%0 = mhlo.constant dense<(1.000000e+00,-1.000000e+00)> : tensor<128x32x4xcomplex<f32>>
  %1 = mhlo.constant dense<(1.000000e+00,1.000000e+00)> : tensor<8x64x128xcomplex<f32>>
  %2 = "mhlo.einsum"(%1, %0) <{einsum_config = "abc,cde->abde"}> : (tensor<8x64x128xcomplex<f32>>, tensor<128x32x4xcomplex<f32>>) -> tensor<8x64x32x4xcomplex<f32>>
  return %2 : tensor<8x64x32x4xcomplex<f32>>
}

// -----

  func.return %0 : tensor<?x2x4x5xf32>
}

// CHECK: func @xla_dot_v2
// CHECK: %[[einsum:.*]] = "tf.Einsum"(%arg0, %arg1) <{equation = "abc,cde->abde"}> : (tensor<?x2x3xf32>, tensor<3x4x5xf32>) -> tensor<?x2x4x5xf32>
// CHECK: return %[[einsum]] : tensor<?x2x4x5xf32>

// -----

// dimension_numbers: {
//   offset_dims: 0
//   collapsed_slice_dims: 1

          func_aliases.values(), [quantize_model_test_base.FUNC_ALIAS]
      )

  @parameterized.parameters(
      testing.parameter_combinations([{
          'equation': (
              'abc,cde->abde',
              'abc,dce->abde',
          ),
      }])
  )
  def test_einsum_ptq_model(
      self,
      equation: str,
  ):
    _, y_shape, bias_shape, x_signature, y_signature = (

}


// -----

module {
  func.func @float_einsum(%arg0: tensor<?x64x32xf32>, %arg1: tensor<32x2x16xf32>) -> (tensor<?x64x2x16xf32>) {
    %0 = "tf.Einsum"(%arg0, %arg1) {equation = "abc,cde->abde"} : (tensor<?x64x32xf32>, tensor<32x2x16xf32>) -> tensor<?x64x2x16xf32>
    func.return %0 : tensor<?x64x2x16xf32>
  }

// CHECK-LABEL: func @float_einsum
// CHECK: -------- Quantization Summary --------

        input_tensor=ops.convert_to_tensor(input_data)
    )

    self.assertAllClose(expected_outputs, got_outputs, atol=0.05)

  @parameterized.parameters(
      ('abc,cde->abde', quant_opts_pb2.XLA),
      ('abc,dce->abde', quant_opts_pb2.XLA),
  )
  def test_einsum_ptq_model(
      self,
      equation: str,
      target_opset: quant_opts_pb2.OpSet,
  ):

	AMVN

	/*
	 * Do not reorder or fragment the conditional branch
	 * opcodes, or the predication code will break
	 */
	ABEQ
	ABNE
	ABCS
	ABHS
	ABCC
	ABLO
	ABMI
	ABPL
	ABVS
	ABVC
	ABHI
	ABLS
	ABGE
	ABLT
	ABGT
	ABLE

	AMOVWD
	AMOVWF
	AMOVDW
	AMOVFW
	AMOVFD
	AMOVDF
	AMOVF
	AMOVD

	ACMPF
	ACMPD
	AADDF
	AADDD

func InvertBranch(as obj.As) obj.As {
	switch as {
	case ABEQ:
		return ABNE
	case ABEQZ:
		return ABNEZ
	case ABGE:
		return ABLT
	case ABGEU:
		return ABLTU
	case ABGEZ:
		return ABLTZ
	case ABGT:
		return ABLE
	case ABGTU:
		return ABLEU
	case ABGTZ:
		return ABLEZ
	case ABLE:
		return ABGT
	case ABLEU:
		return ABGTU
	case ABLEZ:
		return ABGTZ

	asmeq, invasmun bool
}{
	ssa.BlockPPC64EQ: {ppc64.ABEQ, ppc64.ABNE, false, false},
	ssa.BlockPPC64NE: {ppc64.ABNE, ppc64.ABEQ, false, false},

	ssa.BlockPPC64LT: {ppc64.ABLT, ppc64.ABGE, false, false},
	ssa.BlockPPC64GE: {ppc64.ABGE, ppc64.ABLT, false, false},
	ssa.BlockPPC64LE: {ppc64.ABLE, ppc64.ABGT, false, false},
	ssa.BlockPPC64GT: {ppc64.ABGT, ppc64.ABLE, false, false},

	ASRAI
	ALUI
	AAUIPC
	AADD
	ASLT
	ASLTU
	AAND
	AOR
	AXOR
	ASLL
	ASRL
	ASUB
	ASRA

	// 2.5: Control Transfer Instructions
	AJAL
	AJALR
	ABEQ
	ABNE
	ABLT
	ABLTU
	ABGE
	ABGEU

	// 2.6: Load and Store Instructions
	ALW
	ALWU
	ALH
	ALHU
	ALB
	ALBU
	ASW
	ASH
	ASB

	// 2.7: Memory Ordering Instructions
	AFENCE

package cmd

import (
	"context"
	"os"
	"path/filepath"
	"testing"

	"github.com/google/uuid"
)

var testUUID = uuid.MustParse("f5c58c61-7175-4018-ab5e-a94fe9c2de4e")

func BenchmarkCrcHash(b *testing.B) {
	cases := []struct {
		key int
	}{
		{16},
		{64},
		{128},
		{256},
		{512},
		{1024},
	}
	for _, testCase := range cases {