-
Reading from External Memory
Authors:
Ruslan Savchenko
Abstract:
Modern external memory is represented by several device classes. At present, HDD, SATA SSD and NVMe SSD are widely used. Recently ultra-low latency SSD such as Intel Optane became available on the market. Each of these types exhibits it's own pattern for throughput, latency and parallelism. To achieve the highest performance one has to pick an appropriate I/O interface provided by the operating sy…
▽ More
Modern external memory is represented by several device classes. At present, HDD, SATA SSD and NVMe SSD are widely used. Recently ultra-low latency SSD such as Intel Optane became available on the market. Each of these types exhibits it's own pattern for throughput, latency and parallelism. To achieve the highest performance one has to pick an appropriate I/O interface provided by the operating system. In this work we present a detailed overview and evaluation of modern storage reading performance with regard to available Linux synchronous and asynchronous interfaces. While throughout this work we aim for the highest throughput we also measure latency and CPU usage. We provide this report in hope the detailed results could be interesting to both researchers and practitioners.
△ Less
Submitted 22 February, 2021;
originally announced February 2021.
-
On the Complexity of Hub Labeling
Authors:
Maxim Babenko,
Andrew V. Goldberg,
Haim Kaplan,
Ruslan Savchenko,
Mathias Weller
Abstract:
Hub Labeling (HL) is a data structure for distance oracles. Hierarchical HL (HHL) is a special type of HL, that received a lot of attention from a practical point of view. However, theoretical questions such as NP-hardness and approximation guarantee for HHL algorithms have been left aside. In this paper we study HL and HHL from the complexity theory point of view. We prove that both HL and HHL ar…
▽ More
Hub Labeling (HL) is a data structure for distance oracles. Hierarchical HL (HHL) is a special type of HL, that received a lot of attention from a practical point of view. However, theoretical questions such as NP-hardness and approximation guarantee for HHL algorithms have been left aside. In this paper we study HL and HHL from the complexity theory point of view. We prove that both HL and HHL are NP-hard, and present upper and lower bounds for the approximation ratios of greedy HHL algorithms used in practice. We also introduce a new variant of the greedy HHL algorithm and a proof that it produces small labels for graphs with small highway dimension.
△ Less
Submitted 11 January, 2015;
originally announced January 2015.
-
Separating Hierarchical and General Hub Labelings
Authors:
Andrew V. Goldberg,
Ilya Razenshteyn,
Ruslan Savchenko
Abstract:
In the context of distance oracles, a labeling algorithm computes vertex labels during preprocessing. An $s,t$ query computes the corresponding distance from the labels of $s$ and $t$ only, without looking at the input graph. Hub labels is a class of labels that has been extensively studied. Performance of the hub label query depends on the label size. Hierarchical labels are a natural special kin…
▽ More
In the context of distance oracles, a labeling algorithm computes vertex labels during preprocessing. An $s,t$ query computes the corresponding distance from the labels of $s$ and $t$ only, without looking at the input graph. Hub labels is a class of labels that has been extensively studied. Performance of the hub label query depends on the label size. Hierarchical labels are a natural special kind of hub labels. These labels are related to other problems and can be computed more efficiently. This brings up a natural question of the quality of hierarchical labels. We show that there is a gap: optimal hierarchical labels can be polynomially bigger than the general hub labels. To prove this result, we give tight upper and lower bounds on the size of hierarchical and general labels for hypercubes.
△ Less
Submitted 22 June, 2013; v1 submitted 22 April, 2013;
originally announced April 2013.
