Elbert Hu, Philippe Joubert, Robbie King et al. · IBM Journal of Research and Development · 2001
Adaptive Fast Path Architecture (AFPA) is a software architecture that dramatically improves the efficiency, and therefore the capacity, of Web and other network servers. The architecture includes a RAM-based cache that serves static content and a reverse proxy that can distribute requests for dynamic content to multiple servers. These two mechanisms are combined using a flexible layer-7 (content-based) routing facility. The architecture defines interfaces that allow these generic mechanisms to be exploited to accelerate a variety of application protocols, including HTTP. Efficiency is derived from maximizing the number of requests that are handled entirely within the kernel, using a deferred-interrupt context instead of threads wherever possible. AFPA has been implemented on several server platforms including Microsoft Windows NT® and Windows® 2000, OS/390®, AIX®, and most recently Linux. By conservative estimates, AFPA more than doubles cap acity for serving static content compared to conventional server architectures, and has allowed IBM to establish a leadership position in Web server performance. A prototype implementation of AFPA on Linux delivers more than 10000 SPECweb96 operations per second on a single processor.
Cited by 8
Christoph Hansknecht, Alexander T. Richter, Sebastian Stiller · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · 2018
We develop a fast method to compute an optimal robust shortest path in large networks like road networks, a fundamental problem in traffic and logistics under uncertainty. In the robust shortest path problem we are given an s-t-graph D(V,A) and for each arc a nominal length c(a) and a maximal increase d(a) of its length. We consider all scenarios in which for the increased lengths c(a) + bar{d}(a) we have bar{d}(a) <= d(a) and sum_{a in A} (bar{d}(a)/d(a)) <= Gamma. Each path is measured by the length in its worst-case scenario. A classic result [Bertsimas and Sim, 2003] minimizes this path length by solving (|A| + 1)-many shortest path problems. Easily, (|A| + 1) can be replaced by |Theta|, where Theta is the set of all different values d(a) and 0. Still, the approach remains impractical for large graphs. Using the monotonicity of a part of the objective we devise a Divide and Conquer method to evaluate significantly fewer values of Theta. This methods generalizes to binary linear robust problems. Specifically for shortest paths we derive a lower bound to speed-up the Divide and Conquer of Theta. The bound is based on carefully using previous shortest path computations. We combine the approach with non-preprocessing based acceleration techniques for Dijkstra adapted to the robust case. In a computational study we document the value of different accelerations tried in the algorithm engineering process. We also give an approximation scheme for the robust shortest path problem which computes a (1 + epsilon)-approximate solution requiring O(log(d^ / (1 + epsilon))) computations of the nominal problem where d^ := max d(A) / min (d(A)\{0}).
Cited by 3Open access
Tomasz Sroka · Jagiellonian University Repository (Jagiellonian University) · 2015
Cited by 0Open access
C. Priol, Philippe Magarshack · 1991
The authors describe a highly configurable, compiled, fast data path multiplier implemented in the CSAM ASIC technology-independent library based on the GDT/GENESIL silicon compiler environments. The multiplier compiler, which allows independent control of the width of both operands, has been designed for use in high performance applications in data processing with a one stage pipeline option. The authors present the hardware organization combining the modified Booth decoding method and a tree algorithm to reduce the multiplication delay. It is the first time such a tree algorithm compiler is reported for the data path.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 1
Håkon Jarvis Westergård · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2018
This thesis concerns the implementation of an pre-analysis tool in orienteering for the problem of finding the optimal route choice on a leg in orienteering. The problem is modelled by the framework of level sets and an anisotropic travel time equation is proposed. The analysis tool designed in this thesis is based on solving an anisotropic travel time equation, based on a digital elevation model and symbols imported from an orienteering map, using the fast sweeping method.
Cited by 1Open access
Dominik Schultes · 2005
Cited by 0
Yuan-Fang Li, Janelle Harms, Robert C. Holte · 2007
QoS routing has been shown to be NP-hard. A recent study of its hardness shows that the "worst-case" may not occur in practice (Kuipers mieghem, 2003). This suggests that there may exist fast exact algorithms for the multi-constraint shortest path (MCSP) problem, an instance of QoS routing. Search techniques such as A* and IDA* may solve hard problems exactly in polynomial time. In Li et al. (2005), we deploy the idea of iterative deepening search to design IDA*_MCSP, and show its efficiency by extensive empirical study. In this paper, we show that for infeasible cases, where there is no feasible solution, IDA*_MCSP may not be as efficient as A*Prune. This motivates us to design an algorithm that is efficient in both feasible and infeasible cases. We design an exact MCSP algorithm A*_MCSP, which introduces the state notion and dominance relationship between states. Furthermore, we design an exact MCSP algorithm FringeMCSP. It can be regarded as an integration of IDA*_MCSP and A*_MCSP. Extensive empirical study shows that FringeMCSP has good performance in both feasible and infeasible cases; while IDA*_MCSP still shows its superiority among the proposed MCSP algorithms in feasible cases.
Cited by 17
Daniel Jimenez Department · 2003
Microarchitectural prediction based on neural learning has received increasing attention in recent years. However, neural prediction remains impractical because its superior accuracy over conventional predictors is not enough to offset the cost imposed by its high latency. We present a new neural branch predictor that solves the problem from both directions: it is both more accurate and much faster than previous neural predictors. Our predictor improves accuracy by combining path and pattern history to overcome limitations inherent to previous predictors. It also has much lower latency than previous neural predictors. The result is a predictor with accuracy far superior to conventional predictors but with latency comparable to predictors from industrial designs. Our simulations show that a path-based neural predictor improves the instructions-per-cycle (IPC) rate of an aggressively clocked microarchitecture by 16% over the original perceptron predictor.
Cited by 0
George M. ADELSON-VELSKYa, Alexander Gelbukh, Eugene Levner · 2002
Cited by 1
Robert E. Tarjan · Journal of the ACM · 1981
Let G = (V, E) be a directed graph with a distinguished source vertex s. The single-source path expression problem is to find, for each vertex v, a regular expression P(s, v) which represents the set of all paths in G from s to v A solution to this problem can be used to solve shortest path problems, solve sparse systems of linear equations, and carry out global flow analysis. A method is described for computing path expressions by dwidmg G mto components, computing path expressions on the components by Gaussian elimination, and combining the solutions This method requires O(ma(m, n)) time on a reducible flow graph, where n Is the number of vertices m G, m is the number of edges in G, and a is a functional inverse of Ackermann's function The method makes use of an algonthm for evaluating functions defined on paths in trees. A smapllfied version of the algorithm, which runs in O(m log n) time on reducible flow graphs, is quite easy to implement and efficient m practice
Cited by 282Open access
Elliptic · Bulletin of Miscellaneous Information (Royal Gardens Kew) · 2005
The first Drum'n'Bass mix on Zerinnerung has a mellow start, then it's getting darker and harder and kicks ass like hell. The great cover is designed by Ideology's cover artist Resi.
Cited by 0
Roland Schmid, Roger Wattenhofer · arXiv (Cornell University) · 2019
PermitBFT establishes a permissioned byzantine ledger in the partially synchronous networking model. For n replicas, PermitBFT tolerates up to f < n/3 byzantine replicas. It is the first BFT protocol to achieve a latency of just 2 message delays despite tolerating byzantine replicas throughout the "fast track", as long as they are not the leader. The design of PermitBFT relies on two fundamental concepts. First, in PermitBFT the participating nodes do not wait for a distinguished leader to act and subsequently confirm its actions, but send permits to the next leader proactively. Second, PermitBFT achieves a separation of the decision powers that are usually concentrated on a single leader node. A leader in PermitBFT controls which transactions to include in a new block, but not where to append the block in the block graph.
Cited by 0Open access
Myeong-Mo Gu, Bong-Gi Kim · Journal of the Korea Academia-Industrial cooperation Society · 2016
오버레이 멀티캐스트를 이용하는 응용에서는 노드들의 잦은 탈퇴로 인하여 발생하는 전송 경로의 빠른 재구성이 필요하다. 이를 위해 많은 연구들이 진행되어 왔지만 빠른 경로 복구를 위한 연구는 지속적으로 요구된다. 본 논문에서는 오버레이 멀티캐스트에서 부모노드의 탈퇴로 발생하는 전송 경로 재구성 시간을 개선하고 안정적인 경로를 유지하기 위하여 새로운 빠른 경로 복구 방법을 제안한다. 제안 방법에서는 빠른 경로 복구를 위하여 가상 전송 경로와 후보 부모 노드를 이용한다. 모든 노드들은 전송 경로 상 자신과 비슷한 깊이에 존재하는 노드 및 인접한 노드와 주기적으로 RTT(Round Trip Time)정보를 교환한다. 모든 노드들은 후보 부모 노드 리스트를 가지고 있고 각 노드는 이 교환된 RTT 정보를 리스트에 저장한다. 각 노드들은 경로 복구 시간을 줄이기 위하여 저장된 RTT 정보 순으로 후보 부모 노드들을 결정하고 가상 전송 경로를 구성한다. 구성된 가상 전송 경로를 이용하여 부모 노드의 탈퇴 시 빠르게 전송 경로를 복구할 수 있다. 시뮬레이션 결과는 제안 방법의 복구 시간이 일반적인 방법보다 평균 30%로 더 빨랐음을 보여준다. The rapid reconstruction of a transfer path is caused by frequent secession of nodes in many applications using overlay multicast. For this, however, many studies have been proposed, but a study of fast path recovery is required. This paper proposes a new method for fast path recovery to improve the path reconstruction time and for stable path maintain caused by the sudden secession of the parent node on overlay multicast. The proposed method uses the virtual transfer path and the candidate parent node for the fast path recovery. All nodes exchange periodically the RTT (Round Trip Time) information between the other nodes of similar position itself and neighboring node. All nodes have a candidate parent node list, and each node stores the exchanged RTT information on the list. Each node constructs the virtual transfer path to reduce the recovery time after deciding a candidate parent node that is one of them by the RTT order. In this way, when the parent node is seceded, all nodes can recover the transfer path quickly using the virtual transfer path. The simulation result showed that the recovery time of the proposed method is an average 30% faster than the known method.
Cited by 0Open access
Daniel A. Jiménez · 2004
Microarchitectural prediction based on neural learning has received increasing attention in recent years. However, neural prediction remains impractical because its superior accuracy over conventional predictors is not enough to offset the cost imposed by its high latency. We present a new neural branch predictor that solves the problem from both directions: it is both more accurate and much faster than previous neural predictors. Our predictor improves accuracy by combining path and pattern history to overcome limitations inherent to previous predictors. It also has much lower latency than previous neural predictors. The result is a predictor with accuracy for superior to conventional predictors but with latency comparable to predictors from industrial designs. Our simulations show that a path-based neural predictor improves the instructions-per-cycle (IPC) rate of an aggressively clocked microarchitecture by 16% over the original perceptron predictor.
Cited by 71
Jeongjin Lee, Gyehyun Kim, Kyoung Ho Lee et al. · Computers in Biology and Medicine · 2008
Cited by 5
Longquan Dai, Feihu Zhang, Xing Mei et al. · IEEE Signal Processing Letters · 2014
We propose a fast minimax path-based depth interpolation method. The algorithm computes for each target pixel varying contributions from reliable depth seeds, and weighted averaging is used to interpolate missing depths. Compared with state-of-the-art joint geodesic upsampling method which selects the K nearest seeds to interpolate missing depths with O(Kn) complexity, our method does not need to limit the number of seeds to K and reduces the computational complexity to O(n). In addition, the minimax path chooses a path with the smallest maximum immediate pairwise pixel difference on it, so it tends to preserve sharp depth discontinuities better. In contrast to the results of previous depth upsampling algorithms, our approach can provide accurate depths with fewer artifacts.
Cited by 6
Xin Liu, Chengping Zhou, Mingyue Ding et al. · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
A new path planning method for UAV in static workspace is presented. The method can find a nearly optimal path in short time which satisfies the UAV kinematic constraints. The method makes use of the skeletons to construct the graph of the planning space considering the configuration of the obstacles and utilizes the graph to find a shortest collision-free path, and a novel technique is utilized to convert the free path into a feasible path. The method can be applied to different applications and easy to be implemented. Experimental results showed that the path planning can be done in a fraction of second on a contemporary workstation (2-3 seconds) under the condition of satisfying the kinematic constraints.
Cited by 0
Ptfe}ga?h?%r%e H $g$n, James Kuffner, Koichi Nishiwaki et al. · 2007
We present an interactive manipulation path planner under development for providing high-level task-based control for humanoid robots. Using a graphical simulation environment, the software is designed to facilitate the automation of object grasping and manipulation tasks. A randomized path planner is used to directly search the free configuration space of the robot for a collision-free manipulation path. Experimental results are shown using a model of the dynamic humanoid &quot;H6&quot; robot in a virtual world. Key Words: Humanoid Robot, Object Manipulation, Path Planning, Randomized Algorithm Fig.1 The virtual and real dynamic humanoid &apos;H6&apos;. 1. Introduction As the technology and algorithms for real-time 3D vision and tactile sensing improve, humanoid robots will be able to perform tasks that involve complex interactions with the environment (e.g. grasping and manipulating objects). The enabling software for such tasks includes motion planning for obstacle avoidance, and integrating pl...
Cited by 0
Rare & Special e-Zone (The Hong Kong University of Science and Technology) · 2017
Cited by 0
Clément Pêtrès, Yan Pailhas, Yvan Pétillot et al. · 2005
In this paper, new tools for obstacle avoidance and path planning for underwater vehicles are presented. The authors' technique, based on a level set formulation of the path planning problem, extracts optimal paths from complex and continuous environments in a complete and consistent manner. Fast marching algorithm is known to be efficient for finding cost optimal path in mobile robotics because of its reliability, precision, and simple implementation. Fast marching algorithm originally propagates a wave front to isotropically explore the space. We propose an anisotropic version of fast marching by adding directional constraints in a cost function to minimize. We then propose a path planning method able to deal with vectorial fields of force for the first time. Furthermore we explore the relation between the curvature of the optimal path and the cost function generated from scalar and vectorial constraints. This a priori knowledge of the influence of the environment on the final path's curvature allows us to propose a solution to make sure a path is reachable by the vehicle according to its kinematics. A multi-resolution scheme based on an adaptive mesh generation is eventually introduced to speed up the overall algorithm. Results are shown computed from real and simulated underwater environments.
Cited by 81