Prog_Opt_C_Code_on_Blackfin_transcript在Blackfin处理器上编写和优化C代码(2006年1月).pdf

Blackfin Online Learning Development Presentation Title: Programming Optimizing C Code on Blackfin® Presenter Name: Alan Anderson Chapter 1: Concepts and Tools Subchapter 1a: Introduction Subchapter 1b: Useful Tools Chapter 2: Tuning DSP Kernels Subchapter 2a: Tuning techniques Subchapter 2b: Tight Loops Subchapter 2c: Pragmas Subchapter 2d: Volatile etc Chapter 3: Tuning Control code Subchapter 3a: Conditionals Subchapter 3b: Division Subchapter 3c: Advanced Chapter 4: Memory Performance Subchapter 4a: Memory costs Subchapter 4b: Code Speed vs Space Chapter 5: Examples Subchapter 5a: Data Structure Subchapter 5b: Whole Application Chapter 1: Concepts and Tools Subchapter 1a: Introduction Hi. My name is Alan Anderson, I’m a performance engineer with Analog Devices. The talk today is going to be looking at C performance on the Blackfin processor and understanding how you can improve the performance of C programs. We’ll look at the various tools that are available to help you and discuss various scenarios drawn from customer benchmarks that illuminate the situation. There are always a range of techniques available to speed up programs: from engaging automatic optimizations; to intervening in the program; to rewriting parts of the program. The talk will be broken up into a number of sections, beginning with concepts and tools and understanding the general situation on the Blackfin, then we’ll look at tight numeric loops, signal processing kernels vectorization and so on. Then we’ll look at what’s referred to often as control code, decision-making code, conditional branches and the like. The second to last section touches on Blackfin memory performance, because this can be quite significant today in modern processors. And finally, there is an example that will recapitulate some of the material from a rea