Cuda 5 - Others (Library, OpenACC, Dynamic parallelism)

UDACITY教程 Intro to Parallel Programming

Basics on GPU, CUDA, Memory Model

Parallel Algorithms(Reduce, Scan, Histogram, Sort)

Optimize Parallel GPU Programs

Others(Library, OpenACC, Dynamic parallelism)

1. GPU-Accelerated Libraries:

cuBLAS, ArrayFire, OpenCV, etc

Thrust <==> C++ STL

Sort, sort_by_key, scan, reduce, reduce github-wiki

Sort example

2. Other platforms

PyCUDA and MATLAB

3. Cross Platform Solutions

gpu-cross

#pragma acc kernels loop
for( i = 0; i < n; ++i )
	r[i] = a[i]*2.0f;

4. Dynamic parallelism

Nested/Recursive Parallelism

allowed on the compute_35 architecture or above

Quicksort example:

gpu-quicksort