Worked on the gem5/gem5 repository to address a bug in the GPU protocol tester, focusing on improving the accuracy of address printing during protocol validation. The solution involved replacing the previous calculation, which used cache_line_size multiplied by eight, with a more precise approach using floor(log2(cache_line_size)), as required by the protocol. This adjustment reduced false positives in test results and enhanced the reliability of continuous integration feedback. The work leveraged C++ and expertise in GPU computing and protocol testing, resulting in more dependable GPU protocol validation and streamlined debugging processes, although no new features were introduced during this period.