|
| 1 | +//***************************************************************************** |
| 2 | +// Copyright (c) 2026, Intel Corporation |
| 3 | +// All rights reserved. |
| 4 | +// |
| 5 | +// Redistribution and use in source and binary forms, with or without |
| 6 | +// modification, are permitted provided that the following conditions are met: |
| 7 | +// - Redistributions of source code must retain the above copyright notice, |
| 8 | +// this list of conditions and the following disclaimer. |
| 9 | +// - Redistributions in binary form must reproduce the above copyright notice, |
| 10 | +// this list of conditions and the following disclaimer in the documentation |
| 11 | +// and/or other materials provided with the distribution. |
| 12 | +// - Neither the name of the copyright holder nor the names of its contributors |
| 13 | +// may be used to endorse or promote products derived from this software |
| 14 | +// without specific prior written permission. |
| 15 | +// |
| 16 | +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
| 17 | +// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 18 | +// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 19 | +// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE |
| 20 | +// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 21 | +// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 22 | +// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 23 | +// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 24 | +// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 25 | +// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF |
| 26 | +// THE POSSIBILITY OF SUCH DAMAGE. |
| 27 | +//***************************************************************************** |
| 28 | + |
| 29 | +#include <stdexcept> |
| 30 | +#include <type_traits> |
| 31 | +#include <vector> |
| 32 | + |
| 33 | +#include <sycl/sycl.hpp> |
| 34 | + |
| 35 | +#include "dpctl4pybind11.hpp" |
| 36 | +#include <pybind11/pybind11.h> |
| 37 | +#include <pybind11/stl.h> |
| 38 | + |
| 39 | +#include "putmask_kernel.hpp" |
| 40 | + |
| 41 | +#include "../elementwise_functions/simplify_iteration_space.hpp" |
| 42 | + |
| 43 | +// dpctl tensor headers |
| 44 | +#include "utils/offset_utils.hpp" |
| 45 | +#include "utils/output_validation.hpp" |
| 46 | +#include "utils/type_dispatch.hpp" |
| 47 | +#include "utils/type_utils.hpp" |
| 48 | + |
| 49 | +// utils extension headers |
| 50 | +#include "ext/common.hpp" |
| 51 | +#include "ext/validation_utils.hpp" |
| 52 | + |
| 53 | +namespace py = pybind11; |
| 54 | +namespace td_ns = dpctl::tensor::type_dispatch; |
| 55 | + |
| 56 | +using dpctl::tensor::usm_ndarray; |
| 57 | + |
| 58 | +using ext::common::dtype_from_typenum; |
| 59 | +using ext::validation::array_names; |
| 60 | +using ext::validation::check_has_dtype; |
| 61 | +using ext::validation::check_no_overlap; |
| 62 | +using ext::validation::check_num_dims; |
| 63 | +using ext::validation::check_queue; |
| 64 | +using ext::validation::check_same_dtype; |
| 65 | +using ext::validation::check_same_size; |
| 66 | +using ext::validation::check_writable; |
| 67 | + |
| 68 | +namespace dpnp::extensions::indexing |
| 69 | +{ |
| 70 | +using ext::common::init_dispatch_vector; |
| 71 | + |
| 72 | +typedef sycl::event (*putmask_contig_fn_ptr_t)( |
| 73 | + sycl::queue &, |
| 74 | + const std::size_t, // nelems |
| 75 | + char *, // dst |
| 76 | + const char *, // mask |
| 77 | + const char *, // values |
| 78 | + const std::size_t, // values_size |
| 79 | + const std::vector<sycl::event> &); |
| 80 | + |
| 81 | +static putmask_contig_fn_ptr_t putmask_contig_dispatch_vector[td_ns::num_types]; |
| 82 | + |
| 83 | +std::pair<sycl::event, sycl::event> |
| 84 | + py_putmask(const usm_ndarray &dst, |
| 85 | + const usm_ndarray &mask, |
| 86 | + const usm_ndarray &values, |
| 87 | + sycl::queue &exec_q, |
| 88 | + const std::vector<sycl::event> &depends = {}) |
| 89 | +{ |
| 90 | + array_names names = {{&dst, "dst"}, {&mask, "mask"}, {&values, "values"}}; |
| 91 | + |
| 92 | + check_same_dtype(&dst, &values, names); |
| 93 | + check_has_dtype(&mask, td_ns::typenum_t::BOOL, names); |
| 94 | + |
| 95 | + check_same_size({&dst, &mask}, names); |
| 96 | + const int nd = dst.get_ndim(); |
| 97 | + check_num_dims(&mask, nd, names); |
| 98 | + |
| 99 | + check_queue({&dst, &mask, &values}, names, exec_q); |
| 100 | + check_no_overlap({&mask, &values}, {&dst}, names); |
| 101 | + check_writable({&dst}, names); |
| 102 | + |
| 103 | + // values must be 1D |
| 104 | + check_num_dims(&values, 1, names); |
| 105 | + |
| 106 | + auto types = td_ns::usm_ndarray_types(); |
| 107 | + // dst_typeid == values_typeid (check_same_dtype(&dst, &values, names)) |
| 108 | + int dst_values_typeid = types.typenum_to_lookup_id(dst.get_typenum()); |
| 109 | + |
| 110 | + const py::ssize_t *dst_shape = dst.get_shape_raw(); |
| 111 | + const py::ssize_t *mask_shape = mask.get_shape_raw(); |
| 112 | + bool shapes_equal(true); |
| 113 | + std::size_t nelems(1); |
| 114 | + |
| 115 | + for (int i = 0; i < std::max(nd, 1); ++i) { |
| 116 | + const py::ssize_t d = (nd == 0 ? 1 : dst_shape[i]); |
| 117 | + const py::ssize_t m = (nd == 0 ? 1 : mask_shape[i]); |
| 118 | + nelems *= static_cast<std::size_t>(d); |
| 119 | + shapes_equal = shapes_equal && (d == m); |
| 120 | + } |
| 121 | + if (!shapes_equal) { |
| 122 | + throw py::value_error("`mask` and `dst` shapes must match"); |
| 123 | + } |
| 124 | + |
| 125 | + // if nelems is zero, return |
| 126 | + if (nelems == 0) { |
| 127 | + return {sycl::event(), sycl::event()}; |
| 128 | + } |
| 129 | + |
| 130 | + dpctl::tensor::validation::AmpleMemory::throw_if_not_ample(dst, nelems); |
| 131 | + |
| 132 | + char *dst_p = dst.get_data(); |
| 133 | + const char *mask_p = mask.get_data(); |
| 134 | + const char *values_p = values.get_data(); |
| 135 | + const std::size_t values_size = values.get_size(); |
| 136 | + |
| 137 | + // handle C contiguous inputs |
| 138 | + const bool is_dst_c_contig = dst.is_c_contiguous(); |
| 139 | + const bool is_mask_c_contig = mask.is_c_contiguous(); |
| 140 | + const bool is_values_c_contig = values.is_c_contiguous(); |
| 141 | + |
| 142 | + const bool all_c_contig = |
| 143 | + (is_dst_c_contig && is_mask_c_contig && is_values_c_contig); |
| 144 | + |
| 145 | + if (all_c_contig) { |
| 146 | + auto contig_fn = putmask_contig_dispatch_vector[dst_values_typeid]; |
| 147 | + |
| 148 | + if (contig_fn == nullptr) { |
| 149 | + py::dtype dst_values_dtype_py = |
| 150 | + dtype_from_typenum(dst_values_typeid); |
| 151 | + throw std::runtime_error( |
| 152 | + "Contiguous implementation is missing for " + |
| 153 | + std::string(py::str(dst_values_dtype_py)) + "data type"); |
| 154 | + } |
| 155 | + |
| 156 | + auto comp_ev = contig_fn(exec_q, nelems, dst_p, mask_p, values_p, |
| 157 | + values_size, depends); |
| 158 | + sycl::event ht_ev = dpctl::utils::keep_args_alive( |
| 159 | + exec_q, {dst, mask, values}, {comp_ev}); |
| 160 | + |
| 161 | + return std::make_pair(ht_ev, comp_ev); |
| 162 | + } |
| 163 | + |
| 164 | + throw py::value_error("Stride implementation is not implemented yet"); |
| 165 | +} |
| 166 | + |
| 167 | +/** |
| 168 | + * @brief A factory to define pairs of supported types for which |
| 169 | + * putmask function is available. |
| 170 | + * |
| 171 | + * @tparam T Type of input vector `dst` and `values` and of result vector `dst`. |
| 172 | + */ |
| 173 | +template <typename T> |
| 174 | +struct PutMaskOutputType |
| 175 | +{ |
| 176 | + using value_type = typename std::disjunction< |
| 177 | + td_ns::TypeMapResultEntry<T, std::uint8_t>, |
| 178 | + td_ns::TypeMapResultEntry<T, std::int8_t>, |
| 179 | + td_ns::TypeMapResultEntry<T, std::uint16_t>, |
| 180 | + td_ns::TypeMapResultEntry<T, std::int16_t>, |
| 181 | + td_ns::TypeMapResultEntry<T, std::uint32_t>, |
| 182 | + td_ns::TypeMapResultEntry<T, std::int32_t>, |
| 183 | + td_ns::TypeMapResultEntry<T, std::uint64_t>, |
| 184 | + td_ns::TypeMapResultEntry<T, std::int64_t>, |
| 185 | + td_ns::TypeMapResultEntry<T, sycl::half>, |
| 186 | + td_ns::TypeMapResultEntry<T, float>, |
| 187 | + td_ns::TypeMapResultEntry<T, double>, |
| 188 | + td_ns::TypeMapResultEntry<T, std::complex<float>>, |
| 189 | + td_ns::TypeMapResultEntry<T, std::complex<double>>, |
| 190 | + td_ns::DefaultResultEntry<void>>::result_type; |
| 191 | +}; |
| 192 | + |
| 193 | +template <typename fnT, typename T> |
| 194 | +struct PutMaskContigFactory |
| 195 | +{ |
| 196 | + fnT get() |
| 197 | + { |
| 198 | + if constexpr (std::is_same_v<typename PutMaskOutputType<T>::value_type, |
| 199 | + void>) { |
| 200 | + return nullptr; |
| 201 | + } |
| 202 | + else { |
| 203 | + return kernels::putmask_contig_impl<T>; |
| 204 | + } |
| 205 | + } |
| 206 | +}; |
| 207 | + |
| 208 | +static void populate_putmask_dispatch_vectors() |
| 209 | +{ |
| 210 | + init_dispatch_vector<putmask_contig_fn_ptr_t, PutMaskContigFactory>( |
| 211 | + putmask_contig_dispatch_vector); |
| 212 | +} |
| 213 | + |
| 214 | +void init_putmask(py::module_ m) |
| 215 | +{ |
| 216 | + populate_putmask_dispatch_vectors(); |
| 217 | + |
| 218 | + m.def("_putmask", &py_putmask, "", py::arg("dst"), py::arg("mask"), |
| 219 | + py::arg("values"), py::arg("sycl_queue"), |
| 220 | + py::arg("depends") = py::list()); |
| 221 | +} |
| 222 | + |
| 223 | +} // namespace dpnp::extensions::indexing |
0 commit comments