Avaliação da influência da quantidade de núcleos de processadores no desempenho da compilação do Kernel Linux

Abstract

Operational system can be defined as the set of one or more programs that provides a package of services, which creates an interface between applications and computer hardware, which allocates and manages shared resources among multiple processes. Created in 1991 in Finland by Linus Torvalds, Linux is considered a clone of Unix, is an inexpensive and functional alternative, especially for the following advantages: multitasking system and multiuser 32 and 64 bits, X-Window graphics system, support several languages, such as Java, C, C ++, Pascal, Lisp, Prolog, among others, support for network protocols: TCP / IP, IPX, AppleTalk, and NetBIOS, and virtual memory. The kernel is the core of the operational system, this code controls the interconnection between user programs and hardware devices, the scheduling of processes to achieve multitasking and other aspects of the system. The kernel compilation requires items such as: the installation of the C compiler (GCC); the C libraries (LibC); enough hard disk space, about 1GB and be "logged" as root. The use of cores control allows speed up the build process, and releasing the amount of cores to suit the hardware capacity. In aim to evaluate the influence of the number of processor cores in Kernel compilation of performance test was carried out in quadruplicate, varying the amount of cores (0, 2, 4, 6, and 8). To perform the tests used a Macbook Pro, Apple's OSX operating system with El Capitan version 10:11, Parallels Desktop software for Mac. The test results were organized in tables and analyzed by the Action 2.9 software. The parameters analyzed were compile time, hardware utilization and temperature variation throughout the process. It can be concluded at the end of testing, that the most active cores, greater use the hardware utilization, lower compile time, but greater increase in temperature.