This project focused on profiling, instrumentation, and performance analysis to identify bottlenecks in application runtime and memory complexity using perf and Intel's Pin. It was completed for Princeton University's COS375 course: Introduction to Computer Architecture.

Part 1. Profiling with perf
Measure hardware events (e.g., CPU cycles, cache misses).
Gather performance statistics for Sum and Fibonacci programs.

Part 2. Instrumentation with Pin
Develop tools to:
Count dynamic instruction execution per function.
Generate function call graphs.
Track memory access patterns during execution.

Part 3. Performance Analysis
Compare execution times and cache performance.
Analyze runtime complexity and resource utilization.
Explain performance differences using profiling and instrumentation data.

Programs
The project analyzes two sets of programs:
Sum Programs: Compute the sum of the first n natural numbers.
Fibonacci Programs: Compute the nth Fibonacci number.

Tools and Technologies
perf: Linux profiling tool for measuring hardware events.
Intel Pin: A dynamic binary instrumentation framework.

Deliverables
Profiling data and analysis report.
Custom Pin tools:
inst_count.cpp: Counts instructions executed.
call_graph.cpp: Generates function call graphs.
mem_trace.cpp: Tracks memory access patterns.