About    |    Experience    |    Academics    |    Projects    |    Contact  

Hello! I'm Parth Sharma I am final year post graduate student pursuing M.Tech in Computer Science and Engineering from Indian Institute of Technology, Kanpur. Currently, I am doing my thesis under the supervision of Prof. Swarnendu Biswas. We are working on detecting Concurrency bugs and Scoped Races in GPU programs.

RESEARCH EXPERIENCE

• Efficient Dynamic Data Race Detection for GPUs [ Master Thesis ]
  Supervisor : Prof. Swarnendu Biswas

  Prior software-only work does not detect scoped synchronization and is not predictive in nature. Hardware techniques account for scoped synchronization and tries to be predictive using lockset analysis, but reports both false positives and false negatives.
  Recently, there has been a spurt of work on predictive data race detection in the context of CPU programs. We explore predictive race detection for GPUs to improve the effectiveness and coverage of dynamic analysis.
  We are trying to come up with an efficient dynamic data race detection analysis for GPU kernels which accounts for barrier and scoped synchronization.

ACADEMICS

• Indian Institute of Technology, Kanpur
  
  M.Tech in Computer Science and Engineering              ( 2020-Present )

  Courses

  CS610 Programming for Performance	CS633 Parallel Computing	CS622 Advance Computer Architecture
  CS628 Computer system Security	CS641 Modern Cryptology	CS711 Introduction to Machine Learning
  CS738 Advanced Compiler Optimization





• Institute of Engineering and Technology - DAVV, Indore
  
  B.E. in Information Technology              ( 2016-2020 )

PROJECTS

• Simulation of multi-core cache hierarchy [ Prof. Mainak Chauduri ]
  
  Developed a Simulator to model a multi-core cache hierarchy supporting directory-based cache coherence and reported simulated cycles, L1 cache accesses and misses, L2 cache misses based on given machine access traces.
• Halo exchange with neighbouring processes [ Prof. Preeti Malakar ]
  
  Performed computation and communication between MPI processes for different set of configurations.
  Analyzed the performance of data exchange using different methods such as multiple MPI Sends, MPI Pack/Unpack and MPI derived datatypes using MPICH library.
• CUDA kernel for matrix multiplication [ Prof. Swarnendu Biswas ]
  
  Implemented both optimized and naive [O(n^3)] versions of CUDA kernel for square matrix multiplication.
• Exploration of Cache Inclusive, NINE and Exclusive policies [ Prof. Mainak Chauduri ]
  
  Simulated a two-level cache hierarchy. Examined L2, L3 cache misses for cache Inclusive, NINE and Exclusive policies.
  Performed classification of the L3 misses in the case of an inclusive fully associative L3 cache into Cold, Conflict, and Capacity misses using LRU and Belady’s optimal policies.
• Optimization of MPI Blocking Collectives [ Prof. Preeti Malakar ]
  
  Performed different optimizations on MPI Blocking Collectives, such as MPI Bcast, MPI Reduce, MPI Gather, based on the current system state, knowledge of multicore and network topology of the csews cluster of IITK CSE Dept.
• Escaping the caves - Decrypting the Ciphers [ Prof. Manindra Agrawal ]
  
  Applied cryptanalysis techniques on various Cryptosystems viz. Playfair, Substitution, 6-round DES, EAEAE, and RSA.

CONTACT