This paper considers using predictions in the context of the online Joint Replenishment Problem with Deadlines (JRP-D). Prior work includes asymptotically optimal competitive ratios of $O(1)$ for the clairvoyant setting and $O(\sqrt{n})$ of the nonclairvoyant setting, where $n$ is the number of items. The goal of this paper is to significantly reduce the competitive ratio for the nonclairvoyant case by leveraging predictions: when a request arrives, the true deadline of the request is not revealed, but the algorithm is given a predicted deadline. The main result is an algorithm whose competitive ratio is $O(\min(η^{1/3}\log^{2/3}(n), \sqrtη, \sqrt{n}))$, where $n$ is the number of item types and $η\leq n^2$ quantifies how flawed the predictions are in terms of the number of ``instantaneous item inversions.'' Thus, the algorithm is robust, i.e., it is never worse than the nonclairvoyant solution, and it is consistent, i.e., if the predictions exhibit no inversions, then the algorithm behaves similarly to the clairvoyant algorithm. Moreover, if the error is not too large, specifically $η< o(n^{3/2}/\log^2(n))$, then the algorithm obtains an asymptotically better competitive ratio than the nonclairvoyant algorithm. We also show that all deterministic algorithms falling in a certain reasonable class of algorithms have a competitive ratio of $Ω(η^{1/3})$, so this algorithm is nearly the best possible with respect to this error metric.

In recent decades, cloud computing as a service has emerged as a major computing paradigm. These services (e.g., Amazon EC2, Google Compute Engine, Azure Virtual Machines) all offer variations of the following basic model for how storage works: a compute instance can choose between placing data on something that resembles a local disk (e.g., Amazon EBS, Google Block Store, Azure Managed Disks) versus what we will refer to as a global disk (e.g., Amazon S3, Google GCS, Azure Blob Storage). The disks are distinguished by two features:

The classic Earliest Deadline First (EDF) algorithm is widely studied and used due to its simplicity and strong theoretical performance, but has not been rigorously analyzed for systems where jobs may execute critical sections protected by shared locks. Analyzing such systems is often challenging due to unpredictable delays caused by contention. In this paper, we propose a straightforward generalization of EDF, called EDF-Block. In this generalization, the critical sections are executed non-preemptively, but scheduling and lock acquisition priorities are based on EDF. We establish lower bounds on the speed augmentation required for any non-clairvoyant scheduler (EDF-Block is an example of non-clairvoyant schedulers) and for EDF-Block, showing that EDF-Block requires at least 4.11× speed augmentation for jobs and 4× for tasks. We then provide an upper bound analysis, demonstrating that EDF-Block requires speedup of at most 6 to schedule all feasible job and task sets.

Abstract. Contention resolution addresses the problem of coordinating access to a shared communication channel. Time is discretized into synchronized slots, and a packet transmission can be made in any slot. A packet succeeds if it is the only packet transmitted during that slot. If two or more packets are sent in the same slot, then these packets collide and fail. Listening on the channel during a slot provides ternary feedback, indicating whether that slot had (0) silence, (1) a successful transmission, or (2+) noise. No other feedback or exchange of information is available to packets. Packets are (adversarially) injected into the system over time. A packet departs the system once it succeeds. The goal is to ensure all packets succeed, while optimizing throughput, which entails optimizing the fraction of successful slots. Most prior contention resolution algorithms with constant throughput require a short feedback loop, in the sense that a packet’s sending probability in slot [Formula: see text] is fully determined by its internal state at slot [Formula: see text] and the channel feedback at slot [Formula: see text]. This paper answers the question of whether these short feedback loops are necessary; that is, how often must listening and updating occur in order to achieve constant throughput? A shared channel can also suffer random or adversarial noise (modeled as jamming), even when no packets are actually sent. How does noise affect our goal of long feedback loops/energy efficiency? Tying these questions together, we ask the following: What does a contention-resolution algorithm have to sacrifice to reduce channel accesses? Must we give up on constant throughput? What about robustness to noise? Here, we show that we need not concede anything by presenting an algorithm with the following guarantees. Suppose there are [Formula: see text] packets arriving over time and [Formula: see text] jammed slots, where the input is determined by an adaptive adversary. With high probability in [Formula: see text], our algorithm guarantees [Formula: see text] throughput and [Formula: see text] channel accesses (sends or listens) per packet. We also have analogous guarantees when the input stream is infinite—we prove implicit throughput bounds of [Formula: see text] for all time slots [Formula: see text], and this translates to [Formula: see text] guaranteed throughput for any slot [Formula: see text] where the implicit throughput is sufficiently small in [Formula: see text]. As a special case, these throughput results give rise to adversarial-queuing theory guarantees.

Shared communication media are widely used in many applications including safety-critical applications. However, noise and transient errors can cause transmission failures. We consider the problem of designing and minimizing the length of fault-tolerant static schedules for transmitting messages in these media provided the number of errors fall below some upper bound. To transmitnmessages in a medium while tolerating a maximum offfaults, prior work had shown how to construct schedules which had a fault tolerance overhead ofnf/2. In this paper, we provide an efficient constructive algorithm for producing a schedule for n messages with total lengthn+O(f2log2n) that can toleratefmedium errors. We also provide an algorithm for randomly generating fault-tolerant schedules with lengthn+O(flog(f) log(n)) as well as a technique for quickly verifying these on reasonably small inputs.

Contention resolution addresses the problem of coordinating access to a shared communication channel. Time is discretized into synchronized slots, and a packet transmission can be made in any slot. A packet is successfully sent if no other packet is also transmitted during that slot. If two or more packets are sent in the same slot, then these packets collide and fail. Listening on the channel during a slot provides ternary feedback, indicating whether that slot had (0) silence, (1) a successful transmission, or (2+) noise. No other feedback or exchange of information is available to packets. Packets are (adversarially) injected into the system over time. A packet departs the system once it is successfully sent. The goal is to send all packets while optimizing throughput, which is roughly the fraction of successful slots.

This paper presents a randomized algorithm for single-source shortest paths on directed graphs with real (both positive and negative) edge weights. Given an input graph with n vertices and m edges, the algorithm completes in Õ(mn8/9) time with high probability. For real-weighted graphs, this result constitutes the first asymptotic improvement over the classic O(mn)-time algorithm variously attributed to Shimbel, Bellman, Ford, and Moore.

Learning Electronic Health Records (EHRs) representation is a preeminent yet under-discovered research topic. It benefits various clinical decision support applications, e.g., medication outcome prediction or patient similarity search. Current approaches focus on task-specific label supervision on vectorized sequential EHR, which is not applicable to large-scale unsupervised scenarios. Recently, contrastive learning has shown great success in self-supervised representation learning problems. However, complex temporality often degrades the performance. We propose Graph Kernel Infomax, a self-supervised graph kernel learning approach on the graphical representation of EHR, to overcome the previous problems. Unlike the state-of-the-art, we do not change the graph structure to construct augmented views. Instead, we use Kernel Subspace Augmentation to embed nodes into two geometrically different manifold views. The entire framework is trained by contrasting nodes and graph representations on those two manifold views through the commonly used contrastive objectives. Empirically, using publicly available benchmark EHR datasets, our approach yields performance on clinical downstream tasks that exceeds the state-of-the-art. Theoretically, the variation in distance metrics naturally creates different views as data augmentation without changing graph structures. Practically, our method is non-ad hoc and confirms superior performance on commonly used graph benchmark datasets.

The best known solutions for k -message broadcast in dynamic networks of size n require Ω ( nk ) rounds. In this article, we see if these bounds can be improved by smoothed analysis. To do so, we study perhaps the most natural randomized algorithm for disseminating tokens in this setting: at every timestep, choose a token to broadcast randomly from the set of tokens you know. We show that with even a small amount of smoothing (i.e., one random edge added per round), this natural strategy solves k -message broadcast in \(\tilde{O}(n+k^3)\) rounds, with high probability, beating the best known bounds for \(k=o(\sqrt {n})\) and matching the Ω ( n + k ) lower bound for static networks for k = O ( n 1/3 ) (ignoring logarithmic factors). In fact, the main result we show is even stronger and more general: Given ℓ-smoothing (i.e., ℓ random edges added per round), this simple strategy terminates in \(O(kn^{2/3}\log ^{1/3}(n)\ell ^{-1/3})\) rounds. We then prove this analysis close to tight with an almost-matching lower bound. To better understand the impact of smoothing on information spreading, we next turn our attention to static networks, proving a tight bound of \(\tilde{O}(k\sqrt {n})\) rounds to solve k -message broadcast, which is better than what our strategy can achieve in the dynamic setting. This confirms the intuition that although smoothed analysis reduces the difficulties induced by changing graph structures, it does not eliminate them altogether. Finally, we apply tools developed to support our smoothed analysis to prove an optimal result for k -message broadcast in so-called well-mixed networks in the absence of smoothing. By comparing this result to an existing lower bound for well-mixed networks, we establish a formal separation between oblivious and strongly adaptive adversaries with respect to well-mixed token spreading, partially resolving an open question on the impact of adversary strength on the k -message broadcast problem.

This paper presents a randomized parallel single-source shortest paths (SSSP) algorithm for directed graphs with non-negative integer edge weights that solves the problem exactly in O(m) work and n1/2+o(1) span, with high probability. All previous exact SSSP algorithms with nearly linear work have linear span, even for undirected unweighted graphs. Our main technical contribution is to show a reduction from the exact SSSP to directed hopsets [6] using the iterative gradual rounding technique [9]. An (h, ε)-hopset is a set of weighted edges (sometimes called shortcuts) that when added to the graph admit h-hop paths with weights no more than (1 + ε) times the true shortest path distances. Furthermore, we show how to combine this algorithm with Forster and Nanongkai's framework [15] to improve the distributed exact SSSP algorithm. Specifically, we obtain an -round algorithm in the CONGEST model for exact SSSP in directed graphs with non-negative integer edge weights, where D is the unweighted diameter of the underlying undirected graph.

The active-time scheduling problem considers the problem of scheduling preemptible jobs with windows (release times and deadlines) on a parallel machine that can schedule up to g jobs during each timestep. The goal in the active-time problem is to minimize the number of active steps, i.e., timesteps in which at least one job is scheduled.

This paper presents a parallel version of Goldberg's algorithm for the problem of single-source shortest paths with integer (including negatives) edge weights. Given an input graph with n vertices, m edges, and integer weights ≥-N, our algorithms solves the problem with Õ(m √n log N) work and n5/4+o(1) log N span, both with high probability. Our algorithm thus has work similar to Goldberg's algorithm while also achieving at least m1/4-o(1) parallelism. To generate our parallel version of Goldberg's algorithm, we solve two specific distance-limited shortest-path problems, both with work Õ(m) and span √L · n1/2+o(1), where L is the distance limit.

The best known solutions for $k$-message broadcast in dynamic networks of size $n$ require $Ω(nk)$ rounds. In this paper, we see if these bounds can be improved by smoothed analysis. We study perhaps the most natural randomized algorithm for disseminating tokens in this setting: at every time step, choose a token to broadcast randomly from the set of tokens you know. We show that with even a small amount of smoothing (one random edge added per round), this natural strategy solves $k$-message broadcast in $\tilde{O}(n+k^3)$ rounds, with high probability, beating the best known bounds for $k=o(\sqrt{n})$ and matching the $Ω(n+k)$ lower bound for static networks for $k=O(n^{1/3})$ (ignoring logarithmic factors). In fact, the main result we show is even stronger and more general: given $\ell$-smoothing (i.e., $\ell$ random edges added per round), this simple strategy terminates in $O(kn^{2/3}\log^{1/3}(n)\ell^{-1/3})$ rounds. We then prove this analysis close to tight with an almost-matching lower bound. To better understand the impact of smoothing on information spreading, we next turn our attention to static networks, proving a tight bound of $\tilde{O}(k\sqrt{n})$ rounds to solve $k$-message broadcast, which is better than what our strategy can achieve in the dynamic setting. This confirms that although smoothed analysis reduces the difficulties induced by changing graph structures, it does not eliminate them altogether. Finally, we apply our tools to prove an optimal result for $k$-message broadcast in so-called well-mixed networks in the absence of smoothing. By comparing this result to an existing lower bound for well-mixed networks, we establish a formal separation between oblivious and strongly adaptive adversaries with respect to well-mixed token spreading, partially resolving an open question on the impact of adversary strength on the $k$-message broadcast problem.

This article presents I/O-efficient algorithms for topologically sorting a directed acyclic graph and for the more general problem identifying and topologically sorting the strongly connected components of a directed graph G = ( V, E ). Both algorithms are randomized and have I/O-costs O ( sort ( E ) · poly(log V)), with high probability, where sort ( E ) = O( E / B log M / B ( E/B )) is the I/O cost of sorting an | E |-element array on a machine with size- B blocks and size- M cache/internal memory. These are the first algorithms for these problems that do not incur at least one I/O per vertex, and as such these are the first I/O-efficient algorithms for sparse graphs. By applying the technique of time-forward processing, these algorithms also imply I/O-efficient algorithms for most problems on directed acyclic graphs, such as shortest paths, as well as the single-source reachability problem on arbitrary directed graphs.

This brief announcement presents an algorithm for (1+ε) approximate single-source shortest paths for directed graphs with non-negative real edge weights in the CONGEST model that runs in Õ ((n^1/2 +D+n^2/5+o(1) D^2/5 )log W / ε^2) rounds, where W is the ratio between the largest and smallest non-zero edge weights.

Shared communication media are widely used in many applications including safety critical applications such as control systems on flights or autonomous vehicles. Noise and transient errors can cause transmission failures. We consider the problem of designing fault tolerant static schedules for transmitting messages in these media. In particular, we assume that the schedule of transmission over all messages must be computed in advance and must guarantee that all messages will be delivered as long as the number of medium errors falls below a provided upper bound, regardless of when the medium errors occur. It is crucial that the messages be delivered in a timely manner, and hence we are interested in minimizing the length of the schedule that achieves the desired level of fault tolerance. In this paper, we provide an efficient algorithm for producing a schedule for n messages with total length n + O(f2log2n) that can tolerate f medium errors. We also prove that fault-tolerant schedules with length n+O(f logf logn) exist. Since n steps are required to transmit n messages, the overhead of fault tolerance is characterized by the additive terms of O(f2log2n) and O(f logf logn), respectively. Both of these terms are sublinear in n and represent asymptotic improvements to the previously best known schedule, which has overhead fn/2.

In the contention-resolution problem, multiple players contend for access to a shared resource. Contention resolution is used in wireless networks, where messages must be transmitted on a shared communication channel. When two or more messages are transmitted at the same time, a collision occurs, and none of the transmissions succeed. Much of the theoretical work on contention resolution has focused on efficiently resolving collisions in order to obtain throughput guarantees.

In this paper we develop optimal algorithms in the binary-forking model for a variety of fundamental problems, including sorting, semisorting, list ranking, tree contraction, range minima, and ordered set union, intersection and difference. In the binary-forking model, tasks can only fork into two child tasks, but can do so recursively and asynchronously. The tasks share memory, supporting reads, writes and test-and-sets. Costs are measured in terms of work (total number of instructions), and span (longest dependence chain). The binary-forking model is meant to capture both algorithm performance and algorithm-design considerations on many existing multithreaded languages, which are also asynchronous and rely on binary forks either explicitly or under the covers. In contrast to the widely studied PRAM model, it does not assume arbitrary-way forks nor synchronous operations, both of which are hard to implement in modern hardware. While optimal PRAM algorithms are known for the problems studied herein, it turns out that arbitrary-way forking and strict synchronization are powerful, if unrealistic, capabilities. Natural simulations of these PRAM algorithms in the binary-forking model (i.e., implementations in existing parallel languages) incur an $\Omega(\log n)$ overhead in span. This paper explores techniques for designing optimal algorithms when limited to binary forking and assuming asynchrony. All algorithms described in this paper are the first algorithms with optimal work and span in the binary-forking model. Most of the algorithms are simple. Many are randomized.

This brief announcement presents parallel algorithms with a tradeoff between work and span for single source reachability and approximate shortest paths on directed graphs. Both algorithms have ~O(mρ2 + nρ4) work and achieve n1/2+ o(1)/ρ span for all ρ ∈ [1,√n].

The approximate single-source shortest-path problem is as follows: given a graph with nonnegative edge weights and a designated source vertex s, return estimates of the distances from s to each other vertex such that the estimate falls between the true distance and (1+є) times the distance. This paper provides the first nearly work-efficient parallel algorithm with sublinear span (also called depth) for the approximate shortest-path problem on directed graphs. Specifically, for constant є and polynomially-bounded edge weights, our algorithm has work Õ(m) and span n 1/2+o(1). Several algorithms were previously known for the case of undirected graphs, but none of the techniques seem to translate to the directed setting.

One of the simplest problems on directed graphs is that of identifying the set of vertices reachable from a designated source vertex. This problem can be solved easily sequentially by performing a graph search, but efficient parallel algorithms have eluded researchers for decades. For sparse high-diameter graphs in particular, there is no known work-efficient parallel algorithm with nontrivial parallelism. This amounts to one of the most fundamental open questions in parallel graph algorithms: Is there a parallel algorithm for digraph reachability with nearly linear work? This article shows that the answer is yes, presenting a randomized parallel algorithm for digraph reachability and related problems with expected work $\tilde{O}(m)$ and span $\tilde{O}(n^{2/3})$, and hence parallelism $\tilde{\Omega}(m/n^{2/3}) = \tilde{\Omega}(n^{1/3})$, on any graph with $n$ vertices and $m$ arcs. This is the first parallel algorithm having both nearly linear work and strongly sublinear span, i.e., span $\tilde{O}(n^{1-\epsilon})$ for any constant $\epsilon>0$. The algorithm can be extended to produce a directed spanning tree, determine whether the graph is acyclic, topologically sort the strongly connected components of the graph, or produce a directed ear decomposition, all with work $\tilde{O}(m)$ and span $\tilde{O}(n^{2/3})$. The main technical contribution is an efficient Monte Carlo algorithm that, through the addition of $\tilde{O}(n)$ shortcuts, reduces the diameter of the graph to $\tilde{O}(n^{2/3})$ with high probability. While both sequential and parallel algorithms are known with those combinatorial properties, even the sequential algorithms are not efficient, having sequential runtime $\Omega(mn^{\Omega(1)})$. This article presents a surprisingly simple sequential algorithm that achieves the stated diameter reduction and runs in $\tilde{O}(m)$ time. Parallelizing that algorithm yields the main result, but doing so involves overcoming several other challenges.

This paper addresses the problem of provably efficient and practically good on-the-fly determinacy race detection in task parallel programs that use futures. Prior works on determinacy race detection have mostly focused on either task parallel programs that follow a series-parallel dependence structure or ones with unrestricted use of futures that generate arbitrary dependences. In this work, we consider a restricted use of futures and show that we can detect races more efficiently than with general use of futures.

This paper presents I/O-efficient algorithms for topologically sorting a directed acyclic graph and for the more general problem identifying and topologically sorting the strongly connected components of a directed graph G = (V, E). Both algorithms are randomized and have I/O-costs O(sort(E) · poly(log V)), with high probability, where sort(E) = O( logM/B(E/B)) is the I/O cost of sorting an |E|-element array on a machine with size-B blocks and size-M cache/internal memory. These are the first algorithms for these problems that do not incur at least one I/O per vertex, and as such these are the first I/O-efficient algorithms for sparse graphs. By applying the technique of time-forward processing, these algorithms also imply I/O-efficient algorithms for most problems on directed acyclic graphs, such as shortest paths, as well as the single-source reachability problem on arbitrary directed graphs.

Randomized exponential backoff is a widely deployed technique for coordinating access to a shared resource. A good backoff protocol should, arguably, satisfy three natural properties: (1) it should provide constant throughput, wasting as little time as possible; (2) it should require few failed access attempts, minimizing the amount of wasted effort; and (3) it should be robust, continuing to work efficiently even if some of the access attempts fail for spurious reasons. Unfortunately, exponential backoff has some well-known limitations in two of these areas: it can suffer subconstant throughput under bursty traffic, and it is not robust to adversarial disruption. The goal of this article is to “fix” exponential backoff by making it scalable, particularly focusing on the case where processes arrive in an online, worst-case fashion. We present a relatively simple backoff protocol, R e -B ackoff , that has, at its heart, a version of exponential backoff. It guarantees expected constant throughput with dynamic process arrivals and requires only an expected polylogarithmic number of access attempts per process. R e -B ackoff is also robust to periods where the shared resource is unavailable for a period of time. If it is unavailable for D time slots, R e -B ackoff provides the following guarantees. For n packets, the expected number of access attempts for successfully sending a packet is O (log 2 ( n + D )). For the case of an infinite number of packets, we provide a similar result in terms of the maximum number of processes that are ever in the system concurrently.

The future of main memory appears to lie in the direction of new technologies that provide strong capacity-to-performance ratios, but have write operations that are much more expensive than reads in terms of latency, bandwidth, and energy. Motivated by this trend, we propose sequential and parallel algorithms to solve graph connectivity problems using significantly fewer writes than conventional algorithms. Our primary algorithmic tool is the construction of an o(n)-sized implicit decomposition of a bounded-degree graph G on n nodes, which combined with read-only access to G enables fast answers to connectivity and biconnectivity queries on G. The construction breaks the linear-write "barrier", resulting in costs that are asymptotically lower than conventional algorithms while adding only a modest cost to querying time. For general non-sparse graphs on m edges, we also provide the first o(m) writes and O(m) operations parallel algorithms for connectivity and biconnectivity. These algorithms provide insight into how applications can efficiently process computations on large graphs in systems with read-write asymmetry.

A program is said to have a determinacy race if logically parallel parts of a program access the same memory location and one of the accesses is a write. These races are generally bugs in the program since they lead to non-deterministic program behavior — different schedules of the program can lead to different results. Most prior work on detecting these races focuses on a subclass of programs with series-parallel or nested parallelism. This paper presents a race-detection algorithm for detecting races in a more general class of programs, namely programs that include arbitrary ordering constraints in additional to the series-parallel constructs. The algorithm performs a serial execution of the program, augmented to detect races, in O(T1 + k2) time, where T1 is the sequential running time of the original program and k is the number of non series-parallel constraints. The main technical novelty of this paper is a new data structure, R-Sketch, for answering reachability queries in nearly series-parallel (SP) directed acyclic graphs (DAGs). Given as input a graph comprising an n-node series parallel graph and k additional non-SP edges, the total construction time of the data structure is O(n + k2), and each reachability query can be answered in O(1) time. The data structure is traversally incremental, meaning that it supports the insertion of nodes/edges, but only as they are discovered through a graph traversal.

We introduce and study the online house numbering problem, where houses are added arbitrarily along a road and must be assigned labels to maintain their ordering along the road. The online house numbering problem is related to classic online list labeling problems, except that the optimization goal here is to minimize the maximum number of times that any house is relabeled. We provide several algorithms that achieve interesting tradeoffs between upper bounds on the number of maximum relabels per element and the number of bits used by labels.

Load balancing is an important activity in distributed systems. At a high-level of abstraction, this problem assumes processes in the system begin with a potentially uneven assignment of “work” which they must then attempt to spread evenly. There are many different approaches to solving this problem. In this paper, we focus on local load balancing-an approach in which work is balanced in an iterative and distributed manner by having processes exchange work in each round with neighbors in an underlying communication network. The goal with local load balancing is to prove that these local exchanges will lead over time to a global balance. We describe and analyze a new local load balancing strategy called max neighbor, and prove a bound on the number of rounds required for it to obtain a parameterized level of balance, with high probability, in a general dynamic network topology. We then prove this analysis tight (within logarithmic factors) by describing a network and initial work distribution for which max neighbor matches its upper bound, and then build on this to prove that no load balancing algorithm in which every node exchanges work with at most one partner per round can converge asymptotically faster than max neighbor.

Dictionaries remain the most well studied class of data structures. A dictionary supports insertions, deletions, membership queries, and usually successor, predecessor, and extract-min. In a RAM, all such operations take O(log n) time on n elements. Dictionaries are often cross-referenced as follows. Consider a set of tuples {〈ai,bi,ci…〉}. A database might include more than one dictionary on such a set, for example, one indexed on the a ‘s, another on the b‘s, and so on. Once again, in a RAM, inserting into a set of L cross-referenced dictionaries takes O(L log n) time, as does deleting. The situation is more interesting in external memory. On a Disk Access Machine (DAM), B-trees achieve O(logB N) I/Os for insertions and deletions on a single dictionary and K-element range queries take optimal O(logB N + K/B) I/Os. These bounds are also achievable by a B-tree on cross-referenced dictionaries, with a slowdown of an L factor on insertion and deletions. In recent years, both the theory and practice of external- memory dictionaries has been revolutionized by write- optimization techniques. A dictionary is write optimized if it is close to a B-tree for query time while beating B-trees on insertions. The best (and optimal) dictionaries achieve a substantially improved insertion and deletion cost of amortized I/Os on a single dictionary while maintaining optimal O(log1+B∊ N + K/B)- I/O range queries. Although write optimization still helps for insertions into cross-referenced dictionaries, its value for deletions would seem to be greatly reduced. A deletion into a cross- referenced dictionary only specifies a key a. It seems to be necessary to look up the associated values b, c … in order to delete them from the other dictionaries. This takes Ω(logB N) I/Os, well above the per-dictionary write-optimization budget of So the total deletion cost is In short, for deletions, write optimization offers an advantage over B-trees in that L multiplies a lower order term, but when L = 2, write optimization seems to offer no asymptotic advantage over B-trees. That is, no known query- optimal solution for pairs of cross-referenced dictionaries seem to beat B-trees for deletions. In this paper, we show a lower bound establishing that a pair of cross-referenced dictionaries that are optimal for range queries and that supports deletions cannot match the write optimization bound available to insert-only dictionaries. This result thus establishes a limit to the applicability of write-optimization techniques on which many new databases and file systems are based.

This paper gives a new deamortized solution to the sequential-file-maintenance problem. The data structure uses several new tools for solving this historically complicated problem. These tools include an unbalanced ternary-tree layout embedded in the sparse table, one-way rebalancing, and extra structural properties to keep interaction among rebalances to a minimum.

Databases allocate and free blocks of storage on disk. Freed blocks introduce holes where no data is stored. Allocation systems attempt to reuse such deallocated regions in order to minimize the footprint on disk. When previously allocated blocks cannot be moved, this problem is called the memory allocation problem. The competitive ratio for this problem has matching upper and lower bounds that are logarithmic in the number of requests and in the ratio of the largest to smallest requests. This article defines the storage reallocation problem, where previously allocated blocks can be moved, or reallocated , but at some cost. This cost is determined by the allocation/reallocation cost function . The objective is to minimize the storage footprint, that is, the largest memory address containing an allocated object, while simultaneously minimizing the reallocation costs. This article gives asymptotically optimal algorithms for storage reallocation, in which the storage footprint is at most (1+ ϵ) times optimal, and the reallocation cost is O ((1/ϵ)log (1/ϵ)) times the original allocation cost, that is, it is within a constant factor of optimal when ϵ is a constant. The algorithms are cost oblivious , which means they achieve these bounds with no knowledge of the allocation/reallocation cost function, as long as the cost function is subadditive.

In several emerging technologies for computer memory (main memory), the cost of reading is significantly cheaper than the cost of writing. Such asymmetry in memory costs poses a fundamentally different model from the RAM for algorithm design. In this paper we study lower and upper bounds for various problems under such asymmetric read and write costs. We consider both the case in which all but O(1) memory has asymmetric cost, and the case of a small cache of symmetric memory. We model both cases using the (M,omega)-ARAM, in which there is a small (symmetric) memory of size M and a large unbounded (asymmetric) memory, both random access, and where reading from the large memory has unit cost, but writing has cost omega >> 1. For FFT and sorting networks we show a lower bound cost of Omega(omega*n*log_{omega*M}(n)), which indicates that it is not possible to achieve asymptotic improvements with cheaper reads when omega is bounded by a polynomial in M. Moreover, there is an asymptotic gap (of min(omega,log(n)/log(omega*M)) between the cost of sorting networks and comparison sorting in the model. This contrasts with the RAM, and most other models, in which the asymptotic costs are the same. We also show a lower bound for computations on an n*n diamond DAG of Omega(omega*n^2/M) cost, which indicates no asymptotic improvement is achievable with fast reads. However, we show that for the minimum edit distance problem (and related problems), which would seem to be a diamond DAG, we can beat this lower bound with an algorithm with only O(omega*n^2/(M*min(omega^{1/3},M^{1/2}))) cost. To achieve this we make use of a "path sketch" technique that is forbidden in a strict DAG computation. Finally, we show several interesting upper bounds for shortest path problems, minimum spanning trees, and other problems. A common theme in many of the upper bounds is that they require redundant computation and a tradeoff between reads and writes.

In this paper, we study upper and lower bounds for contention resolution on a single hop fading channel; i.e., a channel where receive behavior is determined by a signal to interference and noise ratio (SINR) equation. The best known previous solution solves the problem in this setting in O(log2 n log log n) rounds, with high probability in the system size n. We describe and analyze an algorithm that solves the problem in O(log n + log R) rounds, where R is the ratio between the longest and shortest link, and is a value upper bounded by a polynomial in n for most feasible deployments. We complement this result with an Ω(log n) lower bound that proves the bound tight for reasonable R. We note that in the classical radio network model (which does not include signal fading), high probability contention resolution requires Ω(log 2 n) rounds. Our algorithm, therefore, affirms the conjecture that the spectrum reuse enabled by fading should allow distributed algorithms to achieve a significant improvement on this log2 n speed limit. In addition, we argue that the new techniques required to prove our upper and lower bounds are of general use for analyzing other distributed algorithms in this increasingly well-studied fading channel setting.

In this paper, we consider the classical contention resolution problem in which an unknown subset of n possible nodes are activated and connected to a shared channel. The problem is solved in the first round that an active node transmits alone (thus breaking symmetry). Contention resolution has been an active research topic for over four decades. Accordingly, tight upper and lower bounds are known for most major model assumptions. There remains, however, an important case that is unresolved: contention resolution with multiple channels and collision detection. (Tight bounds are known for contention resolution with multiple channels, and contention resolution with collision detection, but not for the combination of both assumptions.)

Randomized exponential backoff is a widely deployed technique for coordinating access to a shared resource. A good backoff protocol should, arguably, satisfy three natural properties: (i) it should provide constant throughput, wasting as little time as possible; (ii) it should require few failed access attempts, minimizing the amount of wasted effort; and (iii) it should be robust, continuing to work efficiently even if some of the access attempts fail for spurious reasons. Unfortunately, exponential backoff has some well-known limitations in two of these areas: it provides poor (sub-constant) throughput (in the worst case), and is not robust (to adversarial disruption). The goal of this paper is to “fix” exponential backoff by making it scalable, particularly focusing on the case where processes arrive in an on-line, worst-case fashion. We present a relatively simple backoff protocol, Re-Backoff, that has, at its heart, a version of exponential backoff. It guarantees expected constant throughput with dynamic process arrivals and requires only an expected polylogarithmic number of access attempts per process. Re-Backoff is also robust to periods where the shared resource is unavailable for a period of time. If it is unavailable for D time slots, Re-Backoff provides the following guarantees. When the number of packets is a finite n, the average expected number of access attempts for successfully sending a packet is O(log2(n + D)). In the infinite case, the average expected number of access attempts for successfully sending a packet is O(log2(η + D)) where η is the maximum number of processes that are ever in the system concurrently.

Motivated by the significantly higher cost of writing than reading in emerging memory technologies, we consider parallel algorithm design under such asymmetric read-write costs, with the goal of reducing the number of writes while preserving work-efficiency and low span. We present a nested-parallel model of computation that combines (i) small per-task stack-allocated memories with symmetric read-write costs and (ii) an unbounded heap-allocated shared memory with asymmetric read-write costs, and show how the costs in the model map efficiently onto a more concrete machine model under a work-stealing scheduler. We use the new model to design reduced write, work-efficient, low span parallel algorithms for a number of fundamental problems such as reduce, list contraction, tree contraction, breadth-first search, ordered filter, and planar convex hull. For the latter two problems, our algorithms are output-sensitive in that the work and number of writes decrease with the output size. We also present a reduced write, low span minimum spanning tree algorithm that is nearly work-efficient (off by the inverse Ackermann function). Our algorithms reveal several interesting techniques for significantly reducing shared memory writes in parallel algorithms without asymptotically increasing the number of shared memory reads.

Memory efficiency and locality have substantial impact on the performance of programs, particularly when operating on large data sets. Thus, memory- or I/O-efficient algorithms have received significant attention both in theory and practice. The widespread deployment of multicore machines, however, brings new challenges. Specifically, since the memory (RAM) is shared across multiple processes, the effective memory-size allocated to each process fluctuates over time. This paper presents techniques for designing and analyzing algorithms in a cache-adaptive setting, where the RAM available to the algorithm changes over time. These techniques make analyzing algorithms in the cache-adaptive model almost as easy as in the external memory, or DAM model. Our techniques enable us to analyze a wide variety of algorithms --- Master-Method-style algorithms, Akra-Bazzi-style algorithms, collections of mutually recursive algorithms, and algorithms, such as FFT, that break problems of size N into subproblems of size Theta(Nc).

If a parallel program has determinacy race(s), different schedules can result in memory accesses that observe different values --- various race-detection tools have been designed to find such bugs. A key component of race detectors is an algorithm for series-parallel (SP) maintenance, which identifies whether two accesses are logically parallel. This paper describes an asymptotically optimal algorithm, called WSP-Order, for performing SP maintenance in programs with fork-join (or nested) parallelism. Given a fork-join program with T1 work and T∞ span, WSP-Order executes it while also maintaining SP relationships in O(T1/P + T∞) time on P processors, which is asymptotically optimal. At the heart of WSP-Order is a work-stealing scheduler designed specifically for SP maintenance.

We consider the problem of detecting a cycle in a directed graph that grows by arc insertions and the related problems of maintaining a topological order and the strong components of such a graph. For these problems, we give two algorithms, one suited to sparse graphs, the other to dense graphs. The former takes O (min { m 1/2 , n 2/3 } m ) time to insert m arcs into an n -vertex graph; the latter takes O ( n 2 log n ) time. Our sparse algorithm is substantially simpler than a previous O ( m 3/2 )-time algorithm; it is also faster on graphs of sufficient density. The time bound of our dense algorithm beats the previously best time bound of O ( n 5/2 ) for dense graphs. Our algorithms rely for their efficiency on vertex numberings weakly consistent with topological order: we allow ties. Bounds on the size of the numbers give bounds on running time.

We show that simple sequential randomized iterative algorithms for random permutation, list contraction, and tree contraction are highly parallel. In particular, if iterations of the algorithms are run as soon as all of their dependencies have been resolved, the resulting computations have logarithmic depth (parallel time) with high probability. Our proofs make an interesting connection between the dependence structure of two of the problems and random binary trees. Building upon this analysis, we describe linear-work, polylogarithmic-depth algorithms for the three problems. Although asymptotically no better than the many prior parallel algorithms for the given problems, their advantages include very simple and fast implementations, and returning the same result as the sequential algorithm. Experiments on a 40-core machine show reasonably good performance relative to the sequential algorithms.

In a reallocating-scheduler problem, jobs may be inserted and deleted from the system over time. Unlike in traditional online scheduling problems, where a job's placement is immutable, in reallocation problems the schedule may be adjusted, but at some cost. The goal is to maintain an approximately optimal schedule while also minimizing the reallocation cost for changing the schedule.

Emerging memory technologies have a significant gap between the cost, both in time and in energy, of writing to memory versus reading from memory. In this paper we present models and algorithms that account for this difference, with a focus on write-efficient sorting algorithms. First, we consider the PRAM model with asymmetric write cost, and show that sorting can be performed in O(n) writes, O(n log n) reads, and logarithmic depth (parallel time). Next, we consider a variant of the External Memory (EM) model that charges k > 1 for writing a block of size B to the secondary memory, and present variants of three EM sorting algorithms (multi-way merge sort, sample sort, and heap sort using buffer trees) that asymptotically reduce the number of writes over the original algorithms, and perform roughly k block reads for every block write. Finally, we define a variant of the Ideal-Cache model with asymmetric write costs, and present write-efficient,cache-oblivious parallel algorithms for sorting, FFTs, and matrix multiplication. Adapting prior bounds for work-stealing and parallel-depth-first schedulers to the asymmetric setting, these yield provably good bounds for parallel machines with private caches or with a shared cache, respectively.

The recently proposed Integrated Stockpile Evaluation (ISE) problem extends a classic offline scheduling problem where n jobs, each with arrival times and deadlines, must be scheduled nonpreemptively on m machines. The additional constraint in the ISE problem is that a machine may only be used if it has been calibrated recently. The goal is to minimize the number of calibrations necessary to complete all the jobs before their deadlines.

This paper studies the problem of scheduling a streaming pipeline on a multicore machine with private caches to maximize throughput. The theoretical contribution includes lower and upper bounds in the parallel external-memory model. We show that a simple greedy scheduling strategy is asymptotically optimal with a constant-factor memory augmentation. More specifically, we show that if our strategy has a running time of Q cache misses on a machine with size-M caches, then every “static” scheduling policy must have time at least that of Q(Q) cache misses on a machine with size-M/6 caches. Our experimental study considers the question of whether scheduling based on cache effects is more important than scheduling based on only the number of computation steps. Using synthetic pipelines with a range of parameters, we compare our cache-based partitioning against several other static schedulers that load-balance computation. In most cases, the cache-based partitioning indeed beats the other schedulers, but there are some cases that go the other way. We conclude that considering cache effects is a good idea, but other features of the streaming pipeline are also important.

Finding a maximal independent set (MIS) is a classic problem in graph theory that has been widely studied in the context of distributed algorithms. Standard distributed solutions to the MIS problem focus on time complexity. In this paper, we also consider fairness. For a given MIS algorithm A and graph G, we define the inequality factor for A on G to be the largest ratio between the probabilities of the nodes joining an MIS in the graph. We say an algorithm is fair with respect to a family of graphs if it achieves a constant inequality factor for all graphs in the family. In this paper, we seek efficient and fair algorithms for common graph families. We begin by describing an algorithm that is fair and runs in O(log* n)-time in rooted trees of size n. Moving to unrooted trees, we describe a fair algorithm that runs in O(log n) time. Generalizing further to bipartite graphs, we describe a third fair algorithm that requires O(log2 n) rounds. We also show a fair algorithm for planar graphs that runs in O(log2 n) rounds, and describe an algorithm that can be run in any graph, yielding good bounds on inequality in regions that can be efficiently colored with a small number of colors. We conclude our theoretical analysis with a lower bound that identifies a graph where all MIS algorithms achieve an inequality bound in Ω(n)-eliminating the possibility of an MIS algorithm that is fair in all graphs. Finally, to motivate the need for provable fairness guarantees, we simulate both our tree algorithm and Luby's MIS algorithm [13] in a variety of different tree topologies-some synthetic and some derived from real world data. Whereas our algorithm always yield an inequality factor ≤3.25 in these simulations, Luby's algorithms yields factors as large as 168.

Databases allocate and free blocks of storage on disk. Freed blocks introduce holes where no data is stored. Allocation systems attempt to reuse such deallocated regions in order to minimize the footprint on disk. When previously allocated blocks cannot be moved, this problem is called the memory allocation problem. It is known to have a logarithmic overhead in the footprint size. This paper defines the storage reallocation problem, where previously allocated blocks can be moved, or reallocated, but at some cost. This cost is determined by the allocation/reallocation cost function. The algorithms presented here are cost oblivious, in that they work for a broad and reasonable class of cost functions, even when they do not know what the cost function actually is.

This poster proposes an efficient runtime scheduler that provides provable performance guarantees to parallel programs that use data structures through the use of implicit batching.

We introduce the cache-adaptive model, which generalizes the external-memory model to apply to environments in which the amount of memory available to an algorithm can fluctuate. The cache-adaptive model applies to operating systems, databases, and other systems where the allocation of memory to processes changes over time. We prove that if an optimal cache-oblivious algorithm has a particular recursive structure, then it is also an optimal cache-adaptive algorithm. Cache-oblivious algorithms having this form include Floyd-Warshall all pairs shortest paths, naïve recursive matrix multiplication, matrix transpose, and Gaussian elimination. While the cache-oblivious sorting algorithm Lazy Funnel Sort does not have this recursive structure, we prove that it is nonetheless optimally cache-adaptive. We also establish that if a cache-oblivious algorithm is optimal on “square”” (well-behaved) memory profiles then, given resource augmentation it is optimal on all memory profiles. We give paging algorithms for the case where the cache size changes dynamically. We prove that LRU with 4-memory and 4-speed augmentation is competitive with optimal. Moreover, Belady's algorithm remains optimal even when the cache size changes. Cache-obliviousness is distinct from cache-adaptivity. We exhibit a cache-oblivious algorithm that is not cache-adaptive and a cache-adaptive algorithm for a problem having no optimal cache-oblivious solution.

This paper considers the problem of cache-obliviously scheduling streaming pipelines on uniprocessors with the goal of minimizing cache misses. Our recursive algorithm is not parameterized by cache size, yet it achieves the asymptotically minimum number of cache misses with constant factor memory augmentation.

Although concurrent data structures are commonly used in practice on shared-memory machines, even the most efficient concurrent structures often lack performance theorems guaranteeing linear speedup for the enclosing parallel program. Moreover, efficient concurrent data structures are difficult to design. In contrast, parallel batched data structures do provide provable performance guarantees, since processing a batch in parallel is easier than dealing with the arbitrary asynchrony of concurrent accesses. They can limit programmability, however, since restructuring a parallel program to use batched data structure instead of concurrent data structure can often be difficult or even infeasible.

The running time of nested parallel programs on shared memory machines depends in significant part on how well the scheduler mapping the program to the machine is optimized for the organization of caches and processors on the machine. Recent work proposed ``space-bounded schedulers'' for scheduling such programs on the multi-level cache hierarchies of current machines. The main benefit of this class of schedulers is that they provably preserve locality of the program at every level in the hierarchy, resulting (in theory) in fewer cache misses and better use of bandwidth than the popular work-stealing scheduler. On the other hand, compared to work-stealing, space-bounded schedulers are inferior at load balancing and may have greater scheduling overheads, raising the question as to the relative effectiveness of the two schedulers in practice.

Finding a maximal independent set (MIS) is a classic problem in graph theory that has been widely study in the context of distributed algorithms. Standard distributed MIS solutions focus on time complexity. Here we focus on a novel attribute, fairness, where we consider an MIS algorithm fair if all nodes have similar probabilities of joining the set. In many contexts, fairness is important because a node's election to the MIS can have an impact on the resources it consumes. This paper addresses fairness by providing a provably fair and efficient distributed MIS algorithm for unrooted trees. The algorithm runs in O(logn) time and guarantees a correct MIS such that each node enters the set with probability at least 1/4 - ε, for arbitrarily small ε.

No abstract available.

In traditional on-line problems, such as scheduling, requests arrive over time, demanding available resources. As each request arrives, some resources may have to be irrevocably committed to servicing that request. In many situations, however, it may be possible or even necessary to reallocate previously allocated resources in order to satisfy a new request. This reallocation has a cost. This paper shows how to service the requests while minimizing the reallocation cost.

Memory contention can be a serious performance bottleneck in concurrent programs on shared-memory multicore architectures. Having all threads write to a small set of shared locations, for example, can lead to orders of magnitude loss in performance relative to all threads writing to distinct locations, or even relative to a single thread doing all the writes. Shared write access, however, can be very useful in parallel algorithms, concurrent data structures, and protocols for communicating among threads.

The virtues of deterministic parallelism have been argued for decades and many forms of deterministic parallelism have been described and analyzed. Here we are concerned with one of the strongest forms, requiring that for any input there is a unique dependence graph representing a trace of the computation annotated with every operation and value. This has been referred to as internal determinism, and implies a sequential semantics---i.e., considering any sequential traversal of the dependence graph is sufficient for analyzing the correctness of the code. In addition to returning deterministic results, internal determinism has many advantages including ease of reasoning about the code, ease of verifying correctness, ease of debugging, ease of defining invariants, ease of defining good coverage for testing, and ease of formally, informally and experimentally reasoning about performance. On the other hand one needs to consider the possible downsides of determinism, which might include making algorithms (i) more complicated, unnatural or special purpose and/or (ii) slower or less scalable.

This paper considers the problem of scheduling streaming applications on uniprocessors in order to minimize the number of cache-misses. Streaming applications are represented as a directed graph (or multigraph), where nodes are computation modules and edges are channels. When a module fires, it consumes some data-items from its input channels and produces some items on its output channels. In addition, each module may have some state (either code or data) which represents the memory locations that must be loaded into cache in order to execute the module. We consider synchronous dataflow graphs where the input and output rates of modules are known in advance and do not change during execution. We also assume that the state size of modules is known in advance.

The greedy sequential algorithm for maximal independent set (MIS) loops over the vertices in an arbitrary order adding a vertex to the resulting set if and only if no previous neighboring vertex has been added. In this loop, as in many sequential loops, each iterate will only depend on a subset of the previous iterates (i.e. knowing that any one of a vertex's previous neighbors is in the MIS, or knowing that it has no previous neighbors, is sufficient to decide its fate one way or the other). This leads to a dependence structure among the iterates. If this structure is shallow then running the iterates in parallel while respecting the dependencies can lead to an efficient parallel implementation mimicking the sequential algorithm.

This announcement describes the problem based benchmark suite (PBBS). PBBS is a set of benchmarks designed for comparing parallel algorithmic approaches, parallel programming language styles, and machine architectures across a broad set of problems. Each benchmark is defined concretely in terms of a problem specification and a set of input distributions. No requirements are made in terms of algorithmic approach, programming language, or machine architecture. The goal of the benchmarks is not only to compare runtimes, but also to be able to compare code and other aspects of an implementation (e.g., portability, robustness, determinism, and generality). As such the code for an implementation of a benchmark is as important as its runtime, and the public PBBS repository will include both code and performance results.

For nested-parallel computations with low depth (span, critical path length) analyzing the work, depth, and sequential cache complexity suffices to attain reasonably strong bounds on the parallel runtime and cache complexity on machine models with either shared or private caches. These bounds, however, do not extend to general hierarchical caches, due to limitations in (i) the cache-oblivious (CO) model used to analyze cache complexity and (ii) the schedulers used to map computation tasks to processors. This paper presents the parallel cache-oblivious (PCO) model, a relatively simple modification to the CO model that can be used to account for costs on a broad range of cache hierarchies. The first change is to avoid capturing artificial data sharing among parallel threads, and the second is to account for parallelism-memory imbalances within tasks. Despite the more restrictive nature of PCO compared to CO, many algorithms have the same asymptotic cache complexity bounds.

Several existing cache-oblivious dynamic dictionaries achieve O(logB N) (or slightly better memory transfers per operation, where N is the number of items stored, M is the memory size, and B is the block size, which matches the classic B-tree data structure. One recent structure achieves the same query bound and a sometimes-better amortized update bound of memory transfers. This paper presents a new data structure, the xDict, implementing predecessor queries in worst-case memory transfers and insertions and deletions in amortized memory transfers, for any constant ε with 0 < ε < 1. For example, the xDict achieves subconstant amortized update cost when N = M B°(B1−∊), whereas the B-tree's is subconstant only when N = o(MB), and the previously obtained is subconstant only when . The xDict attains the optimal tradeoff between insertions and queries, even in the broader external-memory model, for the range where inserts cost between and O(1/lg3 N) memory transfers.

Let G = (V, E) be a directed acyclic graph (dag) with n = |V| and m = |E|. We say that a total ordering ≺ on vertices V is a topological ordering if for every edge (u, v) ∊ E, we have u ≺ v. In this paper, we consider the problem of maintaining a topological ordering subject to dynamic changes to the underlying graph. That is, we begin with an empty graph G = (V, ) consisting of n nodes. The adversary adds m edges to the graph G, one edge at a time. Throughout this process, we maintain an online topological ordering of the graph G. In this paper, we present a new algorithm that has a total cost of O(n2 log n) for maintaining the topological ordering throughout all the edge additions. At the heart of our algorithm is a new approach for maintaining the ordering. Instead of attempting to place the nodes in an ordered list, we assign each node a label that is consistent with the ordering, and yet can be updated efficiently as edges are inserted. When the graph is dense, our algorithm is more efficient than existing algorithms. By way of contrast, the best known prior algorithms achieve only O (min(m1.5, n2.5)) cost.

This paper introduces a storage format for sparse matrices, called compressed sparse blocks (CSB), which allows both Ax and A,x to be computed efficiently in parallel, where A is an n×n sparse matrix with nnzen nonzeros and x is a dense n-vector. Our algorithms use Θ(nnz) work (serial running time) and Θ(√nlgn) span (critical-path length), yielding a parallelism of Θ(nnz/√nlgn), which is amply high for virtually any large matrix. The storage requirement for CSB is the same as that for the more-standard compressed-sparse-rows (CSR) format, for which computing Ax in parallel is easy but A,x is difficult. Benchmark results indicate that on one processor, the CSB algorithms for Ax and A,x run just as fast as the CSR algorithm for Ax, but the CSB algorithms also scale up linearly with processors until limited by off-chip memory bandwidth.

This paper investigates adding transactions with nested parallelism and nested transactions to a dynamically multithreaded parallel programming language that generates only series-parallel programs. We describe XConflict, a data structure that facilitates conflict detection for a software transactional memory system which supports transactions with nested parallelism and unbounded nesting depth. For languages that use a Cilk-like work-stealing scheduler, XConflict answers concurrent conflict queries in O(1) time and can be maintained efficiently. In particular, for a program with T1 work and a span (or critical-path length) of T∞, the running time on p processors of the program augmented with XConflict is only O(T1/p + pT∞).

This paper presents improved approximation algorithms for the problem of multiprocessor scheduling under uncertainty (SUU), in which the execution of each job may fail probabilistically. This problem is motivated by the increasing use of distributed computing to handle large, computationally intensive tasks. In the SUU problem we are given n unit-length jobs and m machines, a directed acyclic graph G of precedence constraints among jobs, and unrelated failure probabilities qij for each job j when executed on machine i for a single timestep. Our goal is to find a schedule that minimizes the expected makespan.

A streaming B-tree is a dictionary that efficiently implements insertions and range queries. We present two cache-oblivious streaming B-trees, the shuttle tree, and the cache-oblivious lookahead array (COLA).

This paper presents concurrent cache-oblivious (CO) B-trees. We extend the cache-oblivious model to a parallel or distributed setting and present three concurrent CO B-trees. Our first data structure is a concurrent lock-based exponential CO B-tree. This data structure supports insertions and non-blocking searches/successor queries. The second and third data structures are lock-based and lock-free variations, respectively, on the packed-memory CO B-tree. These data structures support range queries and deletions in addition to the other operations. Each data structure achieves the same serial performance as the original data structure on which it is based. In a concurrent setting, we show that these data structures are linearizable, meaning that completed operations appear to an outside viewer as though they occurred in some serialized order. The lock-based data structures are also deadlock free, and the lock-free data structure guarantees forward progress by at least one process.

A key capability of data-race detectors is to determine whether one thread executes logically in parallel with another or whether the threads must operate in series. This paper provides two algorithms, one serial and one parallel, to maintain series-parallel (SP) relationships "on the fly" for fork-join multithreaded programs. The serial SP-order algorithm runs in O(1) amortized time per operation. In contrast, the previously best algorithm requires a time per operation that is proportional to Tarjan's functional inverse of Ackermann's function. SP-order employs an order-maintenance data structure that allows us to implement a more efficient "English-Hebrew" labeling scheme than was used in earlier race detectors, which immediately yields an improved determinacy-race detector. In particular, any fork-join program running in T1 time on a single processor can be checked on the fly for determinacy races in O(T1) time. Corresponding improved bounds can also be obtained for more sophisticated data-race detectors, for example, those that use locks.By combining SP-order with Feng and Leiserson's serial SP-bags algorithm, we obtain a parallel SP-maintenance algorithm, called SP-hybrid. Suppose that a fork-join program has n threads, T1 work, and a critical-path length of T∞. When executed on P processors, we prove that SP-hybrid runs in O((T1/P +PT,/i>∞)lg