• W
    Cudnn convolution reconstruction (#18284) (#18776) · 1b22dd2a
    wangchaochaohu 提交于
    * rewrite the conv_op using cudnn_conv_helper
    
    * add workspace limit for v7 test=develop
    
    * fix test=develop
    
    * add half float test=develop
    
    * fix test=develop
    
    * fix test=develop
    
    * revise code style test=develop
    
    * fix test=develop
    1b22dd2a
conv_cudnn_op.cu.cc 21.0 KB