alexarmbr

I am having a very confusing problem with tf2onnx giving me `TensorScatterAdd` nodes. The tf code I am trying to convert to onnx used `tf.tensor_scatter_nd_add`. I replaced this with `tf.scatter_nd`...

I believe that trying to get TensorRT to use a plugin that implements `NonZero` exactly as described in the ONNX specification ([here](https://github.com/onnx/onnx/blob/master/docs/Operators.md#NonZero)) is impossible with current TensorRT, because the shape...

A workaround that worked for me was to change the source code of the model I was exporting. In my case, `tf.where` was causing the ONNX NonZero operation. This function...

[here](https://github.com/armbuster/matmul-playground/blob/main/src/kernel4.cu) is the code. Right now my kernel is compute bound (no longer memory bound) and I think all these integer instructions are crowding out the issuing of tensor core...

apologies try now!