if (one_by_one && trivial_strides && trivial_dilations) {
    return "NHWC";
  }

  // If filter spatial dimensions are unknown or not 1x1 we prefer NCHW, because
  // it's the fastest option on NVIDIA GPUs with cuDNN library support.
  return "NCHW";
}

//===----------------------------------------------------------------------===//
// Conv2dBackpropFilterOp

in the gradient computation.}]>:$reserve_space_1,
    Res<TF_Float32Tensor, [{A 1D Tensor for the computed batch variance (inverted variance
in the cuDNN case), to be reused in the gradient computation.}]>:$reserve_space_2
  );

  TF_DerivedOperandTypeAttr T = TF_DerivedOperandTypeAttr<0>;

  let hasCanonicalizer = 1;

  let hasVerifier = 1;
}

build:rbe_linux_cpu_old --host_crosstool_top="@ubuntu20.04-gcc9_manylinux2014-cuda11.2-cudnn8.1-tensorrt7.2_config_cuda//crosstool:toolchain"
build:rbe_linux_cpu_old --crosstool_top="@ubuntu20.04-gcc9_manylinux2014-cuda11.2-cudnn8.1-tensorrt7.2_config_cuda//crosstool:toolchain"
build:rbe_linux_cpu_old --extra_toolchains="@ubuntu20.04-gcc9_manylinux2014-cuda11.2-cudnn8.1-tensorrt7.2_config_cuda//crosstool:toolchain-linux-x86_64"