"carpentry_saw":                        "\U0001fa9a",
	"carrot":                               "\U0001f955",
	"cartwheeling":                         "\U0001f938",
	"cat":                                  "\U0001f431",
	"cat2":                                 "\U0001f408",
	"cayman_islands":                       "\U0001f1f0\U0001f1fe",
	"cd":                                   "\U0001f4bf",

	VADDECUQ T1, RED2, CAR1, CAR2 // VACCCQ T1, RED2, CAR1, CAR2
	VADDEUQM T1, RED2, CAR1, T1   // VACQ   T1, RED2, CAR1, T1
	VADDUQM  T2, CAR2, T2         // VAQ    T2, CAR2, T2

	// ---------------------------------------------------

	VSUBCUQ  T0, PL, CAR1       // VSCBIQ  PL, T0, CAR1
	VSUBUQM  T0, PL, TT0        // VSQ     PL, T0, TT0
	VSUBECUQ T1, PH, CAR1, CAR2 // VSBCBIQ T1, PH, CAR1, CAR2

	VACQ   T1, RED2, CAR1, T1
	VAQ    T2, CAR2, T2

	// Last round
	VPERM T1, T0, SEL1, RED2    // d1 d0 d1 d0
	VPERM ZER, RED2, SEL2, RED1 // 0  d1 d0  0
	VSQ   RED1, RED2, RED2      // Guaranteed not to underflow

	VSLDB $8, T1, T0, T0
	VSLDB $8, T2, T1, T1

	VACCQ  T0, RED1, CAR1
	VAQ    T0, RED1, T0
	VACCCQ T1, RED2, CAR1, CAR2
	VACQ   T1, RED2, CAR1, T1
	VAQ    T2, CAR2, T2

	// corresponds to the mingw "crt2.o" object. The object itself was
	// built using an x86_64 HOST=linux TARGET=windows clang cross
	// compiler based on LLVM 13. More build details can be found at
	// https://github.com/mstorsjo/llvm-mingw/releases.
	f, err := Open("testdata/llvm-mingw-20211002-msvcrt-x86_64-crt2")
	if err != nil {
		t.Errorf("open failed with %v", err)
	}
	defer f.Close()

  %cst2 = arith.constant dense<[1.0, 2.0]> : tensor<2xf32>
  %0 = "tf.Conv2D"(%arg0, %cst0) {T = "tfdtype$DT_FLOAT", data_format = "NHWC", dilations = [1, 2, 3, 1], padding = "SAME", strides = [1, 4, 5, 1]} : (tensor<1x112x112x3xf32>, tensor<1x3x3x2xf32>) -> tensor<1x28x23x2xf32>
  %1 = "tf.Mul"(%0, %cst2) : (tensor<1x28x23x2xf32>, tensor<2xf32>) -> tensor<1x28x23x2xf32>

  %cst2 = arith.constant dense<[1.0, 2.0]> : tensor<2xf32>
  %0 = "tfl.fully_connected"(%arg0, %cst0, %cst1) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<1x3xf32>, tensor<2x3xf32>, tensor<2xf32>) -> tensor<1x2xf32>
  // %cst2 isn't broadcast-compatible to %cst0, but tf.Mul is able to fold them.

import org.gradle.internal.time.Time
import org.slf4j.Logger
import org.slf4j.LoggerFactory
import spock.lang.Specification

class Slf4jLoggingConfigurerTest extends Specification {
    Logger logger = LoggerFactory.getLogger("cat1")
    OutputEventListener listener = Mock()
    Slf4jLoggingConfigurer configurer = new Slf4jLoggingConfigurer(listener)

    def cleanup() {

                }
                variant('runtime1') {
                    attribute('custom2', 'c1')
                    capability('cap2')
                }
                variant('runtime2') {
                    attribute('custom2', 'c2')
                    capability('cap2')
                }
            }
        }

        buildFile << """
            dependencies {
                conf('org:test:1.0') {

Than McIntosh <******@****.***> 1647953392 -0400

// This pattern optimizes:
//   (x + cst1) + cst2 -> x + cst
//   (x - cst1) - cst2 -> x - cst
// Where: cst = cst1 + cst2
foreach BinaryOp = [TF_AddV2Op, TF_SubOp] in {
  def OptimizeConsecutive#BinaryOp : Pat<
    (BinaryOp
      (BinaryOp $x, (TF_ConstOp:$cst1 $cst1_value)),
      (TF_ConstOp:$cst2 $cst2_value)),
    (BinaryOp
      $x, (TF_AddV2Op $cst1, $cst2))>;
}

// This pattern optimizes: