux propriétés des objets et aux variables globales. Puis, la spécialisation de points d’entrée lui permet de passer de l’information de typage des fonctions appellantes aux fonctions appellées. Finalement, la spécialisation des continuations d’appels permet de transmettre le type des valeurs de retour des fonctions appellées aux appellants sans coût dynamique. Nous démontrons empiriquement que ces extensions permettent au versionnement de blocs de base d’éliminer plus de tests de typage dynamiques que toute analyse d’inférence de typage statique.

Anti-mosquito midgut antibodies block development of Plasmodium falciparum and Plasmodium vivax in multiple species of Anopheles mosquitoes and reduce vector fecundity and survivorship

The mosquito midgut plays a central role in the sporogonic development of malaria parasites. We have found that polyclonal sera, produced against mosquito midguts, blocked the passage of Plasmodium falciparum ookinetes across the midgut, leading to a significant reduction of infections in mosquitoes. Anti-midgut mAbs were produced that display broad-spectrum activity, blocking parasite development of both P. falciparum and Plasmodium vivax parasites in five different species of mosquitoes. In addition to their parasite transmission-blocking activity, these mAbs also reduced mosquito survivorship and fecundity. These results reveal that mosquito midgut-based antibodies have the potential to reduce malaria transmission in a synergistic manner by lowering both vector competence, through transmission-blocking effects on parasite development, and vector abundance, by decreasing mosquito survivorship and egg laying capacity. Because the intervention can block transmission of different malaria parasite species in various species of mosquitoes, vaccines against such midgut receptors may block malaria transmission worldwide.

Restarting trains under moving block signalling : an expert system approach

A high peak power demand at substations will result under Moving Block Signalling (MBS) when a dense queue of trains begins to start from a complete stop at the same time in an electrified railway system. This may cause the power supply interruption and in turn affect the train service substantially. In a recent study, measures of Starting Time Delay (STD) and Acceleration Rate Limit (ARL) are the possible approaches to reduce the peak power demand on the supply system under MBS. Nevertheless, there is no well-defined relationship between the two measures and peak power demand reduction (PDR). In order to attain a lower peak demand at substations on different traffic conditions and system requirements, an expert system is one of the possible approaches to procure the appropriate use of peak demand reduction measures. The main objective of this paper is to study the effect of the train re-starting strategies on the power demand at substations and the time delay suffered by the trains with the aid of computer simulation. An expert system is a useful tool to select various adoptions of STD and ARL under different operational conditions and system requirements.

Linearity within the SMS4 Block Cipher

We present several new observations on the SMS4 block cipher, and discuss their cryptographic significance. The crucial observation is the existence of fixed points and also of simple linear relationships between the bits of the input and output words for each component of the round functions for some input words. This implies that the non-linear function T of SMS4 does not appear random and that the linear transformation provides poor diffusion. Furthermore, the branch number of the linear transformation in the key scheduling algorithm is shown to be less than optimal. The main security implication of these observations is that the round function is not always non-linear. Due to this linearity, it is possible to reduce the number of effective rounds of SMS4 by four. We also investigate the susceptibility of SMS4 to further cryptanalysis. Finally, we demonstrate a successful differential attack on a slightly modified variant of SMS4. These findings raise serious questions on the security provided by SMS4.

Ten basic factors to identify suitable subcontractors for construction projects

With unpredictable workloads and a need for a multitude of specialized skills, many main contractors rely heavily on subcontracting to reduce their risks (Bresnen et al., 1985; Beardsworth et al., 1988). This is especially the case In Hong Kong, where the average direct labour content accounts for only around 1% of the total contract sum (Lai, 1987). Extensive usage of subcontracting is also reported in many other countries, including the UK (Gray and Flanagan, 1989) and Japan (Bennett et al., 1987). In addition, and depending upon the scale and complexity of works, it is not uncommon for subcontractors to further sublet their works to lower tier(s) subcontractors. Richter and Mitchell (1982) argued that main contractors can obtain a higher profit margin by reducing their performance costs by subcontracting work to those who have the necessary resources to perform the work more efficiently and economically. Subcontracting is also used strategically to allow firms to employ a minimum work force under fluctuating demand (Usdiken and Sözen, 1985). Through subcontracting, the risks of main contractors are also reduced, as errors in estimating or additional costs caused by delays or extra labour requirements can be absorbed by the subcontractors involved (Woon and Ofori, 2000). Despite these benefits, the quality of work can suffer when incapable or inexperienced subcontractors are employed. Additional problems also exist in the form of bid shopping, unclear accountability, and high fragmentation (Palaneeswaran et al., 2002). A recent CIB TG 23 International Conference, October 2003, Hong Kong report produced by the Hong Kong Construction Industry Review Committee (CIRC) points to development of a framework to help distinguish between capable and incapable subcontractors (Tang, 2001). This paper describes research aims at identifying and prioritising criteria for use in such a framework.

The increasing cost of the basic foods required to promote health in Queensland

Objective: To assess changes in the cost and availability of a standard basket of healthy food items (the Healthy Food Access Basket [HFAB]) in Queensland over time. Design and participants: A series of four cross-sectional surveys (in 1998, 2000, 2001 and 2004) describing the cost and availability of foods in the HFAB over time. In the latest survey, 97 Queensland food stores across the five Australian Bureau of Statistics remoteness categories were compared. Main outcome measures: Cost comparisons for HFAB items by remoteness category for the 97 stores surveyed in 2004; changes in cost and availability of foods in the 81 stores surveyed since 2000; comparisons of food prices in the 56 stores surveyed in 1998, 2000, 2001 and 2004. Results: In 2004, the Queensland mean cost of the HFAB was $395.28 a fortnight. The cost of the HFAB was 29.6%($113.89) higher in “very remote” areas than in “major cities” (P<0.001). Between 2001 and 2004, the Queensland mean cost of the HFAB increased by 14.0% ($48.45), while in very remote areas the cost increased by 18.0% ($76.93) (P<0.001). Since 2000, the annualised per cent increase in cost of the HFAB has been higher than the increase in Consumer Price Index for food in Brisbane. The cost of healthy foods has risen more than the cost of some less nutritious foods, so that the latter are now relatively more affordable. Conclusions: Consumers, particularly those in very remote locations, need to pay substantially more for basic healthy foods than they did a few years ago. Higher prices are likely to be a barrier to good health among people of low socioeconomic status and other vulnerable groups. Interventions to make basic healthy food affordable and accessible to all would help reduce the high burden of chronic disease.

Brief guided imagery and body scanning interventions reduce food cravings

Elaborated Intrusion (EI) Theory proposes that cravings occur when involuntary thoughts about food are elaborated; a key part of elaboration is affectively-charged imagery. Craving can be weakened by working memory tasks that block imagery. EI Theory predicts that cravings should also be reduced by preventing involuntary thoughts being elaborated in the first place. Research has found that imagery techniques such as body scanning and guided imagery can reduce the occurrence of food thoughts. This study tested the prediction that these techniques also reduce craving. We asked participants to abstain from food overnight, and then to carry out 10 min of body scanning, guided imagery, or a control mind wandering task. They rated their craving at 10 points during the task on a single item measure, and before and after the task using the Craving Experience Questionnaire. While craving rose during the task for the mind wandering group, neither the guided imagery nor body scanning group showed an increase. These effects were not detected by the CEQ, suggesting that they are only present during the competing task. As they require no devices or materials and are unobtrusive, brief guided imagery strategies might form useful components of weight loss programmes that attempt to address cravings.

Female greater wax moths reduce sexual display behavior in relation to the potential risk of predation by echolocating bats

Female greater wax moths Galleria mellonella display by wing fanning in response to bursts of ultrasonic calls produced by males. The temporal and spectral characteristics of these calls show some similarities with the echolocation calls of bats that emit frequency-modulated (FM) signals. Female G. mellonella therefore need to distinguish between the attractive signals of male conspecifics, which may lead to mating opportunities, and similar sounds made by predatory bats. We therefore predicted that (1) females would display in response to playbacks of male calls; (2) females would not display in response to playbacks of the calls of echolocating bats (we used the calls of Daubenton's bat Myotis daubentonii as representative of a typical FM echolocating bat); and (3) when presented with male calls and bat calls during the same time block, females would display more when perceived predation risk was lower. We manipulated predation risk in two ways. First, we varied the intensity of bat calls to represent a nearby (high risk) or distant (low risk) bat. Second, we played back calls of bats searching for prey (low risk) and attacking prey (high risk). All predictions were supported, suggesting that female G. mellonella are able to distinguish conspecific male mating calls from bat calls, and that they modify display rate in relation to predation risk. The mechanism (s) by which the moths separate the calls of bat and moth must involve temporal cues. Bat and moth signals differ considerably in duration, and differences in duration could be encoded by the moth's nervous system and used in discrimination.