Skip to content

Dedicated code to copy array to C-contig/F-contig destinations #1850

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
oleksandr-pavlyk merged 7 commits into from
Sep 30, 2024
Merged

Dedicated code to copy array to C-contig/F-contig destinations #1850

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
aad28dd
Add functions to copy array into C-/F- contig destination
oleksandr-pavlyk Sep 23, 2024
2ba9829
Small tweaks to copy_numpy_ndarray_into_usm_ndarray validation code
oleksandr-pavlyk Sep 23, 2024
69e17be
Tests to exercise specialized code-paths for as_c_contig/as_f_contig
oleksandr-pavlyk Sep 23, 2024
f4705c0
Add gh-1850 to change-log
oleksandr-pavlyk Sep 23, 2024
610d88d
Address PR review feedback
oleksandr-pavlyk Sep 26, 2024
0899a07
Add include <stdexcept>
oleksandr-pavlyk Sep 27, 2024
d088227
Corrected comment typos
oleksandr-pavlyk Sep 27, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Address PR review feedback
  • Loading branch information
@oleksandr-pavlyk
oleksandr-pavlyk committed Sep 26, 2024
commit 610d88d6d5fc7887627dc7b49e86e8eaf4e07cd4
43 changes: 37 additions & 6 deletions dpctl/tensor/libtensor/source/copy_as_contig.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,7 @@
#include "kernels/copy_as_contiguous.hpp"
#include "utils/memory_overlap.hpp"
#include "utils/offset_utils.hpp"
#include "utils/output_validation.hpp"
#include "utils/sycl_alloc_utils.hpp"
#include "utils/type_dispatch.hpp"

Expand Down Expand Up @@ -107,10 +108,16 @@ void init_copy_as_contig_dispatch_vectors(void)
namespace
{

template <typename dimT> dimT get_nelems(const std::vector<dimT> &shape)
template <typename dimT> std::size_t get_nelems(const std::vector<dimT> &shape)
{
const dimT nelems = std::accumulate(std::begin(shape), std::end(shape),
dimT(1), std::multiplies<dimT>{});
auto mult_fn = [](std::size_t prod, const dimT &term) -> std::size_t {
return prod * static_cast<std::size_t>(term);
};

constexpr std::size_t unit{1};

const std::size_t nelems =
std::accumulate(std::begin(shape), std::end(shape), unit, mult_fn);
return nelems;
}

Expand Down Expand Up @@ -163,6 +170,14 @@ py_as_c_contig(const dpctl::tensor::usm_ndarray &src,
throw py::value_error("Destination array must be C-contiguous");
}

dpctl::tensor::validation::CheckWritable::throw_if_not_writable(dst);

// check compatibility of execution queue and allocation queue
if (!dpctl::utils::queues_are_compatible(exec_q, {src, dst})) {
throw py::value_error(
"Execution queue is not compatible with allocation queues");
}

const auto &src_strides_vec = src.get_strides_vector();

if (src_nd >= 2) {
Expand All @@ -175,7 +190,7 @@ py_as_c_contig(const dpctl::tensor::usm_ndarray &src,
}
}

const py::ssize_t nelems = get_nelems(src_shape_vec);
const std::size_t nelems = get_nelems(src_shape_vec);

if (nelems == 0) {
// nothing to do
Expand Down Expand Up @@ -254,7 +269,7 @@ py_as_f_contig(const dpctl::tensor::usm_ndarray &src,
const std::vector<sycl::event> &depends)
{
/* Same dimensions, same shape, same data-type
* dst is C-contiguous.
* dst is F-contiguous.
*/
int src_nd = src.get_ndim();
int dst_nd = dst.get_ndim();
Expand Down Expand Up @@ -288,6 +303,14 @@ py_as_f_contig(const dpctl::tensor::usm_ndarray &src,
throw py::value_error("Destination array must be F-contiguous");
}

dpctl::tensor::validation::CheckWritable::throw_if_not_writable(dst);

// check compatibility of execution queue and allocation queue
if (!dpctl::utils::queues_are_compatible(exec_q, {src, dst})) {
throw py::value_error(
"Execution queue is not compatible with allocation queues");
}

const auto &src_strides_vec = src.get_strides_vector();

if (src_nd >= 2) {
Expand All @@ -300,7 +323,7 @@ py_as_f_contig(const dpctl::tensor::usm_ndarray &src,
}
}

const py::ssize_t nelems = get_nelems(src_shape_vec);
const std::size_t nelems = get_nelems(src_shape_vec);

if (nelems == 0) {
// nothing to do
Expand Down Expand Up @@ -433,6 +456,14 @@ py_as_c_contig_f2c(const dpctl::tensor::usm_ndarray &src,
throw py::value_error("Destination array must be C-contiguous");
}

dpctl::tensor::validation::CheckWritable::throw_if_not_writable(dst);

// check compatibility of execution queue and allocation queue
if (!dpctl::utils::queues_are_compatible(exec_q, {src, dst})) {
throw py::value_error(
"Execution queue is not compatible with allocation queues");
}

if (nelems == 0) {
// nothing to do
return std::make_pair(sycl::event(), sycl::event());
Expand Down