Improving web server performance by network aware data buffering and caching

In this paper we propose a new method of data handling for web servers. We call this method Network Aware Buffering and Caching (NABC for short). NABC facilitates reduction of data copies in web server's data sending path, by doing three things: (1) Layout the data in main memory in a way that protocol processing can be done without data copies (2) Keep a unified cache of data in kernel and ensure safe access to it by various processes and kernel and (3) Pass only the necessary meta data between processes so that bulk data handling time spent during IPC can be reduced. We realize NABC by implementing a set of system calls and an user library. The end product of the implementation is a set of APIs specifically designed for use by the web servers. We port an in house web server called SWEET, to NABC APIs and evaluate performance using a range of workloads both simulated and real. The results show a very impressive gain of 12% to 21% in throughput for static file serving and 1.6 to 4 times gain in throughput for lightweight dynamic content serving for a server using NABC APIs over the one using UNIX APIs.

Modified kinetic flux vector splitting (m-KFVS) method for compressible flows

To resolve many flow features accurately, like accurate capture of suction peak in subsonic flows and crisp shocks in flows with discontinuities, to minimise the loss in stagnation pressure in isentropic flows or even flow separation in viscous flows require an accurate and low dissipative numerical scheme. The first order kinetic flux vector splitting (KFVS) method has been found to be very robust but suffers from the problem of having much more numerical diffusion than required, resulting in inaccurate computation of the above flow features. However, numerical dissipation can be reduced by refining the grid or by using higher order kinetic schemes. In flows with strong shock waves, the higher order schemes require limiters, which reduce the local order of accuracy to first order, resulting in degradation of flow features in many cases. Further, these schemes require more points in the stencil and hence consume more computational time and memory. In this paper, we present a low dissipative modified KFVS (m-KFVS) method which leads to improved splitting of inviscid fluxes. The m-KFVS method captures the above flow features more accurately compared to first order KFVS and the results are comparable to second order accurate KFVS method, by still using the first order stencil. (C) 2011 Elsevier Ltd. All rights reserved.

MODLEX: A Multi Objective Data Layout EXploration Framework for Embedded Systems-on-Chip

The memory subsystem is a major contributor to the performance, power, and area of complex SoCs used in feature rich multimedia products. Hence, memory architecture of the embedded DSP is complex and usually custom designed with multiple banks of single-ported or dual ported on-chip scratch pad memory and multiple banks of off-chip memory. Building software for such large complex memories with many of the software components as individually optimized software IPs is a big challenge. In order to obtain good performance and a reduction in memory stalls, the data buffers of the application need to be placed carefully in different types of memory. In this paper we present a unified framework (MODLEX) that combines different data layout optimizations to address the complex DSP memory architectures. Our method models the data layout problem as multi-objective genetic algorithm (GA) with performance and power being the objectives and presents a set of solution points which is attractive from a platform design viewpoint. While most of the work in the literature assumes that performance and power are non-conflicting objectives, our work demonstrates that there is significant trade-off (up to 70%) that is possible between power and performance.

Free Energy Barriers for Escape of Water Molecules from Protein Hydration Layer

Free energy barriers separating interfacial water molecules from the hydration layer at the surface of a protein to the bulk are obtained by using the umbrella sampling method of free energy calculation. We consider hydration layer of chicken villin head piece (HP-36) which has been studied extensively by molecular dynamics simulations. The free energy calculations reveal a strong sensitivity to the secondary structure. In particular, we find a region near the junction of first and second helix that contains a cluster of water molecules which are slow in motion, characterized by long residence times (of the order of 100 ps or more) and separated by a large free energy barrier from the bulk water. However, these ``slow'' water molecules constitute only about 5-10% of the total number of hydration layer water molecules. Nevertheless, they play an important role in stabilizing the protein conformation. Water molecules near the third helix (which is the important helix for biological function) are enthalpically least stable and exhibit the fastest dynamics. Interestingly, barrier height distributions of interfacial water are quite broad for water surrounding all the three helices (and the three coils), with the smallest barriers found for those near the helix-3. For the quasi-bound water molecules near the first and second helices, we use well-known Kramers' theory to estimate the residence time from the free energy surface, by estimating the friction along the reaction coordinate from the diffusion coefficient by using Einstein relation. The agreement found is satisfactory. We discuss the possible biological function of these slow, quasi-bound (but transient) water molecules on the surface.