Home > other >  Why choose Google CUDA, not OpenCL as TensorFlow backend?
Why choose Google CUDA, not OpenCL as TensorFlow backend?

Time:11-20

Why choose Google CUDA, not OpenCL as TensorFlow backend?

CodePudding user response:

AMD to deep learning support is insufficient; And nv as core business, it attaches great importance to, Google select CUDA is the most important reason is to save money, under the unit energy consumption with nv + CUDA GPU solutions is the most cost-effective,
  • Related