- Sort Score
- Result 10 results
- Languages All
Results 1 - 10 of 207 for hardware (0.32 sec)
-
tensorflow/compiler/mlir/lite/experimental/tac/hardwares/target_hardware.h
// By default this is the sum of the cost of individual cost for each op. virtual double GetFuncCost(func::FuncOp* func) const; // Returns true if 'op' can run on this Hardware. virtual bool IsOpSupported(mlir::Operation* op) const; // Switching cost between from hardware and this hardware. // If both the hardwares are the same, the transfer cost is basically 0.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue May 09 21:39:59 UTC 2023 - 7.3K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/hardwares/target_hardware.cc
auto* registered_hardwares = GetRegisteredHardwares(); for (auto& hardware : *registered_hardwares) { if (hardware.unique_name == unique_name) { llvm::errs() << "Ignoring duplicate hardware. Hardware " << unique_name << " already registered\n"; hardware.target_hardware_factory = target_hardware_factory; return; } }
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue May 09 21:39:59 UTC 2023 - 9.9K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/common/targets.h
return GetInferenceTypeEnum(device_name_str); } // InferenceDeviceType is a combination of the hardware with inference type. struct InferenceDeviceType { std::string hardware; InferenceType inference_type; bool operator==(const InferenceDeviceType& other) const { return (hardware == other.hardware) && (inference_type == other.inference_type); }
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue Jun 06 03:08:33 UTC 2023 - 4.7K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/transforms/target_annotation.cc
auto* hardware = this->GetTargetHardware(device); if (hardware == nullptr) continue; if (hardware->IsOpSupported(op)) { SetAnnotation(op, kDevice, device, builder); device_is_set = true; break; } } } else { for (const auto* hardware : module_->GetAvailableHardwares()) { if (hardware == nullptr) continue;
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Fri May 19 19:32:06 UTC 2023 - 5.9K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/runtime_metadata.fbs
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed Jul 21 01:22:53 UTC 2021 - 2.5K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/transforms/cost_model.cc
auto hardware = GetTargetAnnotation(op); if (!hardware) return; float cost = GetCostForOp(op, *hardware); UpdateCost(op, cost, &builder); } }); } } // namespace float GetCostForOp(Operation* op, const std::string& hardware) { auto* device_hardware = GetTargetHardware(hardware); if (device_hardware == nullptr) {
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 7.3K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/hardwares/simple_hardware.h
// The larger the value is, the more preferrable over CPU. // If the value > 1, means the hardware has advantage over CPU. // If the value < 1, means CPU is more preferred. // If we specify 10.0, meaning the hardware is 10x faster than CPU. // The value should be > 0. // TODO(renjieliu): Consider add an interface for more detailed customization,
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed Jul 21 01:22:53 UTC 2021 - 2.5K bytes - Viewed (0) -
tensorflow/compiler/mlir/g3doc/overview.md
hardware operations and types. We expect MLIR to be of interest to many groups, including: * Compiler researchers and implementers looking to optimize performance and memory consumption of machine learning models * Hardware makers looking for a way to connect their hardware to TensorFlow, such as TPUs, portable neural hardware in phones, and other custom ASICs
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Fri Feb 21 01:37:38 UTC 2020 - 1.8K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/execution_metadata_exporter.cc
} }); } // Build the flatbuffer. std::vector<flatbuffers::Offset<flatbuffers::String>> hardwares; for (const auto& kv : *hardware_names) { hardwares.push_back(builder->CreateString(kv.first)); } return CreateHardwareMetadata(*builder, builder->CreateVector(hardwares)); } } // namespace std::optional<std::string> ExportRuntimeMetadata(mlir::ModuleOp module) {
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue Jun 11 06:11:34 UTC 2024 - 7.5K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/experimental/tac/transforms/cost_model.h
// Get the estimated cost for the op under the given hardware spec senario. float GetCostForOp(Operation* op, const std::string& hardware); // Get the estimated cost for the whole function under the given hardware. float GetCostForFunc(func::FuncOp* func, const std::string& hardware); // Get the transfer cost given from & to hardware info. // We will only calculate for the "necessary" tensor transferred.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue Apr 19 00:13:50 UTC 2022 - 2.5K bytes - Viewed (0)