基本信息
浏览量:315
职业迁徙
个人简介
I am broadly interested in software transformation, mostly for performance optimisation and parallelisation, but I have also looked at various techniques to e.g. reduce code size on embedded processors. Many transformations a compiler can apply are closely coupled to some kind of analysis determining whether the transformation is safe, i.e. it doesn’t break the program, and profitable, i.e. the program runs faster as a result of applying the transformation. Static analyses, which do not execute the program, but reason about it using some fancy maths, are often conservative and return statements such as “don’t know” or “maybe”. Obviously, this is not helpful for program transformation. Instead, we could either improve static analyses, but this is tricky (and often in fact intractable), or we could incorporate dynamic information obtained from actually running the program. And this is what I am interested in! Of course, there are lots of interesting problems to solve, e.g. how can we make this dynamic analysis safe and make sure we don’t miss some corner case, or how do we make it fast as we don’t want to spend a lot of time analysing the program at runtime when all we want to do is to actually reduce the program’s execution time.
All of this has a number of applications, for example to drive better code optimisations applied by a compiler, to enable more aggressive parallelisation of sequential legacy code or to improve performance of just-in-time compilers in dynamic binary translators (also known as instruction set simulators or emulators).
研究兴趣
论文共 94 篇作者统计合作学者相似作者
按年份排序按引用量排序主题筛选期刊级别筛选合作者筛选合作机构筛选
时间
引用量
主题
期刊级别
合作者
合作机构
IEEE/ACM International Symposium on Code Generation and Optimizationpp.55-66, (2024)
International Conference on Compiler Constructionpp.86-99, (2024)
Proceedings of the ACM International Conference on Parallel Architectures and Compilation Techniques (2020)
加载更多
作者统计
合作学者
合作机构
D-Core
- 合作者
- 学生
- 导师
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn