return output


def set_tf_cuda_compute_capabilities(environ_cp):
  """Set TF_CUDA_COMPUTE_CAPABILITIES."""
  while True:
    native_cuda_compute_capabilities = get_native_cuda_compute_capabilities(
        environ_cp)
    if not native_cuda_compute_capabilities:
      default_cuda_compute_capabilities = _DEFAULT_CUDA_COMPUTE_CAPABILITIES
    else:
      default_cuda_compute_capabilities = native_cuda_compute_capabilities

  TF_Status* s = TF_NewStatus();
  TF_OpKernelConstruction_GetAttrType(ctx, "foobar", &type, s);
  TF_DeleteStatus(s);
  return NULL;
}

// A compute function. This will never actually get called in this test, it's
// just nice to know that it compiles.
void compute(void* kernel, TF_OpKernelContext* ctx) {
  TF_Tensor* input;
  TF_Status* s = TF_NewStatus();
  TF_GetInput(ctx, 0, &input, s);
  TF_DeleteTensor(input);

  // If provided, a non-null `forward_function` will be used instead of the
  // backward function (`backward_function_getter` /
  // `backward_function_deleter`) to compute jvps for this operation. If
  // `forward_function` is null, a GradientTape is used on the backward function
  // to compute the jvp, which will waste computation when executing eagerly.
  //
  // Unlike GradientTape::RecordOperation, Accumulate runs gradient computation

// newer version is available.
func prepareUpdateMessage(downloadURL string, older time.Duration) string {
	if downloadURL == "" || older <= 0 {
		return ""
	}

	// Compute friendly duration string to indicate time
	// difference between newer and current release.
	t := time.Time{}
	newerThan := humanize.RelTime(t, t.Add(older), "before the latest release", "")

	if globalServerCtxt.JSON {

 * not necessarily the case for declaration providers (e.g. the IDE declaration provider hitting the index without caching).
 *
 * However, since symbol names providers may not be able to compute name sets per their contract, we may have to fall back to the
 * [declaration provider][org.jetbrains.kotlin.analysis.providers.KotlinDeclarationProvider].
 *

build:cuda_clang --@local_config_cuda//:cuda_compiler=clang
# Select supported compute capabilities (supported graphics cards).
# This is the same as the official TensorFlow builds.
# See https://developer.nvidia.com/cuda-gpus#compute
# `compute_XY` enables PTX embedding in addition to SASS. PTX
# is forward compatible beyond the current compute capability major
# release while SASS is only forward compatible inside the current

        this.id = id;
        this.transitive = transitive;
    }

    /**
     * The {@code id} uniquely represents a value for this extensible enum.
     * This id should be used to compute the equality and hash code for the instance.
     *
     * @return the id
     */
    @Nonnull
    public String id() {
        return id;
    }

    public boolean isTransitive() {

// decrypted object stream, we need to read the range (encOff,
// encLength) of the encrypted object stream to decrypt it, and
// compute skipLen, the number of bytes to skip in the beginning of
// the encrypted range.
//
// In addition we also compute the object part number for where the
// requested range starts, along with the DARE sequence number within
// that part. For single part objects, the partStart will be 0.

        List<LifecycleMojo> mojos = goals.getMojos();
        if (mojos != null) {
            for (int i = 0; i < mojos.size(); i++) {
                LifecycleMojo mojo = mojos.get(i);

                // Compute goal coordinates
                String groupId, artifactId, version, goal;
                String[] p = mojo.getGoal().trim().split(":");
                if (p.length == 3) {

                .getOrDefault(artifact.getBaseVersion(), null);
    }

    // groupId -> (artifactId -> (version -> project)))
    private Map<String, Map<String, Map<String, MavenProject>>> getAllProjects() {
        // compute the projects mapping
        if (allProjects == null) {
            List<MavenProject> allProjects = session.getAllProjects();
            if (allProjects != null) {