Apply Cuda C++ code to compare the runtime of finishing one CNN layer on CPU and GPU.
Then, apply different optimization methods, such as, uncoalesced tiling, coalesced tiling, etc, to observe the acceleration effects.
Eventuallly, compare the runtime of Cudacore and Tensorcore based on RTX 2080 ti.
A part of results are shown in the Simple results.pdf file.
ShaoKAi100812/CudaCore_TensorCore_Acceleration
Compare the different runtime of CNN computation on CPU and GPU
C++