Compilation error when compiling Movenet's singlepose thunder tensorflow lite model on IREE 3

[RechieKho]

What happened?

I'm attempting to compile Movenet's singlepose thunder as I'm updating my IREE runtime to IREE 3.0.0. I'm left with these error after importing it with iree-import-tflite.

Internal error while creating host target: Resolution of CPU to CPU-features is not implemented on this target architecture. Pass explicit CPU-features, or implement the missing mapping.

<unknown>:0: error: loc(callsite(callsite("center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1_relu/Relu6;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/bn_Conv1/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise_BN/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise/depthwise;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/block_5_project/Conv2D;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1/Conv2D" at "main") at callsite("center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1_relu/Relu6;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/bn_Conv1/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise_BN/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise/depthwise;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/block_5_project/Conv2D;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1/Conv2D" at "main"))): failed to legalize operation 'stream.executable' that was explicitly marked illegal
<unknown>:0: note: loc(callsite("center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1_relu/Relu6;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/bn_Conv1/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise_BN/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise/depthwise;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/block_5_project/Conv2D;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1/Conv2D" at "main")): called from
<unknown>:0: note: loc("main"): called from
<unknown>:0: note: loc(callsite(callsite("center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1_relu/Relu6;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/bn_Conv1/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise_BN/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise/depthwise;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/block_5_project/Conv2D;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1/Conv2D" at "main") at callsite("center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1_relu/Relu6;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/bn_Conv1/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise_BN/FusedBatchNormV3;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/expanded_conv_depthwise/depthwise;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/block_5_project/Conv2D;center_net_mobile_net_v2fpn_feature_extractor/model_1/model/Conv1/Conv2D" at "main"))): see current operation: 
"stream.executable"() <{sym_name = "jit_eval_dispatch_0", sym_visibility = "private"}> ({
  "stream.executable.export"() <{function_ref = @jit_eval_dispatch_0_transpose_56x27_f32, sym_name = "jit_eval_dispatch_0_transpose_56x27_f32"}> ({
    %6:3 = "flow.dispatch.workgroup_count_from_slice"() : () -> (index, index, index)
    "stream.return"(%6#0, %6#1, %6#2) : (index, index, index) -> ()
  }) : () -> ()
  "builtin.module"() ({
    "func.func"() <{arg_attrs = [{stream.alignment = 64 : index}, {stream.alignment = 64 : index}], function_type = (!stream.binding, !stream.binding) -> (), sym_name = "jit_eval_dispatch_0_transpose_56x27_f32"}> ({
    ^bb0(%arg0: !stream.binding, %arg1: !stream.binding):
      %0 = "arith.constant"() <{value = 0 : index}> : () -> index
      %1 = "stream.binding.subspan"(%arg0, %0) : (!stream.binding, index) -> !flow.dispatch.tensor<readonly:tensor<56x27xf32>>
      %2 = "stream.binding.subspan"(%arg1, %0) : (!stream.binding, index) -> !flow.dispatch.tensor<writeonly:tensor<27x56xf32>>
      %3 = "flow.dispatch.tensor.load"(%1) <{operandSegmentSizes = array<i32: 1, 0, 0, 0, 0>, static_offsets = array<i64: 0, 0>, static_sizes = array<i64: 56, 27>, static_strides = array<i64: 1, 1>}> : (!flow.dispatch.tensor<readonly:tensor<56x27xf32>>) -> tensor<56x27xf32>
      %4 = "tensor.empty"() : () -> tensor<27x56xf32>
      %5 = "linalg.generic"(%3, %4) <{indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>, affine_map<(d0, d1) -> (d1, d0)>], iterator_types = [#linalg.iterator_type<parallel>, #linalg.iterator_type<parallel>], operandSegmentSizes = array<i32: 1, 1>}> ({
      ^bb0(%arg2: f32, %arg3: f32):
        "linalg.yield"(%arg2) : (f32) -> ()
      }) : (tensor<56x27xf32>, tensor<27x56xf32>) -> tensor<27x56xf32>
      "flow.dispatch.tensor.store"(%5, %2) <{operandSegmentSizes = array<i32: 1, 1, 0, 0, 0, 0>, static_offsets = array<i64: 0, 0>, static_sizes = array<i64: 27, 56>, static_strides = array<i64: 1, 1>}> : (tensor<27x56xf32>, !flow.dispatch.tensor<writeonly:tensor<27x56xf32>>) -> ()
      "func.return"() : () -> ()
    }) : () -> ()
  }) : () -> ()
  "stream.executable.end"() : () -> ()
}) : () -> ()

Steps to reproduce your issue

1. Download Movenet's singlepose thunder tensorflow lite model.
2. Import the model with iree-import-tflite singlepose_thunder.tflite -o singlepose_thunder.mlir.
3. Compile the model with iree-compile --iree-input-type=tosa --iree-hal-target-backends=vulkan-spirv singlepose_thunder.mlir -o singlepose_thunder.vulkan.vmfb.
4. Error.

What component(s) does this issue relate to?

Compiler

Version information

IREE (https://iree.dev):
IREE compiler version 3.1.0rc20241225 @ 1f19761
LLVM version 20.0.0git
Optimized build

Additional context

This doesn't happen before 3.0.0. I've tried to compile it with the stable version of compiler from iree-base-compiler and it outputs the same error.

[ScottTodd]

Tagging iree-org/iree-test-suites#5 on this so I/we remember to add test coverage. Should have most of the groundwork done for adding test cases in iree-org/iree-test-suites#59.

[fire]

Is iree 3.1.0 expected to help this?