Search

Memory has become the primary cost driver in cloud data centers. Yet, a significant portion of memory allocated to VMs in public clouds remains unused. To optimize this resource, "cold" memory can be reclaimed from VMs and stored on slower storage or compressed, enabling memory overcommit. Current overcommit systems rely on general-purpose OS swap mechanisms, which are not optimized for virtualized workloads, leading to missed memory-saving opportunities and ineffective use of optimizations like prefetchers. This paper introduces a userspace memory management framework designed for VMs. It enables custom policies that have full control over the virtual machines' memory using a simple userspace API, supports huge page-based swapping to satisfy VM performance requirements, is easy to deploy by leveraging Linux/KVM, and supports zero-copy I/O virtualization with shared VM memory. Our evaluation demonstrates that an overcommit system based on our framework outperforms the state-of-the-art solutions on both micro-benchmarks and commonly used cloud workloads. Specifically our implementation outperforms the Linux Kernel baseline implementation by up to 25% while saving a similar amount of memory. We also demonstrate the benefits of custom policies by implementing workload-specific reclaimers and prefetchers that save $10\%$ additional memory, improve performance in a limited memory scenario by 30% over the Linux baseline, and recover faster from hard limit releases., Comment: 13 pages, 13 figures

We present a real-time dense geometric mapping algorithm for large-scale environments. Unlike existing methods which use pinhole cameras, our implementation is based on fisheye cameras which have larger field of view and benefit some other tasks including Visual-Inertial Odometry, localization and object detection around vehicles. Our algorithm runs on in-vehicle PCs at 15 Hz approximately, enabling vision-only 3D scene perception for self-driving vehicles. For each synchronized set of images captured by multiple cameras, we first compute a depth map for a reference camera using plane-sweeping stereo. To maintain both accuracy and efficiency, while accounting for the fact that fisheye images have a rather low resolution, we recover the depths using multiple image resolutions. We adopt the fast object detection framework YOLOv3 to remove potentially dynamic objects. At the end of the pipeline, we fuse the fisheye depth images into the truncated signed distance function (TSDF) volume to obtain a 3D map. We evaluate our method on large-scale urban datasets, and results show that our method works well even in complex environments., Comment: 7 pages, 10 figures

Project AutoVision aims to develop localization and 3D scene perception capabilities for a self-driving vehicle. Such capabilities will enable autonomous navigation in urban and rural environments, in day and night, and with cameras as the only exteroceptive sensors. The sensor suite employs many cameras for both 360-degree coverage and accurate multi-view stereo; the use of low-cost cameras keeps the cost of this sensor suite to a minimum. In addition, the project seeks to extend the operating envelope to include GNSS-less conditions which are typical for environments with tall buildings, foliage, and tunnels. Emphasis is placed on leveraging multi-view geometry and deep learning to enable the vehicle to localize and perceive in 3D space. This paper presents an overview of the project, and describes the sensor suite and current progress in the areas of calibration, localization, and perception.

We considered a generic case of pre-transitional materials with static stress-generating defects, dislocations and coherent nano-precipitates, at temperatures close but above the starting temperature of martensitic transformation, Ms. Using the Phase Field Microelasticity theory and 3D simulation, we demonstrated that the local stress generated by these defects produces equilibrium nano-size martensitic embryos (MEs) in pre-transitional state, these embryos being orientation variants of martensite. This is a new type of equilibrium: the thermoelastic equilibrium between the MEs and parent phase in which the total volume of MEs and their size are equilibrium internal thermodynamic parameters. This thermoelastic equilibrium exists only in presence of the stress-generating defects. Cooling the pre-transitional state towards Ms or applying the external stimuli, stress or magnetic field, results in a shift of the thermoelastic equilibrium provided by a reversible anhysteretic growth of MEs that results in a giant ME-generated macroscopic strain. In particular, this effect can be associated with the diffuse phase transformations observed in some ferroelectrics above the Curie point. It is shown that the ME-generated strain is giant and describes a superelasticity if the applied field is stress. It describes a super magnetostriction if the martensite (or austenite) are ferromagnetic and the applied field is a magnetic field. In general, the material with defects can be a multiferroic with a giant multiferroic response if the parent and martensitic phase have different ferroic properties. Finally the ME-generated strain may explain or, at least, contribute to the Invar and Elinvar effects that are typically observed in pre-transitional austenite. The thermoelastic equilibrium and all these effects exist only if the interaction between the defects and MEs is infinite-range., Comment: 34 Pages, 10 Figures

We use the state-of-the-arts density-functional-theory method to study various magnetic orders and their effects on the electronic structures of the FeSe. Our calculated results show that, for the spins of the single Fe layer, the striped antiferromagnetic orders with distortion are more favorable in total energy than the checkerboard antiferromagnetic orders with tetragonal symmetry, which is consistent with known experimental data, and the inter-layer magnetic interaction is very weak. We investigate the electronic structures and magnetic property of the distorted phases. We also present our calculated spin coupling constants and discuss the reduction of the Fe magnetic moment by quantum many-body effects. These results are useful to understand the structural, magnetic, and electronic properties of FeSe, and may have some helpful implications to other FeAs-based materials.

We propose a natural two-speed model for the phase dynamics of Si(111) 7$\times$7 phase transition to high temperature unreconstructed phase. We formulate the phase dynamics by using phase-field method and adaptive mesh refinement. Our simulated results show that a 7$\times$7 island decays with its shape kept unchanged, and its area decay rate is shown to be a constant increasing with its initial area. LEEM experiments concerned are explained, which confirms that the dimer chains and corner holes are broken first in the transition, and then the stacking fault is remedied slowly. This phase-field method is a reliable approach to phase dynamics of surface phase transitions., Comment: 4 pages with figures included