Courseware for Improving Undergraduate Students' Debugging Skill in GPU Programming

提高本科生GPU编程调试能力课件

• 批准号: 1245876
• 负责人: Ying Zhu
• 金额: $ 17.4万
• 依托单位: Georgia State University Research Foundation, Inc.
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-09-15 至 2018-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1245876&HistoricalAwards=false
• 关键词: Courseware; Improving; Undergraduate; Students; Debugging

This project is developing educational materials to help undergraduate computer science students improve their debugging skill in GPU (Graphics Processing Unit) programming. As multi-core processor architecture becomes the norm, undergraduate computer science students need to shift their focus to multi-core parallel programming. GPU programming is an important part of multi-core parallel programming, especially in heterogeneous computing environments that use a mixture of CPUs, GPUs, and other specialty cores. To write GPU programs, students need to learn special programming languages and new debugging skills. Debugging is particularly challenging for GPU programming because of the complexity of parallel processing in multi-core environments. However, there is a shortage of educational materials for teaching how to debug GPU programs. The project goals are as follows:1. Software defect pattern analysis: Collect and analyze common mistakes made by novice programmers in GPU programming. 2. Teaching materials: Based on the analysis of the software defects, develop teaching materials such as debugging checklists; create tutorials to teach debugging techniques in GPU programming; develop debugging exercises for students to practice their debugging skills.3. Evaluation: Apply the educational materials to parallel and distributed computing courses and computer graphics courses, and measure the efficacy of this approach.4. Dissemination: Share the educational materials online with students and instructors and publicize their existence through other online venues such as the ACM SIGGRAPH Education Committee site and print venues such as the IEEE Computer Graphics & Applications Education Department.

该项目正在开发教育材料，以帮助计算机科学本科生提高他们在GPU（图形处理单元）编程调试技能。随着多核处理器架构成为常态，计算机科学本科生需要将重点转移到多核并行编程。GPU编程是多核并行编程的重要组成部分，特别是在混合使用CPU、GPU和其他专用核的异构计算环境中。要编写GPU程序，学生需要学习特殊的编程语言和新的调试技能。由于多核环境中并行处理的复杂性，GPU编程尤其具有挑战性。然而，缺乏用于教授如何调试GPU程序的教育材料。项目目标如下：1.软件缺陷模式分析：收集和分析新手程序员在GPU编程中的常见错误。2.教材：在分析软件缺陷的基础上，开发调试检查表等教材;制作GPU编程调试技术教程;开发调试练习，让学生练习调试技能.评估：将教材应用于并行与分布式计算课程和计算机图形学课程中，并测量该方法的有效性.传播：与学生和教师在线共享教育材料，并通过其他在线场所（如ACM SIGGRAPH教育委员会网站）和印刷场所（如IEEE计算机图形应用教育部门）宣传其存在&。