CUDA Implementation Details 1. Define error handler macro 2. Define kernel 3. Allocate space on GPU for input image 4. Copy input image from RAM to GPU DRAM 5. Allocate space on GPU for kernel arguments 6. Copy kernel arguments to GPU DRAM 7. Allocate space on GPU for output image 8. Launch the kernel 9. Handle kernel launch error 10. Wait for GPU to finish 11. Copy the result from GPU DRAM to RAM
