shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general purpose GPUs, and custom-designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer, whereas in previous “parameter server” designs the management of shared state is built into the system, TensorFlow enables developers to...

build:android_arm64 --config=android
build:android_arm64 --cpu=arm64-v8a
build:android_arm64 --fat_apk_cpu=arm64-v8a
build:android_x86 --config=android
build:android_x86 --cpu=x86
build:android_x86 --fat_apk_cpu=x86
build:android_x86_64 --config=android
build:android_x86_64 --cpu=x86_64
build:android_x86_64 --fat_apk_cpu=x86_64

# Build everything statically for Android since all static libs are later

void TFE_SetLogicalCpuDevices(TFE_Context* ctx, int num_cpus,
                              const char* prefix, TF_Status* status) {
  std::vector<std::unique_ptr<tensorflow::Device>> devices;

  if (prefix == nullptr || strlen(prefix) == 0)
    prefix = "/job:localhost/replica:0/task:0";

  tensorflow::SessionOptions sess_options;
  (*sess_options.config.mutable_device_count())["CPU"] = num_cpus;
  status->status =

// through TFE_ContextOptions, this method should not be called.
TF_CAPI_EXPORT extern void TFE_SetLogicalCpuDevices(TFE_Context* ctx,
                                                    int num_cpus,
                                                    const char* prefix,
                                                    TF_Status* status);

// Set configuration key and value using coordination service.

  ASSERT_TRUE(absl::StrContains(device_type_cpu, "CPU")) << device_type_cpu;
  int device_id_cpu = TFE_TensorHandleDeviceID(h_cpu, status.get());
  ASSERT_EQ(TF_OK, TF_GetCode(status.get())) << TF_Message(status.get());
  ASSERT_EQ(0, device_id_cpu) << device_id_cpu;

  TFE_DeleteTensorHandle(h_default);
  TFE_DeleteTensorHandle(h_cpu);
  TFE_Executor* executor = TFE_ContextGetExecutorForThread(ctx);

  TFE_DeleteTensorHandle(retvals[0]);
  TFE_DeleteContext(ctx);
  TF_DeleteStatus(status);
}
TEST(CAPI, Execute_MatMul_CPU_Runtime_Error) {
  Execute_MatMul_CPU_Runtime_Error(false);
}
TEST(CAPI, Execute_MatMul_CPU_Runtime_ErrorAsync) {
  Execute_MatMul_CPU_Runtime_Error(true);
}

void Execute_MatMul_CPU_Type_Error(bool async) {
  TF_Status* status = TF_NewStatus();
  TFE_ContextOptions* opts = TFE_NewContextOptions();

          ./tensorflow/tools/ci_build/ci_build.sh cpu.arm64 bash tensorflow/tools/ci_build/rel/ubuntu/cpu_arm64_test.sh...

TFCI_BAZEL_COMMON_ARGS="--repo_env=TF_PYTHON_VERSION=$TFCI_PYTHON_VERSION --config release_gpu_linux"
TFCI_BAZEL_TARGET_SELECTING_CONFIG_PREFIX=linux_cuda
TFCI_BUILD_PIP_PACKAGE_ARGS="--repo_env=WHEEL_NAME=tensorflow"
TFCI_DOCKER_ARGS="--gpus all"
TFCI_LIB_SUFFIX="-gpu-linux-x86_64"

  TF_DeleteStatus(s);
}

void Deallocator(void* data, size_t, void* arg) {
  tensorflow::cpu_allocator()->DeallocateRaw(data);
  *reinterpret_cast<bool*>(arg) = true;
}

TEST(CAPI, Tensor) {
  const int num_bytes = 6 * sizeof(float);
  float* values =
      reinterpret_cast<float*>(tensorflow::cpu_allocator()->AllocateRaw(
          TF_TensorDefaultAlignment(), num_bytes));
  int64_t dims[] = {2, 3};

# This bazelrc can build a CPU-supporting TF package.

# Convenient cache configurations
# Use a cache directory mounted to /tf/cache. Very useful!
build:sigbuild_local_cache --disk_cache=/tf/cache
# Use the public-access TF DevInfra cache (read only)
build:sigbuild_remote_cache --remote_cache="https://storage.googleapis.com/tensorflow-devinfra-bazel-cache/manylinux2014" --remote_upload_local_results=false