A new prefetch cache scheme

The DRAM-based main memory performance is lagging far behind the performance of modern processors. Achieving the highest possible performance of memory-bound applications on modern processors has been increasingly limited by the average DRAM memory access latency. The employed page policy, also referred to as a row-buffer management policy, has a significant impact on DRAM performance. Modern DRAM-based main memory systems have mainly two-page policies; Closed Page Autoprecharge (CPA) policy and Open Page (OP) policy. The optimal page policy is an application-dependent, hence a static page policy utilization may not always be the perfect choice to provide the lowest possible average DRAM memory access latency for the memory requests. In this paper, a dynamic page policy is proposed, in which the decision to switch between the OP policy and the CPA policy is made at run-time for each memory request through monitoring the memory access characteristics of the application instructions. The proposed approach aims at improving the DRAM bandwidth utilization and reducing the average DRAM memory access latency by lowering the number of time-consuming memory operations. The results obtained from this study revealed significant average Last Level Cache (LLC) miss penalty reduction of 25.8% when utilizing effectively the proposed approach compared to the baseline in the evaluated workloads.

In this work we investigate optimal geographical caching in heterogeneous cellular networks where different types of base stations (BSs) have different cache capacities. Users request files from a content library according to a known probability distribution. The performance metric is the total hit probability, which is the probability that a user at an arbitrary location in the plane will find the content that it requires in one of the BSs that it is covered by. We consider the problem of optimally placing content in all BSs jointly. As this problem is not convex, we provide a heuristic scheme by finding the optimal placement policy for one type of base station conditioned on the placement in all other types. We demonstrate that these individual optimization problems are convex and we provide an analytical solution. As an illustration, we find the optimal placement policy of the small base stations (SBSs) depending on the placement policy of the macro base stations (MBSs). We show how the hit probability evolves as the deployment density of the SBSs varies. We show that the heuristic of placing the most popular content in the MBSs is almost optimal after deploying the SBSs with optimal placement policies. Also, for the SBSs no such heuristic can be used; the optimal placement is significantly better than storing the most popular content. Finally, we show that solving the individual problems to find the optimal placement policies for different types of BSs iteratively, namely repeatedly updating the placement policies, does not improve the performance.

Increases in cache capacity are accompanied by growing wire delays due to technology scaling. Non-uniform cache architecture (NUCA) is one of proposed solutions to reducing the average access latency in such cache designs. While most of the prior NUCA work focuses on data placement, data replacement, and migration related issues, this paper studies the problem of data search (access) in NUCA. In our architecture we arrange sets of banks with equal access latency into rings. Our last access based (LAB) prediction scheme predicts the ring that is expected to contain the required data and checks the banks in that ring first for the data block sought. We compare our scheme to two alternate approaches: searching all rings in parallel, and searching rings sequentially. We show that our LAB ring prediction scheme reduces L2 energy significantly over the sequential and parallel schemes, while maintaining similar performance. Our LAB scheme reduces energy consumption by 15.9% relative to the sequential lookup scheme, and 53.8% relative to the parallel lookup scheme.

Increased device density and working set size are driving a rise in cache capacity, which comes at the cost of high access latency. Based on the characteristic of shared data, which is accessed frequently and consumes a little capacity, a novel two-level directory organization is proposed to minimize the cache access time in this paper. In this scheme, a small Fast Directory is used to offer fast hits for a great fraction of memory accesses. Detailed simulation results show that on a 16-core tiled chip multiprocessor, this approach reduces average access latency by 17.9% compared to the general cache organization, and improves the overall performance by 13.3% on average.

In Peer-to-Peer Video-on-Demand (P2P VoD) applications, users tend to seek to the positions that they are interested in. The frequent seeks raise a great challenge to the design of the prefetching scheme. In this paper, we propose a probabilistic prefetching framework to reduce the seeking distance. Each peer performs prefetching based on the segment access probability, which is estimated from the seeking statistics in the previous sessions. It is a challenging task to collect the seeking statistics in a distributed P2P network. In the proposed framework, we employ FM sketches to represent the seeking statistics, thus greatly reducing the space and time complexity. The simulation results show that the proposed prefetching scheme can approach closer to the desired seeking positions compared to the prefetching scheme neglecting the user viewing pattern.

As more processing cores are integrated into one chip and feature size continues to shrink, the average access latency for remote nodes using directory-based coherence protocol becomes higher, which greatly impacts system performance. Previous techniques such as data replication and data migration optimize the performance of the requesting core, but offer little improvement for neighbor nodes. Other techniques such as in-transit optimization try to reduce latency at the cost of increased storage. This paper introduces hierarchical cache directory into CMP (chip multiprocessor), which divides CMP tiles into multiple regions hierarchically, and combines it with data replication. A new directory organization is proposed to record the share status within a region and assist the regional home to complete operation efficiently. Simulation results show that for a 16-core CMP, compared to traditional directory, hierarchical cache directory reduces average access latency by 9% and on-chip network traffic by 34% on average with less storage. Theoretical analyses show that for a 2 n × 2 n tiled CMP, the average access latency in hierarchical cache directory asymptotically approaches a function that is independent of n, hence the architecture is highly scalable.

: Future CMPs will have more cores and greater onchip cache capacity. The on-chip cache can either be divided into separate private L2 caches for each core, or treated as a large shared L2 cache. Private caches provide low hit latency but low capacity, while shared caches have higher hit latencies but greater capacity. Victim replication was previously introduced as a way of reducing the average hit latency of a shared cache by allowing a processor to make a replica of a primary cache victim in its local slice of the global L2 cache. Although victim replication performs well on multithreaded and single-threaded codes, it performs worse than the private scheme for multiprogrammed workloads where there is little sharing between the different programs running at the same time. In this paper, we propose victim migration, which improves on victim replication by adding an additional set of migration tags on each node which are used to implement an exclusive cache policy for replicas. When a replica has been created on a remote node, it is not also cached on the home node, but only recorded in the migration tags. This frees up space on the home node to store shared global lines or replicas for the local processor. We show that victim migration performs better than private, shared, and victim replication schemes across a range of single threaded, multithreaded, and multiprogrammed workloads, while using less area than a private cache design. Victim migration provides a reduction in average memory access latency of up to 10% over victim replication.

This paper introduces a method of prioritizing orbital debris for future active debris removal missions, based on the evaluation of the total collision probability of each cataloged debris. The United States Space Surveillance Network actively monitors, tracks, and catalogs space debris orbiting Earth. The catalog is updated routinely, and is available to the public in the form of two-line element data. The total collision probability of a debris is defined as the overall probability of the debris colliding with any other debris in a given time window. The proposed method uses the two-line element data pertaining to each debris in the catalog to project the future state of the debris, i.e., the classical orbital elements, at predefined time steps for a given time window. The effects of orbital perturbations are considered wherever applicable. The relative distances between all debris are estimated in each time step, and pairwise collision probabilities are calculated for any two debris objects in the catalog. To obtain the total collision probability for a debris, the pairwise collision probabilities pertaining to the debris are summed. Further, for every debris object the trend in the total collision probability as the time window progresses is quantified, and debris objects are ranked based on their chance of collision in the time window. The outcome of the study is compared with target debris proposed in other studies.

Cache-enabled small base station (SBS) densification is foreseen as a key component of 5G cellular networks. This architecture enables storing popular files at the network edge (i.e., SBS caches), which empowers local communication and alleviates traffic congestion at the core/backhaul network. This paper develops a mathematical framework, based on stochastic geometry, to characterize the hit probability in multi-channel cache-enabled 5G networks with both unicast/multicast capabilities and opportunistic spectrum access. To this end, we first derive the hit probability by characterizing the opportunistic spectrum access success probabilities, service distance distributions, and coverage probabilities. An optimization framework for file caching is then developed to maximize the hit probability. To this end, a simple concave approximation for the hit probability is proposed, which highly reduces the optimization complexity and leads to a closed-form solution. The sub-optimal solution is benchmarked against two widely employed caching distribution schemes, namely, uniform and Zipf caching, through numerical results and extensive simulations. It is shown that the caching strategy should be adapted to the network parameters and capabilities. For instance, diversifying file caching according to the Zipf distribution is better in multicast systems with large number of channels. However, when the number of channels is low and/or the network is restricted to unicast transmissions, it is better to confine caching to the most popular files only.

One of the main challenges for computer architects is how to hide the high average memory access latency from the processor. In this context, Hybrid Memory Cubes (HMCs) can provide substantial energy and bandwidth improvements compared to traditional memory organizations. However, it is not clear how this reduced average memory access latency will impact the LLC. For applications with high cache miss ratios, the latency to search for the data inside the cache memory will impact negatively on the performance. The importance of this overhead depends on the memory access latency. In this paper, we present an evaluation of the L3 cache importance on a high performance processor using HMC also exploring chip area tradeoffs between the cache size and number of processor cores. We show that the high bandwidth provided by HMC memories can eliminate the need for L3 caches, removing hardware and making room for more processing power. Our evaluations show that performance increased 37% and the EDP improved 12% while maintaining the same original chip area in a wide range of parallel applications, when compared to DDR3 memories.

Assessing collision algorithms for the newspace era

An analytic method of space debris cloud evolution and its collision evaluation for constellation satellites

Analytical model for collision probability assessments with large satellite constellations

Data access latency is an important metric of system performance in data grid. By means of efficient replication strategy, the amount of data transferred in a wide area network will decrease, and the average access latency of data will decrease ultimately. The motivation of our research is to solve the optimized replica distribution problem in a data grid; that is, the system should utilize many replicas for every data with storage constraints to minimize the average access latency of data. This paper proposes a model of replication strategy in federated data grid and gives the optimized solution. The analysis results and simulation results show that the optimized replication strategy proposed in this paper is superior to LRU caching strategy, uniform replication strategy, proportional replication strategy and square root replication strategy in terms of wide area network bandwidth requirement and in the average access latency of data.

The number of users downloading video content on the Internet has dramatically increased recently, and obtaining huge rich content items, e.g., movie files, is becoming a popular application of the Internet. The transmission bandwidth consumed by delivering rich content is enormous. Therefore, it is urgent for Internet service providers to design an efficient delivery system that minimizes the amount of network resources consumed. Content delivery networks are widely used to reduce the flow hop length and the response time when obtaining content items. To improve the cache efficiency, various methods of replacing caches have been proposed, with the target of improving the cache hit ratio. The effect of reducing the traffic load on network links by delivering content items from caches depends on the distance from the origin server to the users. However, many existing methods do not consider the origin server distance. To maximize the effect of reducing the link load obtained by caches, we propose here a cache-replacement policy based on the hop distance to the origin servers. To simplify the caching algorithm, we divide the storage capacity of cache severs into multiple virtual caches and manage content items separately on the basis of the hop distance to origin servers. We also propose an optimal method for designing the capacities of virtual caches that maximizes the total expected reduction of flow hop length. Through numerical evaluation, we show that the proposed method can decrease the average link load by about 10% compared with the normal least recently used.