Troubles to run Resnet neural network through IREE

[FlintWangacc]

Hi guys,
I ran into some troubles in running Resnet vmfb file.
I first convert a pre-trained Resnet model to mlir format with torch-mlir.
Then I convert resnet.mlir to vmfb file through IREE.

iree-compile --iree-hal-target-backends=rocm --iree-hip-target=gfx1100 resnet50.mlir -o resnet50_rocm.vmfb

But it turns out the resnet50_rocm.vmfb file has so many arguments.

iree-dump-module resnet50_rocm.vmfb

//===--------------------------------------------------------------------------------------------------------------===//
// @module : version 0
//===--------------------------------------------------------------------------------------------------------------===//

Required Types:
  [  0] i32
  [  1] i64
  [  2] !hal.allocator
  [  3] !hal.buffer
  [  4] !hal.buffer_view
  [  5] !hal.command_buffer
  [  6] !hal.device
  [  7] !hal.executable
  [  8] !hal.fence
  [  9] !hal.file
  [ 10] !vm.buffer

Module Dependencies:
  hal, version >= 5, required

Imported Functions:
  [  0] hal.ex.file.from_memory(!vm.ref<?>, i64, i32, !vm.ref<?>, i64, i64, i32) -> (!vm.ref<?>)
  [  1] hal.allocator.allocate(!vm.ref<?>, i64, i32, i32, i64) -> (!vm.ref<?>)
  [  2] hal.allocator.import(!vm.ref<?>, i32, i64, i32, i32, !vm.ref<?>, i64, i64) -> (!vm.ref<?>)
  [  3] hal.buffer.assert(!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, i64, i32, i32) -> ()
  [  4] hal.buffer_view.create(!vm.ref<?>, i64, i64, i32, i32, tuple<i64>...) -> (!vm.ref<?>)
  [  5] hal.buffer_view.assert(!vm.ref<?>, !vm.ref<?>, i32, i32, tuple<i64>...) -> ()
  [  6] hal.buffer_view.buffer(!vm.ref<?>) -> (!vm.ref<?>)
  [  7] hal.command_buffer.create(!vm.ref<?>, i32, i32, i64, i32) -> (!vm.ref<?>)
  [  8] hal.command_buffer.finalize(!vm.ref<?>) -> ()
  [  9] hal.command_buffer.execution_barrier(!vm.ref<?>, i32, i32, i32) -> ()
  [ 10] hal.command_buffer.fill_buffer(!vm.ref<?>, !vm.ref<?>, i64, i64, i32, i32, i32) -> ()
  [ 11] hal.command_buffer.dispatch(!vm.ref<?>, !vm.ref<?>, i32, i32, i32, i32, i64, tuple<i32>..., tuple<i32, i32, !vm.ref<?>, i64, i64>...) -> ()
  [ 12] hal.device.allocator(!vm.ref<?>) -> (!vm.ref<?>)
  [ 13] hal.device.query.i64(!vm.ref<?>, !vm.ref<?>, !vm.ref<?>) -> (i32, i64)
  [ 14] hal.device.queue.alloca(!vm.ref<?>, i64, !vm.ref<?>, !vm.ref<?>, i32, i32, i32, i64) -> (!vm.ref<?>)
  [ 15] hal.device.queue.dealloca(!vm.ref<?>, i64, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>) -> ()
  [ 16] hal.device.queue.read(!vm.ref<?>, i64, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, i64, !vm.ref<?>, i64, i64, i32) -> ()
  [ 17] hal.device.queue.execute.indirect(!vm.ref<?>, i64, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, tuple<!vm.ref<?>, i64, i64>...) -> ()
  [ 18] hal.devices.count() -> (i32)
  [ 19] hal.devices.get(i32) -> (!vm.ref<?>)
  [ 20] hal.executable.create(!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>) -> (!vm.ref<?>)
  [ 21] hal.fence.create(!vm.ref<?>, i32) -> (!vm.ref<?>)
  [ 22] hal.fence.join(tuple<!vm.ref<?>>...) -> (!vm.ref<?>)
  [ 23] hal.fence.await(i32, tuple<!vm.ref<?>>...) -> (i32)

Exported Functions:
  [  0] main$async(!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>) -> (!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>)
  [  1] main(!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>) -> (!vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>, !vm.ref<?>)
  [  2] __init() -> ()

//===--------------------------------------------------------------------------------------------------------------===//
// Sections
//===--------------------------------------------------------------------------------------------------------------===//

Module State:
  0 bytes, 5 refs, ~80 bytes total

FlatBuffer: 150972 bytes
  Bytecode: 115208 bytes
  .rodata[  0] embedded       13 bytes `hal.device.id`
  .rodata[  1] embedded        3 bytes `hip`
  .rodata[  2] embedded       21 bytes `hal.executable.format`
  .rodata[  3] embedded       13 bytes `rocm-hsaco-fb`
  .rodata[  4] external   313784 bytes (offset 96 / 60h to 4CA18h)
  .rodata[  5] external 178193088 bytes (offset 313984 / 4CA80h to AA3CD40h)
  .rodata[  6] embedded        6 bytes `tensor`

External .rodata: ~178506872 bytes

//===--------------------------------------------------------------------------------------------------------------===//
// Bytecode : version 0
//===--------------------------------------------------------------------------------------------------------------===//

  # | Offset   |   Length | Blocks | i32 # | ref # | Requirements | Aliases
----+----------+----------+--------+-------+-------+--------------+-----------------------------------------------------
  0 | 00000000 |    72162 |      1 |   996 |     4 |              | 
  1 | 000119E8 |    39340 |      1 |   666 |   321 |              | main$async
  2 | 0001B398 |     1962 |      1 |     2 |   315 |              | main
  3 | 0001BB48 |     1722 |     12 |    26 |     6 |              | __init

//===--------------------------------------------------------------------------------------------------------------===//
// Debug Information
//===--------------------------------------------------------------------------------------------------------------===//
// NOTE: debug databases are large and should be stripped in deployed artifacts.

Locations: 2

As you can see, the main exported by vmfb has so many argument.
In pytorch, I use the following to prepare the input tensor.

preprocess = transforms.Compose([                                                                                                                                                                                                                                                 
transforms.Resize(256),                                                                                                                                                                                                                                                       
transforms.CenterCrop(224),                                                                                                                                                                                                                                                   
transforms.ToTensor(),                                                                                                                                                                                                                                                        
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),                                                                                                                                                                                              
])                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          

# Load an image (replace 'path_to_image.jpg' with your image path)                                                                                                                                                                                                            

image = Image.open('/home/hmsjwzb/work/models/resnet/elephant.jpg')                                                                                                                                                                                                           

input_tensor = preprocess(image)                                                                                                                                                                                                                                              

input_batch = input_tensor.unsqueeze(0)  # Create a mini-batch as expected by the model   

But this vmfb file requires hundreds of parameters.

I did some research on it. It seems in pytorch, aot_export_module will lift parameters and buffers on the module into graph input. I think this caused the hundreds of parameter.

Is there any better way to do it? So I can just pass one use paramter to vmfb file at runtime.

[ScottTodd]

I don't have a complete answer on hand right now but wanted to respond to at least say that your expectations are reasonable. The compiled module should have a much simpler function signature. You could try some variations on aot.export from iree-turbine (docs: https://iree.dev/guides/ml-frameworks/pytorch/#ahead-of-time-aot-export), since that gives you finer control over the functions that get exported and then compiled.

[FlintWangacc]

It seems torch-mlir has a bug. All the placeholder node in computation graph are marked as argument to mlir function.
In the aot_export_module:

(Pdb) p len([t.op for t in gm.graph.nodes if t.op == 'placeholder'])
321
(Pdb)p len(graph_signature.parameters)
161

named_buffers are also marked as mlir function parameter.