摘要
随着集成电路规模和复杂度的逐渐提高，百万千万门级以上的集成电路验证消耗了整个芯片开发过程大约70％的时间，不仅需要专职的团队，而且人数通常是设计团队的1．5～2倍。针对当前超大规模集成电路验证的这个瓶颈，在传统验证平台的基础上提出了代码覆盖率和功能覆盖率、随机激励与定向激励结合的验证方法。代码覆盖率确保代码的执行，功能覆盖率确保功能点的覆盖，随机与定向激励结合在验证的各个阶段有针对性地编写测试用例，三者相互结合实现高效率验证。此方法在多协议标签交换转发芯片项目中将验证时间缩短了三个月，而且问题的收敛速度加快，验证的规格更可靠。与传统的验证方法相比，此方法提高了验证效率，缩短了验证周期，增强了可靠性，对今后的项目开发有重要借鉴意义和指导意义。
As the scale and complexity of ICs increasing gradually, verification of million or 10 million VLSI costs about 70% time of whole IC research. Not only professional verification groups are needed, but also they must be a half and one to two times than design groups. Based on the bottleneck of present IC research, methodology of code coverage and function coverage driven verification associating with random and direct stimulus is introduced at the fundamental of common testbench. Code coverage to ensure the implementation of the code, function coverage to ensure implementation of function points, and the combination of random and direct stimulus to code special testcases at the various stages of verification, the combination of the three gets highly efficient verification. The methodology shortens three months of verification at a multi - protocol label swi - tching forwarding project,speeds up the convergence of problem, and gets more reliable function points. Comparing with conventional verification methodology, it improves verification productivity, decreases verification time, increases reliability, and has reference means and guidance means for future projects.
出处
《现代电子技术》
2009年第14期115-119,共5页
Modern Electronics Technique
关键词
芯片验证
代码覆盖率
功能覆盖率
随机激励
定向激励
verification of ICs
code coverage
function coverage
random stimulus
direct stimulus
作者简介
罗登富男，1984年出生，四川人，硕士研究生。2006年9月就读于电子科技大学微电子与固体电子学院，2007年7月至今在（华为技术）深圳海思半导体有限公司从事ASIC验证工作。
