/usr/include/trilinos/Stokhos_DynamicThreadedStorage_cuda.hpp is in libtrilinos-stokhos-dev 12.10.1-3.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 | // @HEADER
// ***********************************************************************
//
// Stokhos Package
// Copyright (2009) Sandia Corporation
//
// Under terms of Contract DE-AC04-94AL85000, there is a non-exclusive
// license for use of this work by or on behalf of the U.S. Government.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact Eric T. Phipps (etphipp@sandia.gov).
//
// ***********************************************************************
// @HEADER
#if defined( __CUDA_ARCH__ )
namespace Stokhos {
template <typename ordinal_t, typename value_t>
class DynamicThreadedStorage<ordinal_t, value_t, Kokkos::Cuda> {
public:
static const bool is_static = false;
static const int static_size = 0;
static const bool supports_reset = true;
typedef ordinal_t ordinal_type;
typedef value_t value_type;
typedef Kokkos::Cuda execution_space;
typedef value_type& reference;
typedef volatile value_type& volatile_reference;
typedef const value_type& const_reference;
typedef const volatile value_type& const_volatile_reference;
typedef value_type* pointer;
typedef volatile value_type* volatile_pointer;
typedef const value_type* const_pointer;
typedef const volatile value_type* const_volatile_pointer;
typedef Stokhos::DynArrayTraits<value_type,execution_space> ds;
//! Turn DynamicThreadedStorage into a meta-function class usable with mpl::apply
template <typename ord_t, typename val_t = value_t , typename dev_t = Kokkos::Cuda >
struct apply {
typedef DynamicThreadedStorage<ord_t,val_t,dev_t> type;
};
//! Constructor
__device__
DynamicThreadedStorage(const ordinal_type& sz = 1,
const value_type& x = value_type(0.0)) :
sz_(sz), stride_(num_threads()), total_sz_(sz_*stride_) {
allocate_coeff_array(coeff_, is_owned_, total_sz_, x);
}
//! Constructor from array
__device__
DynamicThreadedStorage(const ordinal_type& sz, const value_type* x) :
sz_(sz), stride_(num_threads()), total_sz_(sz_*stride_) {
allocate_coeff_array(coeff_, is_owned_, total_sz_, x);
}
//! Constructor for creating a view
__device__
DynamicThreadedStorage(const ordinal_type& sz, pointer v, bool owned) :
coeff_(v), sz_(sz), stride_(num_threads()), total_sz_(sz_*stride_),
is_owned_(owned) {}
//! Constructor
__device__
DynamicThreadedStorage(const DynamicThreadedStorage& s) :
sz_(s.sz_), stride_(s.stride_), total_sz_(s.total_sz_) {
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
//! Constructor
__device__
DynamicThreadedStorage(const volatile DynamicThreadedStorage& s) :
sz_(s.sz_), stride_(s.stride_), total_sz_(s.total_sz_) {
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
//! Destructor
__device__
~DynamicThreadedStorage() {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
}
//! Assignment operator
__device__
DynamicThreadedStorage& operator=(const DynamicThreadedStorage& s) {
if (&s != this) {
if (s.sz_ != sz_) {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
sz_ = s.sz_;
stride_ = s.stride_;
total_sz_ = sz_*stride_;
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
else {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
}
return *this;
}
//! Assignment operator
__device__
DynamicThreadedStorage&
operator=(const volatile DynamicThreadedStorage& s) {
if (&s != this) {
if (s.sz_ != sz_) {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
sz_ = s.sz_;
stride_ = s.stride_;
total_sz_ = sz_*stride_;
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
else {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
}
return *this;
}
//! Assignment operator
__device__
volatile DynamicThreadedStorage&
operator=(const DynamicThreadedStorage& s) volatile {
if (&s != this) {
if (s.sz_ != sz_) {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
sz_ = s.sz_;
stride_ = s.stride_;
total_sz_ = sz_*stride_;
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
else {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
}
return *this;
}
//! Assignment operator
__device__
volatile DynamicThreadedStorage&
operator=(const volatile DynamicThreadedStorage& s) volatile {
if (&s != this) {
if (s.sz_ != sz_) {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
sz_ = s.sz_;
stride_ = s.stride_;
total_sz_ = sz_*stride_;
allocate_coeff_array(coeff_, is_owned_, total_sz_);
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
else {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = s.coeff_[i];
}
}
return *this;
}
//! Initialize values to a constant value
__device__
void init(const_reference v) {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = v;
}
//! Initialize values to a constant value
__device__
void init(const_reference v) volatile {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = v;
}
//! Initialize values to an array of values
__device__
void init(const_pointer v, const ordinal_type& sz = 0) {
ordinal_type my_sz = stride_*sz;
if (sz == 0)
my_sz = total_sz_;
for (ordinal_type i=0; i<my_sz; i+=stride_)
coeff_[i] = v[i];
}
//! Initialize values to an array of values
__device__
void init(const_pointer v, const ordinal_type& sz = 0) volatile {
ordinal_type my_sz = stride_*sz;
if (sz == 0)
my_sz = total_sz_;
for (ordinal_type i=0; i<my_sz; i+=stride_)
coeff_[i] = v[i];
}
//! Load values to an array of values
__device__
void load(pointer v) {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = v[i];
}
//! Load values to an array of values
__device__
void load(pointer v) volatile {
for (ordinal_type i=0; i<total_sz_; i+=stride_)
coeff_[i] = v[i];
}
//! Resize to new size (values are preserved)
__device__
void resize(const ordinal_type& sz) {
if (sz != sz_) {
value_type *coeff_new;
bool owned_new;
ordinal_type total_sz_new = sz*stride_;
allocate_coeff_array(coeff_new, owned_new, total_sz_new);
ordinal_type my_tsz = total_sz_;
if (total_sz_ > total_sz_new)
my_tsz = total_sz_new;
for (ordinal_type i=0; i<my_tsz; i+=stride_)
coeff_new[i] = coeff_[i];
destroy_coeff_array(coeff_, is_owned_, total_sz_);
coeff_ = coeff_new;
sz_ = sz;
total_sz_ = total_sz_new;
is_owned_ = owned_new;
}
}
//! Resize to new size (values are preserved)
__device__
void resize(const ordinal_type& sz) volatile {
if (sz != sz_) {
value_type *coeff_new;
bool owned_new;
ordinal_type total_sz_new = sz*stride_;
allocate_coeff_array(coeff_new, owned_new, total_sz_new);
ordinal_type my_tsz = total_sz_;
if (total_sz_ > total_sz_new)
my_tsz = total_sz_new;
for (ordinal_type i=0; i<my_tsz; i+=stride_)
coeff_new[i] = coeff_[i];
destroy_coeff_array(coeff_, is_owned_, total_sz_);
coeff_ = coeff_new;
sz_ = sz;
total_sz_ = total_sz_new;
is_owned_ = owned_new;
}
}
//! Reset storage to given array, size, and stride
__device__
void shallowReset(pointer v, const ordinal_type& sz,
const ordinal_type& stride, bool owned) {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
coeff_ = v;
sz_ = sz;
stride_ = stride;
total_sz_ = sz_*stride_;
is_owned_ = owned;
}
//! Reset storage to given array, size, and stride
__device__
void shallowReset(pointer v, const ordinal_type& sz,
const ordinal_type& stride, bool owned) volatile {
destroy_coeff_array(coeff_, is_owned_, total_sz_);
coeff_ = v;
sz_ = sz;
stride_ = stride;
total_sz_ = sz_*stride_;
is_owned_ = owned;
}
//! Return size
__device__
ordinal_type size() const { return sz_; }
//! Return size
__device__
ordinal_type size() const volatile { return sz_; }
//! Coefficient access (avoid if possible)
KOKKOS_INLINE_FUNCTION
const_reference operator[] (const ordinal_type& i) const {
return coeff_[i*stride_];
}
//! Coefficient access (avoid if possible)
KOKKOS_INLINE_FUNCTION
const_volatile_reference operator[] (const ordinal_type& i) const volatile {
return coeff_[i*stride_];
}
//! Coefficient access (avoid if possible)
KOKKOS_INLINE_FUNCTION
reference operator[] (const ordinal_type& i) {
return coeff_[i*stride_];
}
//! Coefficient access (avoid if possible)
KOKKOS_INLINE_FUNCTION
volatile_reference operator[] (const ordinal_type& i) volatile {
return coeff_[i*stride_];
}
template <int i>
KOKKOS_INLINE_FUNCTION
reference getCoeff() { return coeff_[i*stride_]; }
template <int i>
KOKKOS_INLINE_FUNCTION
volatile_reference getCoeff() volatile { return coeff_[i*stride_]; }
template <int i>
KOKKOS_INLINE_FUNCTION
const_reference getCoeff() const { return coeff_[i*stride_]; }
template <int i>
KOKKOS_INLINE_FUNCTION
const_volatile_reference getCoeff() const volatile { return coeff_[i*stride_]; }
//! Get coefficients
KOKKOS_INLINE_FUNCTION
const_volatile_pointer coeff() const volatile { return coeff_; }
//! Get coefficients
KOKKOS_INLINE_FUNCTION
const_pointer coeff() const { return coeff_; }
//! Get coefficients
KOKKOS_INLINE_FUNCTION
volatile_pointer coeff() volatile { return coeff_; }
//! Get coefficients
KOKKOS_INLINE_FUNCTION
pointer coeff() { return coeff_; }
protected:
//! Compute number of threads in each block
__device__
ordinal_type num_threads() const {
return blockDim.x*blockDim.y*blockDim.z;
}
//! Compute number of threads in each block
__device__
ordinal_type num_threads() const volatile {
return blockDim.x*blockDim.y*blockDim.z;
}
//! Compute thread index within a block
__device__
ordinal_type thread_index() const {
return threadIdx.x + (threadIdx.y + threadIdx.z*blockDim.y)*blockDim.x;
}
//! Compute thread index within a block
__device__
ordinal_type thread_index() const volatile {
return threadIdx.x + (threadIdx.y + threadIdx.z*blockDim.y)*blockDim.x;
}
//! Allocate coefficient array
__device__
void allocate_coeff_array(pointer& c, bool& owned,
ordinal_type total_size,
const value_type& x = value_type(0.0)) {
// Allocate coefficient array on thread 0
__shared__ pointer ptr;
ordinal_type tidx = thread_index();
if (tidx == 0) {
ptr = ds::get_and_fill(total_size,x);
owned = true;
}
else
owned = false;
__syncthreads();
// Give each thread its portion of the array
c = ptr + tidx;
}
//! Allocate coefficient array
__device__
void allocate_coeff_array(pointer& c, bool& owned,
ordinal_type total_size,
const value_type& x = value_type(0.0)) volatile {
// Allocate coefficient array on thread 0
__shared__ pointer ptr;
ordinal_type tidx = thread_index();
if (tidx == 0) {
ptr = ds::get_and_fill(total_size,x);
owned = true;
}
else
owned = false;
__syncthreads();
// Give each thread its portion of the array
c = ptr + tidx;
}
//! Allocate coefficient array
__device__
void allocate_coeff_array(pointer& c, bool& owned,
ordinal_type total_size,
const value_type* x) {
// Allocate coefficient array on thread 0
__shared__ pointer ptr;
ordinal_type tidx = thread_index();
if (tidx == 0) {
ptr = ds::get_and_fill(x, total_size);
owned = true;
}
else
owned = false;
__syncthreads();
// Give each thread its portion of the array
c = ptr + tidx;
}
//! Destroy coefficient array
__device__
void destroy_coeff_array(pointer c, bool owned, ordinal_type total_size) {
__syncthreads();
if (owned)
ds::destroy_and_release(c, total_size);
}
//! Destroy coefficient array
__device__
void destroy_coeff_array(pointer c, bool owned, ordinal_type total_size) volatile {
__syncthreads();
if (owned)
ds::destroy_and_release(c, total_size);
}
private:
//! Coefficient values
pointer coeff_;
//! Size of array used
ordinal_type sz_;
//! Stride of array
ordinal_type stride_;
//! Total size of array
ordinal_type total_sz_;
//! Do we own the array
bool is_owned_;
};
}
#endif
|