Skip to content

README.md

Latest commit

 

History

History
53 lines (38 loc) · 2.83 KB

README.md

File metadata and controls

53 lines (38 loc) · 2.83 KB

Accelerated Computing

This repository contains my COMPE 596 Accelerated Computing programming assignments. Each folder holds the source code (C, C++, CUDA, HIP), a Makefile (or build script), and any supporting bash scripts to compile and run the assignment.

Assignments

  • P01 – OpenMP Matrix Multiplication

    • C program that multiplies two (N\times N) matrices in parallel using OpenMP
    • Includes a Makefile for building and a bash script to run benchmarks over 1–128 threads and print timing results

  • P02 – Parallel Doubly-Linked List Insertion

    • C implementation of sorted-list insertion with hand-over-hand locking in OpenMP
    • Benchmarks insertion time vs. thread count and list size

  • P03 – OpenMP Simpson’s-Rule Integration

    • C program to approximate using Simpson’s rule in parallel
    • Build and run scripts to measure runtime and error for different thread/partition counts

  • P04 – Jacobi 2D Solver (CPU & CUDA)

    • Hybrid C/CUDA code implementing the Jacobi iterative method on a 2D grid
    • Compares CPU serial, CPU parallel (OpenMP), GPU non-SIMD, and GPU SIMD kernels
    • Makefile and bash scripts to build and run each variant

  • P05 – CUDA Reduction for Array Summation

    • CUDA kernel for parallel reduction to sum large arrays
    • Serial C version included for baseline comparison
    • Build scripts to automate array-size sweeps and print speedup metrics

  • P06 – Naïve vs Tiled GPU Matrix Multiplication

    • Two CUDA kernels: naïve and tiled (16×16 shared-memory) for multiplying (M\times1024) by (1024\times M)
    • Makefile builds both versions, and a runner script measures relative performance

  • P07 – Sobel 5×5 Convolution in CUDA

    • CUDA kernel applying a 5×5 horizontal Sobel filter to an image buffer
    • Includes a minimal C driver and build scripts to compile and execute on test data

  • P08 – cuSOLVER LU Factorization & Solve

    • C++ program using cuSOLVER to factor and solve Hilbert systems of size (2^1)–(2^{10})
    • Scripts to build, run, perturb right-hand sides, and print solver timings

  • P09 – GPU-Accelerated Audio Filtering (cuFFT vs FFTW)

    • C++/CUDA code that reads a WAV file, zeroes out a 10 kHz tone in the frequency domain via FFT
    • Comparison between FFTW (CPU) and cuFFT (GPU) implementations
    • Makefile and run script to build and filter test audio

  • P11 – Dense vs Sparse GEMM with ROCm

    • HIP program timing dense GEMM (rocBLAS) vs. sparse GEMM (rocSPARSE) on large matrices
    • Build scripts to automate runs at various sparsity levels

Note: P10 (OpenCL) was omitted. Each assignment folder contains all source files, build instructions, and scripts needed to reproduce the results. Feel free to clone and explore!