Ramakrishna Upadrasta (U.Ramakrishna)

Brief Bio

Research Interests

• Domain Specific Programming Languages for Parallelization: Building Domain Specific Languages (DSLs) and compilation strategies for applications from various domains. An example is Big-Data applications and High-Performance Computing. We have funding for this.
• LLVM Optimizations: Various aspects of the open-source LLVM compiler. We are working in collaboration with various industries on this. I also run the IITH-LLVM group. Join the Group (only with IITH email-IDs)
  
• Coarse/Fine-Grain Compiler Optimization and Parallelization techniques for modern architectures for multi-core/GPUs: Proposal of new parallelization algorithms and improvement of the range of programs that are amenable to parallelization schemes in existing polyhedral compilation tools. As an example, I am interested in the range of programs that could be accepted by the well known polyhedral compilation tools like Pluto and Polly.
• Improvement of Scalability of Tools in Polyhedral Compilation The above mentioned polyhedral compilation tools use expensive tools that limit their scalability. I am interested in improving their scalability using advanced approximations like (U)TVPI sub-polyhedra, similar and more powerful to our POPL-2013 paper.
• Abstract Interpretation and sub-polyhedral approximations: Abstract interpretation is a formal framework to perform static analysis of programs. It was formalized by the classic work of Cousot-Cousot in the 1970s. Since then, there have been advances, with the Astrée static analyzer being one of its largest successes. It uses a variety of sub-polyhedral approximations; for example Antoine Miné's Octagon domain. I am interested in various aspects of scalability vs. precision issues of these approximations, with a particular focus in loop-programs (Affine Control Programs/SCoPs).
• Any aspects of program analysis.
• ...

Selected Publications

Current (TBD)

Earlier

• Ramakrishna Upadrasta and Albert Cohen. Sub-polyhedral scheduling using (unit-)two-variable-per-inequality polyhedra. In Proceedings of the 40^th annual ACM SIGPLAN-SIGACT symposium on Principles of programming languages (POPL '13). ACM, New York, NY, USA, 483-496. ACM-DL. Recipient of a European Network of Excellence on High Performance and Embedded Architecture and Compilation (HiPEAC) paper award.
• Ramakrishna Upadrasta and Albert Cohen. Potential and Challenges of Two-Variable-Per-Inequality Sub-Polyhedral Compilation. In 1^st International Workshop on Polyhedral Compilation Techniques (IMPACT-2011), in conjunction with CGO'11, Chamonix, France, April 2011. IMPACT-11 from INRIA
• Dibyendu Das, U. Ramakrishna: A practical and fast iterative algorithm for phi-function computation using DJ graphs. ACM Trans. Program. Lang. Syst. 27(3): 426-440 (2005) ACM-DL
• Maya Madhavan, Priti Shankar, Siddhartha Rai, U. Ramakrishna: Extending Graham-Glanville techniques for optimal code generation. ACM Trans. Program. Lang. Syst. 22(6): 973-1001 (2000) ACM-DL

Teaching

Students

• Keyur Joshi, B.Tech 4th year, (Undergraduate honors thesis).
• Pratik Bhatu, Dual (B.Tech+M.Tech) degree 5th year.
• Abhishek Patwardhan M.Tech-RA 2nd year.
• Santanu Das, M.Tech-RA 2nd year.
• Dangeti Tharun Kumar, M.Tech-TA 2nd year.
• Shalini Jain, PhD 2nd year.
• Gayatri PK, PhD 2nd year.
• Utpal Bora, PhD 2nd year (converted from M.Tech-TA program).

• Ajay Brahmakshatriya, B.Tech (Undergraduate honors thesis).
• Kiran Bhos, M.Tech.
• Yogesh Chavan, M.Tech.
• Bhargav Reddy Godala, B.Tech (Undergraduate honors thesis).

Misc