Can CUDA handle one billion rows of text? Yes.
fast.cu takes 16.8 seconds to process one billion rows on a V100.
This is a 60X speedup over a pure C++ baseline in base.cpp.
Check out the blog for a detailed explanation.
Can CUDA handle one billion rows of text? Yes.
fast.cu takes 16.8 seconds to process one billion rows on a V100.
This is a 60X speedup over a pure C++ baseline in base.cpp.
Check out the blog for a detailed explanation.