unable to run vulkan_tensors_3 and vulkan_tensors_4 (using ML-emulation)

[fabian-lunarg]

I'm trying to run the ML-samples using ARM's ML-emulation layer:
https://github.com/arm/ai-ml-emulation-layer-for-vulkan

using this environment:

VK_INSTANCE_LAYERS=VK_LAYER_ML_Graph_Emulation:VK_LAYER_ML_Tensor_Emulation;LD_LIBRARY_PATH=/path/to/arm/lib64/;VK_ADD_LAYER_PATH=/path/to/arm/share/vulkan/explicit_layer.d

I can successfully run vulkan_tensors_1 and vulkan_tensors_2.
but for 3+4 run into an exception when invoking vkCreateDataGraphPipelinesARM, which seems to originate from the ML-layers:

terminate called after throwing an instance of 'std::runtime_error'
  what():  Missing declared constant for id: 30

so this project is prob. the wrong place to report, but wanted to bring it to your attention anyway.

fyi @per-mathisen-arm

[per-mathisen-arm]

I tried to download the latest binary build from https://github.com/arm/ai-ml-emulation-layer-for-vulkan/releases/tag/v0.9.0 and tested with these tests, and I cannot reproduce this problem.

[fabian-lunarg]

ok, that would already help me out. I didn't test with earlier versions and my report was based on a source-build of most current ml-emulation layers.

[per-mathisen-arm]

If you want more content to test, you can also find more at https://github.com/arm/ai-ml-sdk-scenario-runner - again I would recommend taking the tested binary builds, rather than trying to build these things from source.