Aditya Manthramurthy <******@****.***> 1716591923 -0700

Anis Eleuch <******@****.***> 1729936707 +0100

Anis Eleuch <******@****.***> 1729936707 +0100

					n += got
				}
				atomic.AddInt64(&ops, int64(n))
				atomic.AddInt64(&lat, latency)
			})
			spent := time.Since(t)
			if spent > 0 && n > 0 {
				// Since we are benchmarking n parallel servers we need to multiply by n.
				// This will give an estimate of the total ops/s.
				latency := float64(atomic.LoadInt64(&lat)) / float64(time.Millisecond)
				b.ReportMetric(float64(n)*float64(ops)/spent.Seconds(), "vops/s")

  return ops::MatMul(ctx, inputs[0], inputs[1], &outputs[0],
                     /*transpose_a=*/false,
                     /*transpose_b=*/false, "MatMul");
}

absl::Status MulModel(AbstractContext* ctx,
                      absl::Span<AbstractTensorHandle* const> inputs,
                      absl::Span<AbstractTensorHandle*> outputs) {
  return ops::Mul(ctx, inputs[0], inputs[1], &outputs[0], "Mul");
}

// on it must match the list of operations in ops.
type testFile struct {
	ops fileOps
	pos int64
}

func (f *testFile) Read(b []byte) (int, error) {
	if len(b) == 0 {
		return 0, nil
	}
	if len(f.ops) == 0 {
		return 0, io.EOF
	}
	s, ok := f.ops[0].(string)
	if !ok {
		return 0, errors.New("unexpected Read operation")
	}

	n := copy(b, s)
	if len(s) > n {
		f.ops[0] = s[n:]
	} else {
		f.ops = f.ops[1:]

    TF_RETURN_IF_ERROR(
        ops::Sub(ctx, fPlus.get(), fMinus.get(), f_outputs, "sub_top"));
    AbstractTensorHandlePtr fDiff(f_outputs[0]);

    // Calculate using the difference quotient definition:
    // (f(theta + eps) - f(theta - eps)) / (2 * eps).
    TF_RETURN_IF_ERROR(
        ops::Div(ctx, fDiff.get(), two_eps.get(), f_outputs, "diff_quotient"));

					}
				case testReadFrom:
					f := &testFile{ops: tf.ops}
					got, err := tw.readFrom(f)
					if _, ok := err.(testError); ok {
						t.Errorf("test %d, ReadFrom(): %v", i, err)
					} else if got != tf.wantCnt || !equalError(err, tf.wantErr) {
						t.Errorf("test %d, ReadFrom() = (%d, %v), want (%d, %v)", i, got, err, tf.wantCnt, tf.wantErr)
					}
					if len(f.ops) > 0 {

#include "tensorflow/c/experimental/gradients/not_differentiable.h"
#include "tensorflow/c/experimental/gradients/tape/tape_context.h"
#include "tensorflow/c/experimental/ops/array_ops.h"
#include "tensorflow/c/experimental/ops/math_ops.h"
#include "tensorflow/c/tf_status_helper.h"
#include "tensorflow/c/tf_tensor.h"
#include "tensorflow/core/lib/llvm_rtti/llvm_rtti.h"
#include "tensorflow/core/platform/errors.h"

        "//tensorflow/core/kernels/data:optional_ops_util",
        "//tensorflow/core/platform:abi",
    ]),
    alwayslink = 1,
)

tf_cuda_library(
    name = "ops",
    srcs = [
        "ops.cc",
    ],
    hdrs = [
        "ops.h",
    ],
    copts = tf_copts(),
    visibility = ["//visibility:public"],
    deps = [
        ":c_api_macros_hdrs",
        ":tf_datatype",