Enable TPU POD for JAX/1VM by creating devices and local_devices topology.

PiperOrigin-RevId: 289761259
Change-Id: Icbdcca91fd37ea0a04ad16df82aede52e3281ed9

tensorflow/compiler/xla/python/tpu_driver/client/tpu_client.cc

@@ -34,14 +34,34 @@ limitations under the License.
 
 namespace xla {
 
+constexpr char kTpuPlatform[] = "tpu";
+
+TpuDevice::TpuDevice(int id, int host_id, const std::array<int, 3>& coords,
+                     int core_on_chip)
+    : xla::Device(id, /*local_device_state=*/nullptr, kTpuPlatform, host_id),
+      coords_(coords),
+      core_on_chip_(core_on_chip) {}
+
 std::string TpuDevice::DebugString() const {
-  return absl::StrCat("TPU_", id());
+  return absl::StrFormat("TPU_%i(host=%i,(%i,%i,%i,%i))", id(), host_id(),
+                         coords_[0], coords_[1], coords_[2], core_on_chip_);
 }
 
-static std::shared_ptr<Device> MakeDevice(const std::string& platform_name,
-                                          int id) {
-  CHECK_EQ(platform_name, "tpu");
-  return std::make_shared<TpuDevice>(id, /*local_device_state=*/nullptr, "tpu");
+xla::StatusOr<std::vector<std::shared_ptr<xla::Device>>>
+TpuDevice::GetTpuDevices(const tpu_driver::SystemInfo& system_info) {
+  std::vector<std::shared_ptr<Device>> devices;
+  for (const auto& chip : system_info.tpu_chip()) {
+    auto& coord = chip.chip_coord();
+    std::array<int, 3> coords_array = {coord.x(), coord.y(), coord.z()};
+    int host_id = chip.host_id();
+    for (const auto& core : chip.core()) {
+      auto device = std::make_shared<TpuDevice>(
+          core.id(), host_id, coords_array, core.core_on_chip_index());
+      devices.push_back(device);
+    }
+  }
+
+  return devices;
 }
 
 StatusOr<std::shared_ptr<PyTpuClient>> PyTpuClient::Get(
@@ -49,7 +69,6 @@ StatusOr<std::shared_ptr<PyTpuClient>> PyTpuClient::Get(
   tpu_driver::TpuDriverConfig driver_config;
   driver_config.set_worker(worker);
   auto client_status = tpu_driver::TpuDriverRegistry::Open(driver_config);
-
   if (!client_status.ok()) {
     return client_status.status();
   }
@@ -58,19 +77,13 @@ StatusOr<std::shared_ptr<PyTpuClient>> PyTpuClient::Get(
 
   tpu_driver::SystemInfo system_info;
   client->QuerySystemInfo(&system_info);
-  int num_cores =
-      system_info.tpu_chip_size() * system_info.tpu_chip(0).core_size();
 
-  std::vector<std::shared_ptr<Device>> devices;
-  CHECK_GE(num_cores, 1);
-  LOG(INFO) << "Creating " << num_cores << " TPU device(s).";
-  devices.reserve(num_cores);
-  for (int i = 0; i < num_cores; ++i) {
-    devices.push_back(MakeDevice("tpu", i));
-  }
+  TF_ASSIGN_OR_RETURN(std::vector<std::shared_ptr<Device>> devices,
+                      TpuDevice::GetTpuDevices(system_info));
 
-  return std::make_shared<PyTpuClient>("tpu", std::move(client),
-                                       std::move(devices), /*host_id=*/0);
+  return std::make_shared<PyTpuClient>(kTpuPlatform, std::move(client),
+                                       std::move(devices),
+                                       system_info.host_id());
 }
 
 PyTpuClient::PyTpuClient(std::string platform_name,
@@ -81,18 +94,21 @@ PyTpuClient::PyTpuClient(std::string platform_name,
       driver_(std::move(driver)),
       devices_(std::move(devices)),
       host_id_(host_id) {
-  local_devices_.resize(devices_.size());
   for (const std::shared_ptr<Device>& device : devices_) {
     CHECK(id_to_device_.insert({device->id(), device}).second)
         << "Duplicate device id: " << device->id();
 
-    if (device->id() != -1) {
-      int idx = device->id();
-      CHECK(local_devices_[idx] == nullptr) << idx;
-      CHECK_LT(idx, local_devices_.size());
-      local_devices_[idx] = device;
+    if (device->host_id() == host_id_) {
+      LOG(INFO) << "Detected local device, host-id: " << host_id_
+                << ". core-id: " << device->id();
+      local_devices_.push_back(device);
+    } else {
+      VLOG(2) << "Other devices, id: " << device->id();
     }
   }
+  CHECK_GE(local_devices_.size(), 1);
+  LOG(INFO) << "Creating " << local_devices_.size() << " TPU device(s).";
+
   for (int idx = 0; idx < local_devices_.size(); ++idx) {
     CHECK(local_devices_[idx] != nullptr) << idx;
   }
@@ -217,8 +233,8 @@ StatusOr<std::unique_ptr<PyTpuBuffer>> PyTpuBuffer::MakeTuple(
     std::shared_ptr<TpuSharedBuffer> child_device_buffer =
         child_buffer->DeviceBuffer();
     // Merge all definition events from all children, so that anyone using this
-    // tuple must wait for all its children to finish receiving transfers.
-    // This works recursively up a nested tuple tree as well.
+    // tuple must wait for all its children to finish receiving transfers. This
+    // works recursively up a nested tuple tree as well.
     for (std::shared_ptr<tpu_driver::Event> child_event :
          child_device_buffer->wait_for_use) {
       child_events.push_back(std::move(child_event));

tensorflow/compiler/xla/python/tpu_driver/client/tpu_client.h

@@ -38,8 +38,21 @@ namespace xla {
 
 class TpuDevice : public Device {
  public:
-  using Device::Device;
+  TpuDevice(int id, int host_id, const std::array<int, 3>& coords,
+            int core_on_chip);
+
+  const std::array<int, 3>& coords() const { return coords_; }
+  int core_on_chip() const { return core_on_chip_; }
+
   std::string DebugString() const override;
+
+  static xla::StatusOr<std::vector<std::shared_ptr<xla::Device>>> GetTpuDevices(
+      const tpu_driver::SystemInfo& system_info);
+
+ private:
+  const std::array<int, 3> coords_;
+  // Index of the core of the same chip.
+  int core_on_chip_;
 };
 
 // Encapsulates the state of Python session with XLA.
@@ -50,7 +63,7 @@ class PyTpuClient {
   static StatusOr<std::shared_ptr<PyTpuClient>> Get(const std::string& worker);
 
   explicit PyTpuClient(std::string platform_name,
-                       std::unique_ptr<tpu_driver::TpuDriver> client,
+                       std::unique_ptr<tpu_driver::TpuDriver> driver,
                        std::vector<std::shared_ptr<Device>> devices,
                        int host_id);
   virtual ~PyTpuClient() = default;

tensorflow/compiler/xla/python/tpu_driver/client/tpu_client_extension.cc

@@ -206,8 +206,13 @@ PYBIND11_MODULE(tpu_client_extension, m) {
            py::call_guard<py::gil_scoped_release>(), py::arg("arguments"));
 
   py::class_<TpuDevice, Device, std::shared_ptr<TpuDevice>>(m, "TpuDevice")
+      .def_property_readonly("coords", &TpuDevice::coords)
+      .def_property_readonly("core_on_chip", &TpuDevice::core_on_chip)
       .def("__repr__", [](const TpuDevice& device) {
-        return absl::StrFormat("TpuDevice(id=%i)", device.id());
+        return absl::StrFormat(
+            "TpuDevice(id=%i, host_id=%i, coords=(%i,%i,%i), core_on_chip=%i)",
+            device.id(), device.host_id(), device.coords()[0],
+            device.coords()[1], device.coords()[2], device.core_on_chip());
       });
 }  // NOLINT(readability/fn_size)
 

tensorflow/compiler/xla/python/tpu_driver/tpu_driver.proto

@@ -19,15 +19,24 @@ package tpu_driver;
 
 enum MemoryRegion { HBM = 1; }
 
+message ChipCoordinate {
+  required int32 x = 1;
+  required int32 y = 2;
+  required int32 z = 3;
+}
+
 message TpuCoreInfo {
   required int32 id = 1;
-
-  required int64 hbm_bytes_available = 100;
-  required int64 hbm_bytes_allocatable = 101;
+  optional int32 core_on_chip_index = 2;
+  optional int32 core_on_host_index = 3;
+  optional int64 hbm_bytes_available = 100;
+  optional int64 hbm_bytes_allocatable = 101;
 }
 
 message TpuChipInfo {
   repeated TpuCoreInfo core = 1;
+  optional int32 host_id = 2;
+  optional ChipCoordinate chip_coord = 3;
 }
 
 message CpuInfo {
@@ -40,6 +49,11 @@ message CpuInfo {
 message SystemInfo {
   repeated TpuChipInfo tpu_chip = 1;
   required CpuInfo cpu = 2;
+  repeated TpuCoreInfo local_core = 3;
+  optional int32 host_id = 4;
+  optional int32 host_count = 5;
+  optional int32 chip_count = 6;
+  optional int32 core_count = 7;
 }
 
 message TpuDriverConfig {