c-gpu-example: CPU vs GPU Vector Addition Benchmark

Compares sequential C (GCC) with parallel CUDA on 100M-element float arrays. Targets WSL2 with an NVIDIA RTX 500 Ada GPU.

Prepared with the help of Claude Code.

Results

Version	Time (s)	Throughput (M elem/s)	Speedup
CPU	0.7301	137	1×
GPU	0.0176	5691	~41×

Files

File	Description
vector_add_cpu.c	Sequential for loop, timed with clock()
vector_add_gpu.cu	CUDA kernel, 256 threads/block, timed with cudaEvent API
Makefile	Build script for both versions

Requirements

• CPU: GCC
• GPU: NVIDIA GPU + CUDA Toolkit (nvcc in PATH)
• Compile and run from WSL2, not Windows directly

WSL2 CUDA Setup

1. Install NVIDIA drivers on Windows (regular GeForce/Studio drivers).
2. Install CUDA Toolkit in WSL2:

   wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
   sudo dpkg -i cuda-keyring_1.1-1_all.deb
   sudo apt-get update
   sudo apt-get install -y cuda-toolkit-12-8

3. Add to ~/.bashrc:

   export PATH=/usr/local/cuda/bin:$PATH
   export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

Build & Run

make all          # Build both versions
make cpu          # Build CPU version only
make gpu          # Build GPU version only
make run          # Build and run both
make clean        # Remove compiled binaries

From Windows: wsl make run

Key Parameters

Parameter	Location	Default	Notes
N	Both files	100 000 000	Keep below 1B for 4 GB VRAM
THREADS_PER_BLOCK	GPU only	256	—

end of file