Unable to run resize layer on DLA on Xavier

Robotics & Edge Computing Jetson Systems Jetson Xavier NX
1

Hi I have tried optimizing a model containing resize layer on DLA on Xavier, but it was not running on DLA, falling back to GPU, below is the log

[W] [TRT] Resize__916: For bilinear resize, DLA only supports kHALF_PIXEL coordinate transformation with resize selector: kFORMULA or kUPPER but kASYMMETRIC with resize selector: kFORMULA was found.
[W] [TRT] Default DLA is enabled but layer Resize__916 is not supported on DLA, falling back to GPU.
[I] [TRT] ---------- Layers Running on DLA ----------
[I] [TRT] [DlaLayer] {ForeignNode[resnet_v1_50/block4/unit_3/bottleneck_v1/Relu]}
[I] [TRT] ---------- Layers Running on GPU ----------
[I] [TRT] [GpuLayer] RESIZE: Resize__916

I didn’t understand the log, can you explain to me why the resize layer is not running on DLA and how can i change that resize the layer to make it run on DLA.

Below is the model link I am using

3

Hi,

Please find more details in our document below:

https://docs.nvidia.com/deeplearning/tensorrt/archives/tensorrt-821/developer-guide/index.html#dla_layers

Resize layer

  • Only Nearest Neighbor resize mode is supported.
  • The number of scales has to be exactly 4.
  • The first two elements in scales need to be exactly 1 (for unchanged batch and channel dimensions).
  • The last two elements in scales, representing the scale values along height and width dimensions, respectively, need to be integer values in the range of [1, 32].

Thanks.