kinap / canny-edge-detector Goto Github PK

1. Create powerpoint
2. perform timing analysis -> compare serial exe time to parallel exe time:
   a) Sweep block/thread count
   b) Sweep size of input image
3. Finish write-up

Comment each function, cleanup any TODOs, etc.

We want our individual kernels to execute in sequential order. We could call cudaDeviceSynchronize() after each step/kernel, or we could put them all in the same CUDA stream and let the scheduling engines do that work for us. This would best exploit task-level parallelism.

Need to time both serial and parallel implementations. Reorganize common code in cannyEdgeDetector::detect_edges().

Hi I'm trying to implement this proyect but I have a issue, the library ed_pixel.h isn't in the proyect, please can you help me?

I know I've got some duplicate #includes and probably some other #includes that I don't need.

Additionally, we can factor out duplicate code or create functions to generate common variables (such as my blur kernel)

Consider switching to cmake in order to support cross-platform compilation (GNU make, Visual Studio solution files). Might be interesting to run this on more powerful hardware at home.