| /external/pytorch/test/benchmark_utils/ |
| D | callgrind_artifacts.json | 160 …t_user/repos/pytorch/build/aten/src/ATen/Functions.cpp:at::ones(c10::ArrayRef<long>, c10::TensorOp… 161 …"5130822 build/aten/src/ATen/Functions.cpp:at::ones(c10::ArrayRef<long>, c10::TensorOptions const&… 162 …ctions.cpp:at::ones(c10::ArrayRef<long>, c10::optional<c10::ScalarType>, c10::optional<c10::Layout… 163 …spatcher.h:at::ones(c10::ArrayRef<long>, c10::optional<c10::ScalarType>, c10::optional<c10::Layout… 164 …c10::Dispatcher::callWithDispatchKey<at::Tensor, c10::ArrayRef<long>, c10::optional<c10::ScalarTyp… 165 …c10::Dispatcher::callWithDispatchKey<at::Tensor, c10::ArrayRef<long>, c10::optional<c10::ScalarTyp… 166 …c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoRuntimeFunctor_<at::Ten… 167 …c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoRuntimeFunctor_<at::Ten… 168 …us namespace)::ones(c10::ArrayRef<long>, c10::optional<c10::ScalarType>, c10::optional<c10::Layout… 169 …us namespace)::ones(c10::ArrayRef<long>, c10::optional<c10::ScalarType>, c10::optional<c10::Layout… [all …]
|
| /external/pytorch/aten/src/ATen/core/ |
| D | ATen_pch.h | 16 // "impl" headers (e.g BFloat16-inl.h or complex_math.h in c10). 24 // Where the threshold can be tweaked until c10 and some of ATen 66 #include <c10/core/Allocator.h> 67 #include <c10/core/AutogradState.h> 68 #include <c10/core/Backend.h> 69 #include <c10/core/DefaultDtype.h> 70 #include <c10/core/Device.h> 71 #include <c10/core/DeviceType.h> 72 #include <c10/core/DispatchKey.h> 73 #include <c10/core/DispatchKeySet.h> [all …]
|
| D | type_factory.h | 8 #include <c10/macros/Macros.h> 10 namespace c10 { 16 struct TORCH_API TypeFactoryBase<c10::DynamicType> { 18 static c10::DynamicTypePtr create(TypePtr ty, Args&&... args) { 19 return std::make_shared<c10::DynamicType>( 20 c10::DynamicTypeTrait<T>::tagValue(), 21 c10::DynamicType::Arguments(c10::ArrayRef<c10::TypePtr>( 25 static c10::DynamicTypePtr create(const std::vector<c10::TypePtr>& types) { 26 return std::make_shared<c10::DynamicType>( 27 c10::DynamicTypeTrait<T>::tagValue(), [all …]
|
| /external/pytorch/c10/core/ |
| D | DispatchKey.cpp | 1 #include <c10/core/DispatchKey.h> 2 #include <c10/core/DispatchKeySet.h> 6 namespace c10 { namespace 263 c10::DispatchKey parseDispatchKey(const std::string& k) { in parseDispatchKey() 264 static std::unordered_map<std::string, c10::DispatchKey> key_map = { in parseDispatchKey() 265 {"Undefined", c10::DispatchKey::Undefined}, in parseDispatchKey() 266 {"Dense", c10::DispatchKey::Dense}, in parseDispatchKey() 267 {"FPGA", c10::DispatchKey::FPGA}, in parseDispatchKey() 268 {"MAIA", c10::DispatchKey::MAIA}, in parseDispatchKey() 269 {"MPS", c10::DispatchKey::MPS}, in parseDispatchKey() [all …]
|
| /external/pytorch/c10/util/ |
| D | TypeCast.h | 2 #include <c10/macros/Macros.h> 3 #include <c10/util/BFloat16.h> 4 #include <c10/util/Float8_e4m3fn.h> 5 #include <c10/util/Float8_e4m3fnuz.h> 6 #include <c10/util/Float8_e5m2.h> 7 #include <c10/util/Float8_e5m2fnuz.h> 8 #include <c10/util/Half.h> 9 #include <c10/util/complex.h> 21 namespace c10 { 103 struct static_cast_with_inter_type<c10::complex<c10::Half>, c10::BFloat16> { [all …]
|
| D | complex_math.h | 3 …"c10/util/complex_math.h is not meant to be individually included. Include c10/util/complex.h inst… 11 C10_HOST_DEVICE inline c10::complex<T> exp(const c10::complex<T>& x) { in exp() 13 return static_cast<c10::complex<T>>( in exp() 16 return static_cast<c10::complex<T>>( in exp() 22 C10_HOST_DEVICE inline c10::complex<T> log(const c10::complex<T>& x) { in log() 24 return static_cast<c10::complex<T>>( in log() 27 return static_cast<c10::complex<T>>( in log() 33 C10_HOST_DEVICE inline c10::complex<T> log10(const c10::complex<T>& x) { in log10() 35 return static_cast<c10::complex<T>>( in log10() 38 return static_cast<c10::complex<T>>( in log10() [all …]
|
| /external/pytorch/aten/src/ATen/native/cpu/ |
| D | zmath.h | 4 #include <c10/util/complex.h> 5 #include <c10/util/MathConstants.h> 17 inline c10::complex<float> zabs <c10::complex<float>> (c10::complex<float> z) { 18 return c10::complex<float>(std::abs(z)); 22 inline float zabs <c10::complex<float>, float> (c10::complex<float> z) { 27 inline c10::complex<double> zabs <c10::complex<double>> (c10::complex<double> z) { 28 return c10::complex<double>(std::abs(z)); 32 inline double zabs <c10::complex<double>, double> (c10::complex<double> z) { 47 return z < 0 ? c10::pi<double> : 0; in angle_impl() 51 inline c10::complex<float> angle_impl <c10::complex<float>> (c10::complex<float> z) { [all …]
|
| /external/pytorch/test/cpp_extensions/ |
| D | mtia_extension.cpp | 2 #include <c10/core/Device.h> 3 #include <c10/core/Stream.h> 4 #include <c10/core/impl/DeviceGuardImplInterface.h> 5 #include <c10/util/Logging.h> 10 constexpr c10::DeviceType kMTIADeviceType = c10::DeviceType::MTIA; 11 constexpr c10::DeviceIndex kMTIADeviceCount = 2; 12 static thread_local c10::DeviceIndex current_device = 0; 13 static thread_local std::array<c10::Stream, kMTIADeviceCount> current_streams = 14 {c10::Stream::unpack3(0, 0, c10::DeviceType::MTIA), 15 c10::Stream::unpack3(0, 1, c10::DeviceType::MTIA)}; [all …]
|
| /external/pytorch/c10/test/util/ |
| D | complex_test_common.h | 1 #include <c10/macros/Macros.h> 2 #include <c10/util/complex.h> 3 #include <c10/util/hash.h> 21 static_assert(sizeof(c10::complex<float>) == 2 * sizeof(float), ""); in test_size() 22 static_assert(sizeof(c10::complex<double>) == 2 * sizeof(double), ""); in test_size() 26 static_assert(alignof(c10::complex<float>) == 2 * sizeof(float), ""); in test_align() 27 static_assert(alignof(c10::complex<double>) == 2 * sizeof(double), ""); in test_align() 31 static_assert(std::is_standard_layout<c10::complex<float>>::value, ""); in test_pod() 32 static_assert(std::is_standard_layout<c10::complex<double>>::value, ""); in test_pod() 38 c10::complex<float> zz = *reinterpret_cast<c10::complex<float>*>(&z); in TEST() [all …]
|
| D | complex_math_test_common.h | 4 #include <c10/util/complex.h> 20 c10::complex<float> e_i_pi = std::exp(c10::complex<float>(0, float(PI))); in C10_DEFINE_TEST() 25 c10::complex<float> e_i_pi = ::exp(c10::complex<float>(0, float(PI))); in C10_DEFINE_TEST() 30 c10::complex<double> e_i_pi = std::exp(c10::complex<double>(0, PI)); in C10_DEFINE_TEST() 35 c10::complex<double> e_i_pi = ::exp(c10::complex<double>(0, PI)); in C10_DEFINE_TEST() 44 c10::complex<float> x(0.1, 1.2); in C10_DEFINE_TEST() 45 c10::complex<float> e = std::exp(x); in C10_DEFINE_TEST() 52 c10::complex<float> x(0.1, 1.2); in C10_DEFINE_TEST() 53 c10::complex<float> e = ::exp(x); in C10_DEFINE_TEST() 60 c10::complex<double> x(0.1, 1.2); in C10_DEFINE_TEST() [all …]
|
| /external/pytorch/aten/src/ATen/cpu/vec/vec256/ |
| D | vec256_half_neon.h | 9 #include <c10/util/Half.h> 10 #include <c10/util/irange.h> 62 class Vectorized<c10::Half> { 67 // value_type should be c10::Half to fit interface with vec_base.h 68 using value_type = c10::Half; 77 Vectorized<c10::Half> map2( 78 const Vectorized<c10::Half>& second, 79 c10::Half (*const f)(c10::Half, c10::Half)) const { 80 __at_align__ c10::Half tmp_first[size()]; 81 __at_align__ c10::Half tmp_second[size()]; [all …]
|
| D | vec256_complex_double.h | 6 #include <c10/util/complex.h> 7 #include <c10/util/irange.h> 22 template <> class Vectorized<c10::complex<double>> { 26 using value_type = c10::complex<double>; 33 Vectorized(c10::complex<double> val) { in Vectorized() 39 Vectorized(c10::complex<double> val1, c10::complex<double> val2) { in Vectorized() 47 …static Vectorized<c10::complex<double>> blend(const Vectorized<c10::complex<double>>& a, const Vec… in blend() 48 // convert c10::complex<V> index mask to V index mask: xy -> xxyy in blend() 61 …static Vectorized<c10::complex<double>> blendv(const Vectorized<c10::complex<double>>& a, const Ve… in blendv() 62 const Vectorized<c10::complex<double>>& mask) { in blendv() [all …]
|
| D | vec256_qint.h | 10 #include <c10/util/irange.h> 11 #include <c10/util/qint32.h> 12 #include <c10/util/qint8.h> 13 #include <c10/util/quint8.h> 259 struct Vectorized<c10::qint32> : public Vectorizedqi { 274 using int_vec_return_type = std::array<Vectorized<c10::qint32>, 1>; 275 using value_type = c10::qint32::underlying; 284 Vectorized(const c10::qint32& val) { 297 static Vectorized<c10::qint32> loadu(const void* ptr) { 298 return Vectorized<c10::qint32>(ptr); [all …]
|
| D | vec256_complex_float.h | 6 #include <c10/util/complex.h> 7 #include <c10/util/irange.h> 21 template <> class Vectorized<c10::complex<float>> { 25 using value_type = c10::complex<float>; 32 Vectorized(c10::complex<float> val) { in Vectorized() 41 …Vectorized(c10::complex<float> val1, c10::complex<float> val2, c10::complex<float> val3, c10::comp… in Vectorized() 52 …static Vectorized<c10::complex<float>> blend(const Vectorized<c10::complex<float>>& a, const Vecto… in blend() 53 // convert c10::complex<V> index mask to V index mask: xy -> xxyy in blend() 90 …static Vectorized<c10::complex<float>> blendv(const Vectorized<c10::complex<float>>& a, const Vect… in blendv() 91 const Vectorized<c10::complex<float>>& mask) { in blendv() [all …]
|
| /external/pytorch/test/cpp/aoti_abi_check/ |
| D | test_dtype.cpp | 3 #include <c10/util/BFloat16-math.h> 4 #include <c10/util/BFloat16.h> 5 #include <c10/util/Float8_e4m3fn.h> 6 #include <c10/util/Float8_e4m3fnuz.h> 7 #include <c10/util/Float8_e5m2.h> 8 #include <c10/util/Float8_e5m2fnuz.h> 9 #include <c10/util/Half.h> 10 #include <c10/util/complex.h> 16 c10::BFloat16 a = 1.0f; in TEST() 17 c10::BFloat16 b = 2.0f; in TEST() [all …]
|
| /external/pytorch/aten/src/ATen/ |
| D | ExpandUtils.h | 12 #include <c10/util/Exception.h> 13 #include <c10/util/MaybeOwned.h> 14 #include <c10/util/irange.h> 88 // Functions in ExpandUtils return `c10::MaybeOwned<Tensor>` because 91 // `c10::MaybeOwned<Tensor>::borrowed(to_expand)`. However, this means 92 // that you need to be careful: the returned `c10::MaybeOwned<Tensor>` 99 inline c10::MaybeOwned<Tensor> expand_inplace( in expand_inplace() 103 return c10::MaybeOwned<Tensor>::borrowed(to_expand); in expand_inplace() 105 return c10::MaybeOwned<Tensor>::owned( in expand_inplace() 109 inline c10::MaybeOwned<Tensor> expand_inplace( [all …]
|
| /external/pytorch/torch/csrc/utils/ |
| D | python_symnode.h | 3 #include <c10/core/SafePyObject.h> 4 #include <c10/core/SymNodeImpl.h> 30 // This c10::SymNodeImpl simply backends to a Python object that 33 class PythonSymNodeImpl : public c10::SymNodeImpl { 35 PythonSymNodeImpl(py::object pyobj) : c10::SymNodeImpl() { in PythonSymNodeImpl() 36 pyobj_ = std::make_shared<c10::SafePyObject>( in PythonSymNodeImpl() 40 c10::SymNode wrap_int(int64_t num) override { in wrap_int() 43 return c10::make_intrusive<PythonSymNodeImpl>(std::move(r)); in wrap_int() 46 c10::SymNode wrap_float(double num) override { in wrap_float() 49 return c10::make_intrusive<PythonSymNodeImpl>(std::move(r)); in wrap_float() [all …]
|
| /external/pytorch/tools/iwyu/ |
| D | c10.imp | 2 { include: [ "<c10/macros/cmake_macros.h>", private, "<c10/macros/Macros.h>", public ] }, 3 { include: [ "<c10/macros/Export.h>", public, "<c10/macros/Macros.h>", public ] }, 5 { include: [ "<c10/util/BFloat16-inl.h>", private, "<c10/util/BFloat16.h>", public ] }, 6 { include: [ "<c10/util/Half-inl.h>", private, "<c10/util/Half.h>", public ] }, 7 { include: [ "<c10/util/Float8_e5m2-inl.h>", private, "<c10/util/Float8_e5m2.h>", public ] }, 8 … { include: [ "<c10/util/Float8_e4m3fn-inl.h>", private, "<c10/util/Float8_e4m3fn.h>", public ] }, 9 …{ include: [ "<c10/util/Float8_e5m2fnuz-inl.h>", private, "<c10/util/Float8_e5m2fnuz.h>", public ]… 10 …{ include: [ "<c10/util/Float8_e4m3fnuz-inl.h>", private, "<c10/util/Float8_e4m3fnuz.h>", public ]… 12 { include: [ "<c10/util/complex_math.h>", private, "<c10/util/complex.h>", public ] }, 13 { include: [ "<c10/util/complex_utils.h>", private, "<c10/util/complex.h>", public ] }, [all …]
|
| /external/pytorch/aten/src/ATen/core/op_registration/ |
| D | op_registration_test.cpp | 23 using c10::RegisterOperators; 24 using c10::OperatorKernel; 25 using c10::OperatorHandle; 26 using c10::Dispatcher; 27 using c10::IValue; 28 using c10::DispatchKey; 53 …auto registrar = c10::RegisterOperators().op(c10::RegisterOperators::options().schema("_test::dumm… in TEST() 58 callOp(*op, dummyTensor(c10::DispatchKey::CUDA)); in TEST() 64 …auto registrar = c10::RegisterOperators().op(c10::RegisterOperators::options().catchAllKernel<Mock… in TEST() 69 callOp(*op, dummyTensor(c10::DispatchKey::CUDA)); in TEST() [all …]
|
| /external/pytorch/torch/csrc/distributed/rpc/ |
| D | request_callback_no_python.h | 15 c10::intrusive_ptr<JitFuture> processMessage( 17 std::vector<c10::Stream> streams) const override; 24 virtual c10::intrusive_ptr<JitFuture> processScriptCall( 26 const std::vector<c10::Stream>& streams) const; 28 virtual c10::intrusive_ptr<JitFuture> processPythonCall( 30 const std::vector<c10::Stream>& streams) const; 32 c10::intrusive_ptr<JitFuture> assignOwnerRRef( 35 const c10::intrusive_ptr<JitFuture>& valueFuture) const; 37 virtual c10::intrusive_ptr<JitFuture> processScriptRemoteCall( 39 const std::vector<c10::Stream>& streams) const; [all …]
|
| /external/pytorch/torch/csrc/cuda/ |
| D | CUDAPluggableAllocator.h | 3 #include <c10/core/Allocator.h> 4 #include <c10/cuda/CUDAGraphsC10Utils.h> 5 #include <c10/cuda/CUDAMacros.h> 6 #include <c10/cuda/CUDAStream.h> 8 #include <c10/cuda/CUDACachingAllocator.h> 41 using streamType = c10::hip::HIPStream; 43 using streamType = c10::cuda::CUDAStream; 47 c10::cuda::CUDACachingAllocator::CUDAAllocator> 50 c10::cuda::CUDACachingAllocator::CUDAAllocator> 55 const std::shared_ptr<c10::cuda::CUDACachingAllocator::CUDAAllocator>& [all …]
|
| /external/pytorch/aten/src/ATen/cpu/vec/vec512/ |
| D | vec512_complex_double.h | 6 #include <c10/util/complex.h> 7 #include <c10/util/irange.h> 22 template <> class Vectorized<c10::complex<double>> { 27 using value_type = c10::complex<double>; 34 Vectorized(c10::complex<double> val) { in Vectorized() 40 Vectorized(c10::complex<double> val1, c10::complex<double> val2, in Vectorized() 41 c10::complex<double> val3, c10::complex<double> val4) { in Vectorized() 51 static Vectorized<c10::complex<double>> blend(const Vectorized<c10::complex<double>>& a, in blend() 52 const Vectorized<c10::complex<double>>& b) { in blend() 53 // convert c10::complex<V> index mask to V index mask: xy -> xxyy in blend() [all …]
|
| D | vec512_qint.h | 10 #include <c10/util/irange.h> 11 #include <c10/util/qint32.h> 12 #include <c10/util/qint8.h> 13 #include <c10/util/quint8.h> 273 struct Vectorized<c10::qint32> : public Vectorizedqi { 288 using int_vec_return_type = std::array<Vectorized<c10::qint32>, 1>; 289 using value_type = c10::qint32::underlying; 298 Vectorized(const c10::qint32& val) { 311 static Vectorized<c10::qint32> loadu(const void* ptr) { 312 return Vectorized<c10::qint32>(ptr); [all …]
|
| /external/pytorch/torch/csrc/distributed/c10d/ |
| D | ProcessGroup.hpp | 11 #include <c10/macros/Macros.h> 80 const c10::intrusive_ptr<::c10d::Store>& store, 83 c10::intrusive_ptr<Options> options); 113 virtual void startCoalescing(c10::DeviceType deviceType) { in startCoalescing() 120 virtual c10::intrusive_ptr<Work> endCoalescing(c10::DeviceType deviceType) { in endCoalescing() 128 virtual c10::intrusive_ptr<Work> broadcast( in broadcast() 132 c10::Dispatcher::singleton() in broadcast() 135 std::tuple<std::vector<at::Tensor>, c10::intrusive_ptr<Work>>( in broadcast() 137 const c10::intrusive_ptr<::c10d::ProcessGroup>&, in broadcast() 147 c10::intrusive_ptr<ProcessGroup>::unsafe_reclaim_from_nonowning(this), in broadcast() [all …]
|
| /external/pytorch/torch/csrc/lazy/core/ |
| D | tensor_impl.cpp | 3 #include <c10/core/Allocator.h> 4 #include <c10/core/ScalarType.h> 5 #include <c10/core/impl/DeviceGuardImplInterface.h> 6 #include <c10/macros/Macros.h> 7 #include <c10/util/irange.h> 21 thread_local c10::Device g_device(c10::DeviceType::Lazy); 23 struct LTCGuardImpl : public c10::impl::DeviceGuardImplInterface { 28 c10::Device exchangeDevice(c10::Device device) const override { in exchangeDevice() 29 TORCH_INTERNAL_ASSERT(device.type() == c10::DeviceType::Lazy); in exchangeDevice() 35 c10::Device getDevice() const override { in getDevice() [all …]
|