Files
composable_kernel/include/ck/utility/tuple.hpp
arai713 2e3183af4f Codegen hipRTC compilation (#1579)
* updating codegen build for MIOpen access: adding .cmake for codegen component

* updating CMake

* adding in header guards for some headers due to issues with hiprtc compilation in MIOpen

* some more header guards

* putting env file in header guard

* cleaning up some includes

* updated types file for hiprtc purposes

* fixed types file: bit-wise/memcpy issue

* updating multiple utility files to deal with standard header inclusion for hiprtc

* added some more header guards in the utility files, replacing some standard header functionality

* added some more header guards

* fixing some conflicts in utility files, another round of header guards

* fixing errors in data type file

* resolved conflict errors in a few utility files

* added header guards/replicated functionality in device files

* resolved issues with standard headers in device files: device_base and device_grouped_conv_fwd_multiple_abd

* resolved issues with standard headers in device files: device_base.hpp, device_grouped_conv_fwd_multiple_abd.hpp, device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp

* added header guards for gridwise gemm files: gridwise_gemm_multiple_abd_xdl_cshuffle.hpp and gridwise_gemm_multiple_d_xdl_cshuffle.hpp

* fixed issue with numerics header, removed from transform_conv_fwd_to_gemm and added to device_column_to_image_impl, device_grouped_conv_fwd_multiple_abd_xdl_cshuffle, device_grouped_conv_fwd_multiple_abd_xdl_cshuffle_v3, device_image_to_column_impl

* replaced standard header usage and added header guards in block to ctile map and gridwise_gemm_pipeline_selector

* resolved errors in device_gemm_xdl_splitk_c_shuffle files in regards to replacement of standard headers in previous commit

* added replicated functionality for standard header methods in utility files

* replaced standard header functionality in threadwise tensor slice transfer files and added header guards in element_wise_operation.hpp

* temp fix for namespace error in MIOpen

* remove standard header usage in codegen device op

* removed standard header usage in elementwise files, resolved namespace errors

* formatting fix

* changed codegen argument to ON for testing

* temporarily removing codegen compiler flag for testing purposes

* added codegen flag again, set default to ON

* set codegen flag default back to OFF

* replaced enable_if_t standard header usage in data_type.hpp

* added some debug prints to pinpoint issues in MIOpen

* added print outs to debug in MIOpen

* removed debug print outs from device op

* resolved stdexcept include error

* formatting fix

* adding includes to new fp8 file to resolve ck::enable_if_t errors

* made changes to amd_wave_read_first_lane

* updated functionality in type utility file

* fixed end of file issue

* resovled errors in type utility file, added functionality to array utility file

* fixed standard header usage replication in data_type file, resolves error with failing examples on navi3x

* formatting fix

* replaced standard header usage in amd_ck_fp8 file

* added include to random_gen file

* removed and replicated standard header usage from data_type and type_convert files for fp8 changes

* replicated standard unsigned integer types in random_gen

* resolved comments from review: put calls to reinterpret_cast for size_t in header guards

* updated/added copyright headers

* removed duplicate header

* fixed typo in header guard

* updated copyright headers

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
2025-01-31 09:48:39 -08:00

224 lines
6.5 KiB
C++

// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2025, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include "ck/utility/integral_constant.hpp"
#include "ck/utility/sequence.hpp"
#include "ck/utility/type.hpp"
#include "ck/utility/enable_if.hpp"
namespace ck {
namespace detail {
template <index_t>
struct TupleElementKey
{
__host__ __device__ constexpr TupleElementKey() = default;
};
template <typename Key, typename Data>
struct TupleElementKeyData
{
using DataType = Data;
#if 0 // workaround compiler complaint about implicitly-deleted default constructor
__host__ __device__ constexpr TupleElementKeyData() = default;
#else
__host__ __device__ constexpr TupleElementKeyData() : mData{} {}
#endif
template <typename T,
typename enable_if<!is_same<remove_cvref_t<T>, TupleElementKeyData>::value,
bool>::type = false>
__host__ __device__ constexpr TupleElementKeyData(T&& v) : mData(ck::forward<T>(v))
{
}
DataType mData;
};
// for read access of tuple element
template <typename Key, typename Data>
__host__ __device__ constexpr const Data&
get_tuple_element_data_reference(const TupleElementKeyData<Key, Data>& x)
{
return static_cast<const Data&>(x.mData);
}
// for write access of tuple element
template <typename Key, typename Data>
__host__ __device__ constexpr Data&
get_tuple_element_data_reference(TupleElementKeyData<Key, Data>& x)
{
return x.mData;
}
// TODO: not sure the use of reference is correct
template <typename Key, typename Data>
__host__ __device__ constexpr Data&&
get_tuple_element_data_reference(TupleElementKeyData<Key, Data>&& x)
{
return static_cast<Data&&>(x.mData);
}
// for infering type of tuple element
template <typename Key, typename Data>
__host__ __device__ constexpr Data get_tuple_element_data(const TupleElementKeyData<Key, Data>& x)
{
return ck::forward(x.mData);
}
template <typename Indices, typename... Xs>
struct TupleImpl;
template <index_t... Is, typename... Xs>
struct TupleImpl<Sequence<Is...>, Xs...> : TupleElementKeyData<TupleElementKey<Is>, Xs>...
{
__host__ __device__ constexpr TupleImpl() = default;
template <typename Y,
typename enable_if<sizeof...(Is) == 1 && sizeof...(Xs) == 1 &&
!is_same<remove_cvref_t<Y>, TupleImpl>::value,
bool>::type = false>
__host__ __device__ constexpr TupleImpl(Y&& y)
: TupleElementKeyData<TupleElementKey<Is>, Xs>(ck::forward<Y>(y))...
{
}
template <typename... Ys, typename enable_if<sizeof...(Ys) >= 2, bool>::type = false>
__host__ __device__ constexpr TupleImpl(Ys&&... ys)
: TupleElementKeyData<TupleElementKey<Is>, Xs>(ck::forward<Ys>(ys))...
{
static_assert(sizeof...(Is) == sizeof...(Xs) && sizeof...(Is) == sizeof...(Ys),
"wrong! inconsistent size");
}
__host__ __device__ static constexpr index_t Size() { return sizeof...(Xs); }
template <index_t I>
__host__ __device__ constexpr const auto& GetElementDataByKey(TupleElementKey<I>) const
{
return get_tuple_element_data_reference<TupleElementKey<I>>(*this);
}
template <index_t I>
__host__ __device__ constexpr auto& GetElementDataByKey(TupleElementKey<I>)
{
return get_tuple_element_data_reference<TupleElementKey<I>>(*this);
}
};
} // namespace detail
template <typename... Xs>
struct Tuple : detail::TupleImpl<typename arithmetic_sequence_gen<0, sizeof...(Xs), 1>::type, Xs...>
{
using base =
detail::TupleImpl<typename arithmetic_sequence_gen<0, sizeof...(Xs), 1>::type, Xs...>;
__host__ __device__ constexpr Tuple() = default;
template <typename Y,
typename enable_if<sizeof...(Xs) == 1 && !is_same<remove_cvref_t<Y>, Tuple>::value,
bool>::type = false>
__host__ __device__ constexpr Tuple(Y&& y) : base(ck::forward<Y>(y))
{
}
template <typename... Ys,
typename enable_if<sizeof...(Ys) == sizeof...(Xs) && sizeof...(Ys) >= 2, bool>::type =
false>
__host__ __device__ constexpr Tuple(Ys&&... ys) : base(ck::forward<Ys>(ys)...)
{
}
__host__ __device__ static constexpr index_t Size() { return sizeof...(Xs); }
// read access
template <index_t I>
__host__ __device__ constexpr const auto& At(Number<I>) const
{
static_assert(I < base::Size(), "wrong! out of range");
return base::GetElementDataByKey(detail::TupleElementKey<I>{});
}
// write access
template <index_t I>
__host__ __device__ constexpr auto& At(Number<I>)
{
static_assert(I < base::Size(), "wrong! out of range");
return base::GetElementDataByKey(detail::TupleElementKey<I>{});
}
// read access
template <index_t I>
__host__ __device__ constexpr const auto& operator[](Number<I> i) const
{
return At(i);
}
// write access
template <index_t I>
__host__ __device__ constexpr auto& operator()(Number<I> i)
{
return At(i);
}
template <typename T>
__host__ __device__ constexpr auto operator=(const T& a)
{
static_assert(T::Size() == Size(), "wrong! size not the same");
static_for<0, Size(), 1>{}([&](auto i) { operator()(i) = a[i]; });
return *this;
}
__host__ __device__ static constexpr bool IsStaticBuffer() { return true; }
__host__ __device__ static constexpr bool IsTuple() { return true; }
};
template <>
struct Tuple<>
{
__host__ __device__ constexpr Tuple() = default;
__host__ __device__ static constexpr index_t Size() { return 0; }
template <typename T>
__host__ __device__ constexpr auto operator=(const T&)
{
return *this;
}
__host__ __device__ static constexpr bool IsStaticBuffer() { return true; }
};
template <index_t I, typename TTuple>
struct tuple_element
{
// type should keep the cv/ref qualifier of original tuple element
using type = decltype(detail::get_tuple_element_data<detail::TupleElementKey<I>>(TTuple{}));
};
template <index_t I, typename TTuple>
using tuple_element_t = typename tuple_element<I, TTuple>::type;
template <typename... Xs>
__host__ __device__ constexpr auto make_tuple(Xs&&... xs)
{
return Tuple<remove_cvref_t<Xs>...>(ck::forward<Xs>(xs)...);
}
// https://en.cppreference.com/w/cpp/utility/tuple/tie
template <typename... Args>
constexpr Tuple<Args&...> tie(Args&... args) noexcept
{
return {args...};
}
} // namespace ck