Hi, I had a question about the cudnnGetConvolutionForwardAlgorithm_v7() api. I am using CUDA 11.0 and CuDNN 8.0.
cudnnGetConvolutionForwardAlgorithm() is removed in cudnn 8. I am looking into using cudnnGetConvolutionForwardAlgorithm_v7() instead. However, I am running into trouble replicating my current workflow when using the v7 api. cudnnGetConvolutionForwardAlgorithm() allows you to pass in a cudnnConvolutionFwdPreference_t and a memory limit. Previously, I had been able to pass CUDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT and a workspace limit of 0 to select a convolution algorithm that required no workspace. I do not see an obvious way to do this with the v7 api.
Is there a way to select a no-workspace conv algorithm with the v7 api ( cudnnGetForwardAlgorithm_v7() ) ?