COMPSCI 61C Lecture Notes - Lecture 2: Cpu Cache, Raspberry Pi, Synchronous Dynamic Random-Access Memory
21 Jan 2017
School
Department
Course
Professor
Document Summary
Lab 0 online soon, including instructions for raspberry pi. O ce hours to be announced soon (on webpage) Understand 6 great ideas in computer architecture to enable high performance programming via parallelism. Understand computers from transistor-level up to the programming-language level. Work with small computing boards for connection to the internet of things and. Machine interpretation: hardware architecture description (e. g. block diagrams) Every two years, the transistors you can put in a chip doubles, so the performance doubles, leading to exponential growth. Not true anymore since single-task performance improvements hit a brick wall years ago. Edo, sd-ram, ddr-sdram, rd-ram, and more (physical memory) Intuitive for multiple cores, but single cores can switch between tasks when waiting for data to be received. Not everything can be parallelized, because of dependencies. Failure rate of disks: 2% to 10% per year. On average, a disk fails once per hour. Example of redundancy: nasa fixing rover"s flash memory.
