aboutsummaryrefslogblamecommitdiffstats
path: root/include/os/linux/ioctl_ctrl.h
blob: 3e1f798bf65def7913424ee5a2a76604328d0ede (plain) (tree)


























                                                                                  
/*
 * Copyright (c) 2011-2017, NVIDIA CORPORATION.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
#ifndef __NVGPU_IOCTL_CTRL_H__
#define __NVGPU_IOCTL_CTRL_H__

int gk20a_ctrl_dev_open(struct inode *inode, struct file *filp);
int gk20a_ctrl_dev_release(struct inode *inode, struct file *filp);
long gk20a_ctrl_dev_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
int gk20a_ctrl_dev_mmap(struct file *filp, struct vm_area_struct *vma);

void nvgpu_hide_usermode_for_poweroff(struct gk20a *g);
void nvgpu_restore_usermode_for_poweron(struct gk20a *g);

#endif
f='#n144'>144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
/* Please note:
 * This code is the optimized version of the first version of Matrix
 * Stressmark. It uses less temporary vectors and vsariables, thus reduce
 * memory allocation/deallocation overhead. the simulation is faster
 */
/*
 *  Sample code for the DIS Matrix Stressmark
 *
 * This source code is the completely correct source code based on
 * the example codes provided by Atlantic Aerospace Division, Titan
 * Systems Corporation, 2000.
 *
 * If you just compile and generate the executables from this source
 * code, this code would be enough. However, if you wish to get a complete
 * understanding of this stressmark, it is strongly suggested that you
 * read the Benchmark Analysis and Specifications Document Version 1.0
 * before going on since the detailed comments are given in this documents.
 * the comments are not repeated here.
 */

/*
 *  The Sparse Matrix Storage is implemented by Compact Row Storage Scheme
 *  In the code, the data is first generated by randomNonzeroFloat()
 *  the data is first stored in a full-space matrix with size of dim*dim
 *  then the data is transfered to the Compact Row Matrix,
 *  the data value is kept in *value,
 *  the columns corresponding to the value are stored in *col_ind,
 *  the start element of each row is stored in *row_start.
 */

/*
 * Please note:
 * the total number of data is numberNonzero +dim
 * among which, NumberNonzero because this is symmetric matrix
 * dim because the diagonal elements
 */

#include "DISstressmarkRNG.h"
#include "extra.h"
#include <assert.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>

#define MIN_SEED -2147483647
#define MAX_SEED -1
#define MIN_DIM 1
#define MAX_DIM 32768
#define MAX_ITERATIONS 65536
#define MIN_TOLERANCE 1e-7 // 0.000007
#define MAX_TOLERANCE 0.5
#define MIN_NUMBER -3.4e10 / dim
#define MAX_NUMBER 3.4e10 / dim
#define EPSI 1.0e-10
#define MIN_DIG_NUMBER 1.0e-10
#define MAX_DIG_NUMBER 3.4e10

/*
 *  External variable, dimension
 */

static int dim;

/*
 *  matrix * vector
 */

void matrixMulvector(double *value, int *col_ind, int *row_start,
                     double *vector, double *out) {
  int l, ll;
  double sum;
  int tmp_rs, tmp_re;

  for (l = 0; l < dim; l++) {
    *(out + l) = 0;
    tmp_rs = row_start[l];

    if (tmp_rs != -1) {
      tmp_re = row_start[l + 1]; /*
                                  *get the start and ending elements of
                                  *  each row
                                  */
      for (ll = tmp_rs; ll < tmp_re; ll++) {
        *(out + l) += value[ll] * vector[col_ind[ll]];
      }
    }
  }
  return;
}

/*
 *    vector1 - vector2
 */

void vectorSub(double *vector1, double *vector2, double *vector) {

  int l;

  for (l = 0; l < dim; l++) {
    *(vector + l) = *(vector1 + l) - *(vector2 + l);
  }
  return;
}

/*
 * vector1 + vector2
 */

void vectorAdd(double *vector1, double *vector2, double *vector) {

  int l;

  for (l = 0; l < dim; l++) {
    *(vector + l) = *(vector1 + l) + *(vector2 + l);
  }
  return;
}

/*
 * vector1 * vector2
 */

double vectorMul(double *vector1, double *vector2) {

  int l;
  double product;

  product = 0;

  for (l = 0; l < dim; l++) {
    product += (*(vector1 + l)) * (*(vector2 + l));
  }
  return product;
}

/*
 * /vector/
 */

double vectorValue(double *vector) {

  double value;
  int l;

  value = 0;

  for (l = 0; l < dim; l++) {
    value += (*(vector + l)) * (*(vector + l));
  }

  return (sqrt(value));
}

/*
 * transpose(vector)
 * In fact, we return the original vector here
 */

void transpose(double *vector, double *vect) {

  int l;

  for (l = 0; l < dim; l++) {
    *(vect + l) = *(vector + l);
  }
  return;
}

/*
 * value * <vector>
 */
void valueMulvector(double value, double *vector, double *vect) {

  int l;
  int lll, i;
  double tmp;

  for (l = 0; l < dim; l++) {
    *(vect + l) = *(vector + l) * value;
  }
  return;
}

/*
 * generate the data distributed sparsely in matrix
 */

void initMatrix(double *matrix, int dim, int numberNonzero) {

  int k, l, ll;
  int i, j;

  int lll;
  double sum;

  for (k = 0; k < dim * dim; k++) {
    *(matrix + k) = 0;
  }

  for (l = 0; l < numberNonzero / 2; l++) {

    i = randomUInt(1, dim - 1);
    j = randomUInt(0, i - 1);

    while (*(matrix + i * dim + j) != 0) {

      i++;
      if (i == dim) {
        j++;
        if (j == dim - 1) {
          j = 0;
          i = 1;
        } else {
          i = j + 1;
        }
      }
    }

    if (*(matrix + i * dim + j) == 0) {
      *(matrix + i * dim + j) =
          (double)randomNonZeroFloat(MIN_NUMBER, MAX_NUMBER, EPSI);
      *(matrix + j * dim + i) = *(matrix + i * dim + j);
    }
  }

  for (ll = 0; ll < dim; ll++) {

    *(matrix + ll * dim + ll) = (double)randomNonZeroFloat(
        -MAX_DIG_NUMBER, MAX_DIG_NUMBER, MIN_DIG_NUMBER);

    sum = 0;

    for (lll = 0; lll < dim; lll++) {
      if (lll != ll) {
        sum += *(matrix + lll * dim + ll);
      }
    }

    if (*(matrix + ll * dim + ll) < sum) {
      *(matrix + ll * dim + ll) += sum;
    }
  }

  return;
}

/*
 * generate the data value in the vectors
 */

void initVector(double *vector, int dim) {

  int l;

  for (l = 0; l < dim; l++) {
    *(vector + l) = (double)randomFloat(MIN_NUMBER, MAX_NUMBER);
  }

  return;
}

/*
 * make a vector contains value of zero
 */

void zeroVector(double *vector, int dim) {
  int l;

  for (l = 0; l < dim; l++) {
    *(vector + l) = 0;
  }
  return;
}

/*
 * return a vector which is the copy of the vect
 */

void equalVector(double *vect, double *vect1) {

  int l;

  for (l = 0; l < dim; l++) {
    *(vect1 + l) = *(vect + l);
  }
  return;
}

void biConjugateGradient(double *value, int *col_ind, int *row_start,
                         double *vectorB, double *vectorX,
                         double errorTolerance, int maxIterations,
                         double *actualError, int *actualIteration, int dim,
                         // Start internal matrixes
                         double *vectorP, double *vectorR, double *nextVectorR,
                         double *tmpVector1, double *tmpVector2,
                         double *tmpVector3)
/*
 * in the code, we use a lot of temparary vectors and variables
 * this is just for simple and clear
 * you can optimize these temporary variables and vectors
 * based on your need
 *
 */
{
  double error;
  int iteration;
  double alpha, beta;

  double tmpValue1, tmpValue2;
  int i;
  int l;
  int ll;

  alpha = 0;
  beta = 0;

  /*
   * vectorR = vectorB - matrixA*vectorX
   */
  matrixMulvector(value, col_ind, row_start, vectorX, tmpVector1);

  vectorSub(vectorB, tmpVector1, vectorR);

  /*
   * vectorP = vectorR
   */

  equalVector(vectorR, vectorP);

  /*
   * error = |matrixA * vectorX - vectorB| / |vectorB|
   */
  vectorSub(tmpVector1, vectorB, tmpVector1);

  error = vectorValue(tmpVector1) / vectorValue(vectorB);

  iteration = 0;

  while ((iteration < maxIterations) && (error > errorTolerance)) {

    /*
     *   alpha = (transpose(vectorR) * vectorR) /
     *           (transpose(vectorP) * (matrixA * vectorP)
     */

    matrixMulvector(value, col_ind, row_start, vectorP, tmpVector1);
    transpose(vectorR, tmpVector2);
    transpose(vectorP, tmpVector3);
    tmpValue1 = vectorMul(tmpVector3, tmpVector1);
    tmpValue2 = vectorMul(tmpVector2, vectorR);
    alpha = tmpValue2 / tmpValue1;

    /*
     * nextVectorR = vectorR - alpha*(matrixA * vectorP)
     */

    valueMulvector(alpha, tmpVector1, tmpVector2);
    vectorSub(vectorR, tmpVector2, tmpVector1);
    equalVector(tmpVector1, nextVectorR);

    /*
     * beta = (transpose(nextVectorR) * nextVectorR) /
     *           (transpose(vectorR) * vectorR)
     */

    transpose(nextVectorR, tmpVector3);
    tmpValue1 = vectorMul(tmpVector3, nextVectorR);
    transpose(vectorR, tmpVector2);
    tmpValue2 = vectorMul(tmpVector2, vectorR);
    beta = tmpValue1 / tmpValue2;

    /*
     * vectorX = vectorX + alpha * vectorP
     */
    valueMulvector(alpha, vectorP, tmpVector1);
    vectorAdd(vectorX, tmpVector1, vectorX);

    /*
     *vectorP = nextVectorR + beta*vectorP
     */
    valueMulvector(beta, vectorP, tmpVector1);
    vectorAdd(nextVectorR, tmpVector1, tmpVector1);

    for (ll = 0; ll < dim; ll++) {
      *(vectorP + ll) = *(tmpVector1 + ll);
    }

    /*
     * vectorR = nextVectorR
     */

    for (l = 0; l < dim; l++) {
      *(vectorR + l) = *(nextVectorR + l);
    }

    /*
     * error = |matrixA * vectorX - vectorB| / |vectorB|
     */
    matrixMulvector(value, col_ind, row_start, vectorX, tmpVector1);
    vectorSub(tmpVector1, vectorB, tmpVector1);
    error = vectorValue(tmpVector1) / vectorValue(vectorB);

    iteration++;
  }

  *actualError = error;
  *actualIteration = iteration;

  return;
}

/*
 * This is the function to transfer the data from the matrix of dense storage
 * to Compact Row Storage
 */
void create_CRS(double *matrixA, double *value, int *col_ind, int *row_start,
                int dim, int numberNonzero) {

  int i, j, k;
  int cnt;
  double tmp;

  /*
   *initialize the row_start
   */

  for (k = 0; k < dim; k++) {
    row_start[k] = -1;
  }

  /*
   * make the end of the last row to be numberNonzero + dim.
   */

  row_start[dim] = numberNonzero + dim;

  /*
   * initialize the col_ind
   */

  for (k = 0; k < numberNonzero + dim; k++) {
    col_ind[k] = -1;
  }

  cnt = 0;

  for (i = 0; (cnt < numberNonzero + dim) && (i < dim); i++) {
    for (j = 0; (cnt < numberNonzero + dim) && (j < dim); j++) {

      tmp = *(matrixA + i * dim + j);

      if (tmp != 0) {

        value[cnt] = tmp;
        col_ind[cnt] = j;

        if (row_start[i] == -1)
          row_start[i] = cnt;

        cnt += 1;
      }
    }
  }
  row_start[i] = cnt;

  return;
}

int main(int argc, char **argv) {
  int seed;
  int numberNonzero;
  int maxIterations;
  float errorTolerance;
  double actualError;
  int actualIteration;

  time_t beginTime;
  time_t endTime;

  double *matrixA;
  double *vectorB;
  double *vectorX;

  double *value;
  int *col_ind;
  int *row_start;
  double sum;
  int k;
  // Internal vects for biConj
  double *vectorR, *vectorP, *nextVectorR;
  double *tmpVector1, *tmpVector2, *tmpVector3