Skip to content

Commit b005db7

Browse files
ndgrigorianndgrigorian
committed
Refactor size_t to std::size_t and include cstddef everywhere it is used
Maintains stylistic consistency and removes reliance on compiler interpretation
1 parent f7cb1b1 commit b005db7

File tree

150 files changed

+1869
-1730
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

150 files changed

+1869
-1730
lines changed

‎dpctl/_host_task_util.hpp

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -31,6 +31,7 @@
3131

3232
#pragma once
3333
#include <exception>
34+
#include <stddef.h>
3435
#include <sycl/sycl.hpp>
3536

3637
#include "Python.h"

‎dpctl/apis/include/dpctl4pybind11.hpp

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -27,8 +27,10 @@
2727

2828
#include "dpctl_capi.h"
2929
#include <complex>
30+
#include <cstddef> // for std::size_t for C++ linkage
3031
#include <memory>
3132
#include <pybind11/pybind11.h>
33+
#include <stddef.h> // for size_t for C linkage
3234
#include <stdexcept>
3335
#include <sycl/sycl.hpp>
3436
#include <utility>
@@ -759,7 +761,7 @@ class usm_memory : public py::object
759761
* lifetime of the USM allocation.
760762
*/
761763
usm_memory(void *usm_ptr,
762-
size_t nbytes,
764+
std::size_t nbytes,
763765
const sycl::queue &q,
764766
std::shared_ptr<void> shptr)
765767
{
@@ -819,7 +821,7 @@ class usm_memory : public py::object
819821
return reinterpret_cast<char *>(MRef);
820822
}
821823

822-
size_t get_nbytes() const
824+
std::size_t get_nbytes() const
823825
{
824826
auto const &api = ::dpctl::detail::dpctl_capi::get();
825827
Py_MemoryObject *mem_obj = reinterpret_cast<Py_MemoryObject *>(m_ptr);

‎dpctl/tensor/libtensor/include/kernels/accumulators.hpp

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -24,6 +24,7 @@
2424

2525
#pragma once
2626
#include <array>
27+
#include <cstddef>
2728
#include <cstdint>
2829
#include <limits>
2930
#include <new>

‎dpctl/tensor/libtensor/include/kernels/alignment.hpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,7 @@ namespace kernels
3030
namespace alignment_utils
3131
{
3232

33-
static constexpr size_t required_alignment = 64UL;
33+
static constexpr std::size_t required_alignment = 64UL;
3434

3535
template <std::uintptr_t alignment, typename Ptr> bool is_aligned(Ptr p)
3636
{

‎dpctl/tensor/libtensor/include/kernels/boolean_advanced_indexing.hpp

Lines changed: 20 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -23,6 +23,7 @@
2323
//===---------------------------------------------------------------------===//
2424

2525
#pragma once
26+
#include <cstddef>
2627
#include <cstdint>
2728
#include <limits>
2829
#include <sycl/sycl.hpp>
@@ -55,7 +56,7 @@ struct MaskedExtractStridedFunctor
5556
MaskedExtractStridedFunctor(const dataT *src_data_p,
5657
const indT *cumsum_data_p,
5758
dataT *dst_data_p,
58-
size_t masked_iter_size,
59+
std::size_t masked_iter_size,
5960
const OrthogIndexerT &orthog_src_dst_indexer_,
6061
const MaskedSrcIndexerT &masked_src_indexer_,
6162
const MaskedDstIndexerT &masked_dst_indexer_,
@@ -81,7 +82,7 @@ struct MaskedExtractStridedFunctor
8182

8283
const std::size_t max_offset = masked_nelems + 1;
8384
for (std::uint32_t i = l_i; i < lacc.size(); i += lws) {
84-
const size_t offset = masked_block_start + i;
85+
const std::size_t offset = masked_block_start + i;
8586
lacc[i] = (offset == 0) ? indT(0)
8687
: (offset < max_offset) ? cumsum[offset - 1]
8788
: cumsum[masked_nelems - 1] + 1;
@@ -99,9 +100,10 @@ struct MaskedExtractStridedFunctor
99100
if (mask_set && (masked_i < masked_nelems)) {
100101
const auto &orthog_offsets = orthog_src_dst_indexer(orthog_i);
101102

102-
const size_t total_src_offset = masked_src_indexer(masked_i) +
103-
orthog_offsets.get_first_offset();
104-
const size_t total_dst_offset =
103+
const std::size_t total_src_offset =
104+
masked_src_indexer(masked_i) +
105+
orthog_offsets.get_first_offset();
106+
const std::size_t total_dst_offset =
105107
masked_dst_indexer(current_running_count - 1) +
106108
orthog_offsets.get_second_offset();
107109

@@ -113,7 +115,7 @@ struct MaskedExtractStridedFunctor
113115
const dataT *src = nullptr;
114116
const indT *cumsum = nullptr;
115117
dataT *dst = nullptr;
116-
const size_t masked_nelems = 0;
118+
const std::size_t masked_nelems = 0;
117119
// has nd, shape, src_strides, dst_strides for
118120
// dimensions that ARE NOT masked
119121
const OrthogIndexerT orthog_src_dst_indexer;
@@ -136,7 +138,7 @@ struct MaskedPlaceStridedFunctor
136138
MaskedPlaceStridedFunctor(dataT *dst_data_p,
137139
const indT *cumsum_data_p,
138140
const dataT *rhs_data_p,
139-
size_t masked_iter_size,
141+
std::size_t masked_iter_size,
140142
const OrthogIndexerT &orthog_dst_rhs_indexer_,
141143
const MaskedDstIndexerT &masked_dst_indexer_,
142144
const MaskedRhsIndexerT &masked_rhs_indexer_,
@@ -157,12 +159,12 @@ struct MaskedPlaceStridedFunctor
157159
const std::uint32_t l_i = ndit.get_local_id(1);
158160
const std::uint32_t lws = ndit.get_local_range(1);
159161

160-
const size_t masked_i = ndit.get_global_id(1);
161-
const size_t masked_block_start = masked_i - l_i;
162+
const std::size_t masked_i = ndit.get_global_id(1);
163+
const std::size_t masked_block_start = masked_i - l_i;
162164

163165
const std::size_t max_offset = masked_nelems + 1;
164166
for (std::uint32_t i = l_i; i < lacc.size(); i += lws) {
165-
const size_t offset = masked_block_start + i;
167+
const std::size_t offset = masked_block_start + i;
166168
lacc[i] = (offset == 0) ? indT(0)
167169
: (offset < max_offset) ? cumsum[offset - 1]
168170
: cumsum[masked_nelems - 1] + 1;
@@ -180,9 +182,10 @@ struct MaskedPlaceStridedFunctor
180182
if (mask_set && (masked_i < masked_nelems)) {
181183
const auto &orthog_offsets = orthog_dst_rhs_indexer(orthog_i);
182184

183-
const size_t total_dst_offset = masked_dst_indexer(masked_i) +
184-
orthog_offsets.get_first_offset();
185-
const size_t total_rhs_offset =
185+
const std::size_t total_dst_offset =
186+
masked_dst_indexer(masked_i) +
187+
orthog_offsets.get_first_offset();
188+
const std::size_t total_rhs_offset =
186189
masked_rhs_indexer(current_running_count - 1) +
187190
orthog_offsets.get_second_offset();
188191

@@ -194,7 +197,7 @@ struct MaskedPlaceStridedFunctor
194197
dataT *dst = nullptr;
195198
const indT *cumsum = nullptr;
196199
const dataT *rhs = nullptr;
197-
const size_t masked_nelems = 0;
200+
const std::size_t masked_nelems = 0;
198201
// has nd, shape, dst_strides, rhs_strides for
199202
// dimensions that ARE NOT masked
200203
const OrthogIndexerT orthog_dst_rhs_indexer;
@@ -450,8 +453,8 @@ sycl::event masked_extract_some_slices_strided_impl(
450453

451454
const std::size_t lws = get_lws(masked_extent);
452455

453-
const size_t n_groups = ((masked_extent + lws - 1) / lws);
454-
const size_t orthog_extent = static_cast<size_t>(orthog_nelems);
456+
const std::size_t n_groups = ((masked_extent + lws - 1) / lws);
457+
const std::size_t orthog_extent = static_cast<std::size_t>(orthog_nelems);
455458

456459
sycl::range<2> gRange{orthog_extent, n_groups * lws};
457460
sycl::range<2> lRange{1, lws};
@@ -809,7 +812,7 @@ sycl::event non_zero_indexes_impl(sycl::queue &exec_q,
809812
const std::size_t masked_block_start = group_i * lws;
810813

811814
for (std::uint32_t i = l_i; i < lacc.size(); i += lws) {
812-
const size_t offset = masked_block_start + i;
815+
const std::size_t offset = masked_block_start + i;
813816
lacc[i] = (offset == 0) ? indT1(0)
814817
: (offset - 1 < masked_extent)
815818
? cumsum_data[offset - 1]

‎dpctl/tensor/libtensor/include/kernels/clip.hpp

Lines changed: 18 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -25,6 +25,7 @@
2525
#pragma once
2626
#include <algorithm>
2727
#include <complex>
28+
#include <cstddef>
2829
#include <cstdint>
2930
#include <sycl/sycl.hpp>
3031
#include <type_traits>
@@ -85,14 +86,14 @@ template <typename T,
8586
class ClipContigFunctor
8687
{
8788
private:
88-
size_t nelems = 0;
89+
std::size_t nelems = 0;
8990
const T *x_p = nullptr;
9091
const T *min_p = nullptr;
9192
const T *max_p = nullptr;
9293
T *dst_p = nullptr;
9394

9495
public:
95-
ClipContigFunctor(size_t nelems_,
96+
ClipContigFunctor(std::size_t nelems_,
9697
const T *x_p_,
9798
const T *min_p_,
9899
const T *max_p_,
@@ -110,30 +111,30 @@ class ClipContigFunctor
110111
if constexpr (is_complex<T>::value || !enable_sg_loadstore) {
111112
const std::uint16_t sgSize =
112113
ndit.get_sub_group().get_local_range()[0];
113-
const size_t gid = ndit.get_global_linear_id();
114+
const std::size_t gid = ndit.get_global_linear_id();
114115
const uint16_t nelems_per_sg = sgSize * nelems_per_wi;
115116

116-
const size_t start =
117+
const std::size_t start =
117118
(gid / sgSize) * (nelems_per_sg - sgSize) + gid;
118-
const size_t end = std::min(nelems, start + nelems_per_sg);
119+
const std::size_t end = std::min(nelems, start + nelems_per_sg);
119120

120-
for (size_t offset = start; offset < end; offset += sgSize) {
121+
for (std::size_t offset = start; offset < end; offset += sgSize) {
121122
dst_p[offset] = clip(x_p[offset], min_p[offset], max_p[offset]);
122123
}
123124
}
124125
else {
125126
auto sg = ndit.get_sub_group();
126127
const std::uint16_t sgSize = sg.get_max_local_range()[0];
127128

128-
const size_t base =
129+
const std::size_t base =
129130
nelems_per_wi * (ndit.get_group(0) * ndit.get_local_range(0) +
130131
sg.get_group_id()[0] * sgSize);
131132

132133
if (base + nelems_per_wi * sgSize < nelems) {
133134
sycl::vec<T, vec_sz> dst_vec;
134135
#pragma unroll
135136
for (std::uint8_t it = 0; it < n_vecs * vec_sz; it += vec_sz) {
136-
const size_t idx = base + it * sgSize;
137+
const std::size_t idx = base + it * sgSize;
137138
auto x_multi_ptr = sycl::address_space_cast<
138139
sycl::access::address_space::global_space,
139140
sycl::access::decorated::yes>(&x_p[idx]);
@@ -162,8 +163,8 @@ class ClipContigFunctor
162163
}
163164
}
164165
else {
165-
const size_t lane_id = sg.get_local_id()[0];
166-
for (size_t k = base + lane_id; k < nelems; k += sgSize) {
166+
const std::size_t lane_id = sg.get_local_id()[0];
167+
for (std::size_t k = base + lane_id; k < nelems; k += sgSize) {
167168
dst_p[k] = clip(x_p[k], min_p[k], max_p[k]);
168169
}
169170
}
@@ -175,7 +176,7 @@ template <typename T, int vec_sz, int n_vecs> class clip_contig_kernel;
175176

176177
typedef sycl::event (*clip_contig_impl_fn_ptr_t)(
177178
sycl::queue &,
178-
size_t,
179+
std::size_t,
179180
const char *,
180181
const char *,
181182
const char *,
@@ -184,7 +185,7 @@ typedef sycl::event (*clip_contig_impl_fn_ptr_t)(
184185

185186
template <typename T>
186187
sycl::event clip_contig_impl(sycl::queue &q,
187-
size_t nelems,
188+
std::size_t nelems,
188189
const char *x_cp,
189190
const char *min_cp,
190191
const char *max_cp,
@@ -199,10 +200,10 @@ sycl::event clip_contig_impl(sycl::queue &q,
199200
sycl::event clip_ev = q.submit([&](sycl::handler &cgh) {
200201
cgh.depends_on(depends);
201202

202-
size_t lws = 64;
203+
std::size_t lws = 64;
203204
constexpr std::uint8_t vec_sz = 4;
204205
constexpr std::uint8_t n_vecs = 2;
205-
const size_t n_groups =
206+
const std::size_t n_groups =
206207
((nelems + lws * n_vecs * vec_sz - 1) / (lws * n_vecs * vec_sz));
207208
const auto gws_range = sycl::range<1>(n_groups * lws);
208209
const auto lws_range = sycl::range<1>(lws);
@@ -258,7 +259,7 @@ template <typename T, typename IndexerT> class ClipStridedFunctor
258259

259260
void operator()(sycl::id<1> id) const
260261
{
261-
size_t gid = id[0];
262+
std::size_t gid = id[0];
262263
auto offsets = indexer(static_cast<ssize_t>(gid));
263264
dst_p[offsets.get_fourth_offset()] = clip(
264265
x_p[offsets.get_first_offset()], min_p[offsets.get_second_offset()],
@@ -270,7 +271,7 @@ template <typename T, typename IndexerT> class clip_strided_kernel;
270271

271272
typedef sycl::event (*clip_strided_impl_fn_ptr_t)(
272273
sycl::queue &,
273-
size_t,
274+
std::size_t,
274275
int,
275276
const char *,
276277
const char *,
@@ -285,7 +286,7 @@ typedef sycl::event (*clip_strided_impl_fn_ptr_t)(
285286

286287
template <typename T>
287288
sycl::event clip_strided_impl(sycl::queue &q,
288-
size_t nelems,
289+
std::size_t nelems,
289290
int nd,
290291
const char *x_cp,
291292
const char *min_cp,

‎dpctl/tensor/libtensor/include/kernels/constructors.hpp

Lines changed: 9 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -29,6 +29,7 @@
2929
#include "utils/strided_iters.hpp"
3030
#include "utils/type_utils.hpp"
3131
#include <complex>
32+
#include <cstddef>
3233
#include <sycl/sycl.hpp>
3334

3435
namespace dpctl
@@ -97,7 +98,7 @@ template <typename Ty> class LinearSequenceStepFunctor
9798
*/
9899
template <typename Ty>
99100
sycl::event lin_space_step_impl(sycl::queue &exec_q,
100-
size_t nelems,
101+
std::size_t nelems,
101102
Ty start_v,
102103
Ty step_v,
103104
char *array_data,
@@ -123,10 +124,10 @@ template <typename Ty, typename wTy> class LinearSequenceAffineFunctor
123124
Ty *p = nullptr;
124125
Ty start_v;
125126
Ty end_v;
126-
size_t n;
127+
std::size_t n;
127128

128129
public:
129-
LinearSequenceAffineFunctor(char *dst_p, Ty v0, Ty v1, size_t den)
130+
LinearSequenceAffineFunctor(char *dst_p, Ty v0, Ty v1, std::size_t den)
130131
: p(reinterpret_cast<Ty *>(dst_p)), start_v(v0), end_v(v1),
131132
n((den == 0) ? 1 : den)
132133
{
@@ -188,7 +189,7 @@ template <typename Ty, typename wTy> class LinearSequenceAffineFunctor
188189
*/
189190
template <typename Ty>
190191
sycl::event lin_space_affine_impl(sycl::queue &exec_q,
191-
size_t nelems,
192+
std::size_t nelems,
192193
Ty start_v,
193194
Ty end_v,
194195
bool include_endpoint,
@@ -238,7 +239,7 @@ sycl::event lin_space_affine_impl(sycl::queue &exec_q,
238239
*/
239240
template <typename dstTy>
240241
sycl::event full_contig_impl(sycl::queue &q,
241-
size_t nelems,
242+
std::size_t nelems,
242243
dstTy fill_v,
243244
char *dst_p,
244245
const std::vector<sycl::event> &depends)
@@ -294,7 +295,7 @@ template <typename Ty, typename IndexerT> class FullStridedFunctor
294295
template <typename dstTy>
295296
sycl::event full_strided_impl(sycl::queue &q,
296297
int nd,
297-
size_t nelems,
298+
std::size_t nelems,
298299
const ssize_t *shape_strides,
299300
dstTy fill_v,
300301
char *dst_p,
@@ -321,7 +322,7 @@ sycl::event full_strided_impl(sycl::queue &q,
321322
/* ================ Eye ================== */
322323

323324
typedef sycl::event (*eye_fn_ptr_t)(sycl::queue &,
324-
size_t nelems, // num_elements
325+
std::size_t nelems, // num_elements
325326
ssize_t start,
326327
ssize_t end,
327328
ssize_t step,
@@ -375,7 +376,7 @@ template <typename Ty> class EyeFunctor
375376
*/
376377
template <typename Ty>
377378
sycl::event eye_impl(sycl::queue &exec_q,
378-
size_t nelems,
379+
std::size_t nelems,
379380
const ssize_t start,
380381
const ssize_t end,
381382
const ssize_t step,

0 commit comments

Comments
 (0)