Effective matrix transpose

Improve matrix transpose step by step.

Please see my blogpost for a detailed explanation.

Performance Comparison

Kernel	Bandwidth (GB/s)	% of Max Bandwidth	Implementation
transpose_naive	875.46	26.5291%	Custom
transpose_swizzle	1251.76	37.9323%	Custom
transpose_swizzle_batched	2771.35	83.9804%	Custom

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
Makefile		Makefile
README.md		README.md
swizzle.cu		swizzle.cu
transpose_naive.cu		transpose_naive.cu
transpose_swizzle.cu		transpose_swizzle.cu
transpose_swizzle_batched.cu		transpose_swizzle_batched.cu
transpose_swizzle_batched_for_profile.cu		transpose_swizzle_batched_for_profile.cu
utils.h		utils.h