Dataset Viewer
Auto-converted to Parquet Duplicate
file_path
stringlengths
19
84
content
stringlengths
235
1.29M
./openacc-vv/serial_loop_reduction_multiply_general.F90
#ifndef T1 !T1:serial,reduction,combined-constructs,loop,V:2.6-2.7 LOGICAL FUNCTION test1() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" REAL(8),DIMENSION(10):: a, b REAL(8):: reduced, host_reduced INTEGER:: errors, x, y errors = 0 SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) DO y = 1, LOOPCOUNT CALL RANDOM_NUMBER(a) CALL RANDOM_NUMBER(b) reduced = 1 host_reduced = 1 DO x = 1, 10 host_reduced = host_reduced * (a(x) + b(x)) END DO !$acc data copyin(a(1:10), b(1:10)) !$acc serial loop reduction(*:reduced) DO x = 1, 10 reduced = reduced * (a(x) + b(x)) END DO !$acc end data IF (abs(host_reduced - reduced) .gt. PRECISION) THEN errors = errors + 1 END IF END DO IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
./openacc-vv/atomic_x_eqv_expr_end.F90
#ifndef T1 !T1:construct-independent,atomic,V:2.0-2.7 LOGICAL FUNCTION test1() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x, y !Iterators REAL(8),DIMENSION(LOOPCOUNT, 10):: randoms LOGICAL,DIMENSION(LOOPCOUNT, 10):: a !Data LOGICAL,DIMENSION(LOOPCOUNT):: totals, totals_comparison INTEGER :: errors = 0 !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(randoms) DO x = 1, LOOPCOUNT DO y = 1, 10 IF (randoms(x, y) > .5) THEN a(x, y) = .TRUE. ELSE a(x, y) = .FALSE. END IF END DO END DO totals = .FALSE. totals_comparison = .FALSE. !$acc data copyin(a(1:LOOPCOUNT,1:10)) copy(totals(1:LOOPCOUNT)) !$acc parallel !$acc loop DO x = 1, LOOPCOUNT DO y = 1, 10 !$acc atomic totals(x) = totals(x) .EQV. a(x, y) !$acc end atomic END DO END DO !$acc end parallel !$acc end data DO x = 1, LOOPCOUNT DO y = 1, 10 totals_comparison(x) = totals_comparison(x) .EQV. a(x, y) END DO END DO DO x = 1, LOOPCOUNT IF (totals_comparison(x) .NEQV. totals(x)) THEN errors = errors + 1 WRITE(*, *) totals_comparison(x) END IF END DO IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
./openacc-npb/CG/CG/cg.c
//-------------------------------------------------------------------------// // // // This benchmark is a serial C version of the NPB CG code. This C // // version is developed by the Center for Manycore Programming at Seoul // // National University and derived from the serial Fortran versions in // // "NPB3.3-SER" developed by NAS. // // // // Permission to use, copy, distribute and modify this software for any // // purpose with or without fee is hereby granted. This software is // // provided "as is" without express or implied warranty. // // // // Information on NPB 3.3, including the technical report, the original // // specifications, source code, results and information on how to submit // // new results, is available at: // // // // http://www.nas.nasa.gov/Software/NPB/ // // // // Send comments or suggestions for this C version to cmp@aces.snu.ac.kr // // // // Center for Manycore Programming // // School of Computer Science and Engineering // // Seoul National University // // Seoul 151-744, Korea // // // // E-mail: cmp@aces.snu.ac.kr // // // //-------------------------------------------------------------------------// //-------------------------------------------------------------------------// // Authors: Sangmin Seo, Jungwon Kim, Jun Lee, Jeongho Nah, Gangwon Jo, // // and Jaejin Lee // //-------------------------------------------------------------------------// //-------------------------------------------------------------------------// //// // //// The OpenACC C version of the NAS CG code is developed by the // //// HPCTools Group of University of Houston and derived from the serial // //// C version developed by SNU and Fortran versions in "NPB3.3-SER" // //// developed by NAS. // //// // //// Permission to use, copy, distribute and modify this software for any // //// purpose with or without fee is hereby granted. This software is // //// provided "as is" without express or implied warranty. // //// // //// Send comments or suggestions for this OpenACC version to // //// hpctools@cs.uh.edu // //// //// Information on NPB 3.3, including the technical report, the original // //// specifications, source code, results and information on how to submit // //// new results, is available at: // //// // //// http://www.nas.nasa.gov/Software/NPB/ // //// // ////-------------------------------------------------------------------------// // ////-------------------------------------------------------------------------// //// Authors: Rengan Xu, Sunita Chandrasekaran, Barbara Chapman // ////-------------------------------------------------------------------------// //--------------------------------------------------------------------- // NPB CG OpenACC version //--------------------------------------------------------------------- #include <stdio.h> #include <stdlib.h> #include <math.h> #include "globals.h" #include "randdp.h" #include "timers.h" #include "print_results.h" #include <openacc.h> //--------------------------------------------------------------------- unsigned int nz = (NA*(NONZER+1)*(NONZER+1)); unsigned int naz = (NA*(NONZER+1)); unsigned int na = NA; /* common / main_int_mem / */ static int colidx[NZ]; static int rowstr[NA+1]; static int iv[NA]; static int arow[NA]; static int acol[NAZ]; /* common / main_flt_mem / */ static double aelt[NAZ]; static double a[NZ]; static double x[NA+2]; static double z[NA+2]; static double p[NA+2]; static double q[NA+2]; static double r[NA+2]; /* common / partit_size / */ static int naa; static int nzz; static int firstrow; static int lastrow; static int firstcol; static int lastcol; /* common /urando/ */ static double amult; static double tran; /* common /timers/ */ static logical timeron; //--------------------------------------------------------------------- //--------------------------------------------------------------------- static void conj_grad(int colidx[], int rowstr[], double x[], double z[], double a[], double p[], double q[], double r[], double *rnorm); static void makea(int n, int nz, double a[], int colidx[], int rowstr[], int firstrow, int lastrow, int firstcol, int lastcol, int arow[], int acol[][NONZER+1], double aelt[][NONZER+1], int iv[]); static void sparse(double a[], int colidx[], int rowstr[], int n, int nz, int nozer, int arow[], int acol[][NONZER+1], double aelt[][NONZER+1], int firstrow, int lastrow, int nzloc[], double rcond, double shift); static void sprnvc(int n, int nz, int nn1, double v[], int iv[]); static int icnvrt(double x, int ipwr2); static void vecset(int n, double v[], int iv[], int *nzv, int i, double val); static int conj_calls = 0; static int loop_iter = 0; //--------------------------------------------------------------------- int main(int argc, char *argv[]) { int i, j, k, it; int end; double zeta; double rnorm; double norm_temp1, norm_temp2; double t, mflops, tmax; char Class; int verified; double zeta_verify_value, epsilon, err; char *t_names[T_last]; acc_init(acc_device_default); for (i = 0; i < T_last; i++) { timer_clear(i); } FILE *fp; if ((fp = fopen("timer.flag", "r")) != NULL) { timeron = true; t_names[T_init] = "init"; t_names[T_bench] = "benchmk"; t_names[T_conj_grad] = "conjgd"; fclose(fp); } else { timeron = false; } timer_start(T_init); firstrow = 0; lastrow = NA-1; firstcol = 0; lastcol = NA-1; if (NA == 1400 && NONZER == 7 && NITER == 15 && SHIFT == 10) { Class = 'S'; zeta_verify_value = 8.5971775078648; } else if (NA == 7000 && NONZER == 8 && NITER == 15 && SHIFT == 12) { Class = 'W'; zeta_verify_value = 10.362595087124; } else if (NA == 14000 && NONZER == 11 && NITER == 15 && SHIFT == 20) { Class = 'A'; zeta_verify_value = 17.130235054029; } else if (NA == 75000 && NONZER == 13 && NITER == 75 && SHIFT == 60) { Class = 'B'; zeta_verify_value = 22.712745482631; } else if (NA == 150000 && NONZER == 15 && NITER == 75 && SHIFT == 110) { Class = 'C'; zeta_verify_value = 28.973605592845; } else if (NA == 1500000 && NONZER == 21 && NITER == 100 && SHIFT == 500) { Class = 'D'; zeta_verify_value = 52.514532105794; } else if (NA == 9000000 && NONZER == 26 && NITER == 100 && SHIFT == 1500) { Class = 'E'; zeta_verify_value = 77.522164599383; } else { Class = 'U'; } printf("\n\n NAS Parallel Benchmarks (NPB3.3-ACC-C) - CG Benchmark\n\n"); printf(" Size: %11d\n", NA); printf(" Iterations: %5d\n", NITER); printf("\n"); naa = NA; nzz = NZ; //--------------------------------------------------------------------- // Inialize random number generator //--------------------------------------------------------------------- tran = 314159265.0; amult = 1220703125.0; zeta = randlc(&tran, amult); //--------------------------------------------------------------------- // //--------------------------------------------------------------------- makea(naa, nzz, a, colidx, rowstr, firstrow, lastrow, firstcol, lastcol, arow, (int (*)[NONZER+1])(void*)acol, (double (*)[NONZER+1])(void*)aelt, iv); //--------------------------------------------------------------------- // Note: as a result of the above call to makea: // values of j used in indexing rowstr go from 0 --> lastrow-firstrow // values of colidx which are col indexes go from firstcol --> lastcol // So: // Shift the col index vals from actual (firstcol --> lastcol ) // to local, i.e., (0 --> lastcol-firstcol) //--------------------------------------------------------------------- for (j = 0; j < lastrow - firstrow + 1; j++) { for (k = rowstr[j]; k < rowstr[j+1]; k++) { colidx[k] = colidx[k] - firstcol; } } #pragma acc data copyin(colidx[0:nz],a[0:nz], \ rowstr[0:na+1]) \ create(x[0:na+2],z[0:na+2], \ p[0:na+2],q[0:na+2], \ r[0:na+2]) { //--------------------------------------------------------------------- // set starting vector to (1, 1, .... 1) //--------------------------------------------------------------------- int na_gangs = NA+1; #pragma acc kernels loop gang((na_gangs+127)/128) vector(128) for (i = 0; i < NA+1; i++) { x[i] = 1.0; } end = lastcol - firstcol + 1; #pragma acc kernels loop gang((end+127)/128) vector(128) for (j = 0; j < end; j++) { q[j] = 0.0; z[j] = 0.0; r[j] = 0.0; p[j] = 0.0; } zeta = 0.0; //--------------------------------------------------------------------- //----> // Do one iteration untimed to init all code and data page tables //----> (then reinit, start timing, to niter its) //--------------------------------------------------------------------- for (it = 1; it <= 1; it++) { //--------------------------------------------------------------------- // The call to the conjugate gradient routine: //--------------------------------------------------------------------- conj_grad(colidx, rowstr, x, z, a, p, q, r, &rnorm); //--------------------------------------------------------------------- // zeta = shift + 1/(x.z) // So, first: (x.z) // Also, find norm of z // So, first: (z.z) //--------------------------------------------------------------------- norm_temp1 = 0.0; norm_temp2 = 0.0; #pragma acc parallel loop num_gangs((end+127)/128) num_workers(4) \ vector_length(32) reduction(+:norm_temp2) for (j = 0; j < end; j++) { //norm_temp1 = norm_temp1 + x[j] * z[j]; norm_temp2 = norm_temp2 + z[j] * z[j]; } norm_temp2 = 1.0 / sqrt(norm_temp2); //--------------------------------------------------------------------- // Normalize z to obtain x //--------------------------------------------------------------------- #pragma acc kernels loop gang((end+127)/128) vector(128) for (j = 0; j < end; j++) { x[j] = norm_temp2 * z[j]; } } // end of do one iteration untimed //--------------------------------------------------------------------- // set starting vector to (1, 1, .... 1) //--------------------------------------------------------------------- na_gangs = NA+1; #pragma acc kernels loop gang((na_gangs+127)/128) vector(128) for (i = 0; i < NA+1; i++) { x[i] = 1.0; } zeta = 0.0; timer_stop(T_init); printf(" Initialization time = %15.3f seconds\n", timer_read(T_init)); timer_start(T_bench); //--------------------------------------------------------------------- //----> // Main Iteration for inverse power method //----> //--------------------------------------------------------------------- for (it = 1; it <= NITER; it++) { //--------------------------------------------------------------------- // The call to the conjugate gradient routine: //--------------------------------------------------------------------- conj_grad(colidx, rowstr, x, z, a, p, q, r, &rnorm); //--------------------------------------------------------------------- // zeta = shift + 1/(x.z) // So, first: (x.z) // Also, find norm of z // So, first: (z.z) //--------------------------------------------------------------------- norm_temp1 = 0.0; norm_temp2 = 0.0; #pragma acc parallel loop gang worker vector num_gangs((end+127)/128) num_workers(4) \ vector_length(32) reduction(+:norm_temp1,norm_temp2) for (j = 0; j < end; j++) { norm_temp1 = norm_temp1 + x[j]*z[j]; norm_temp2 = norm_temp2 + z[j]*z[j]; } norm_temp2 = 1.0 / sqrt(norm_temp2); zeta = SHIFT + 1.0 / norm_temp1; if (it == 1) printf("\n iteration ||r|| zeta\n"); printf(" %5d %20.14E%20.13f\n", it, rnorm, zeta); //--------------------------------------------------------------------- // Normalize z to obtain x //--------------------------------------------------------------------- #pragma acc kernels loop gang((end+127)/128) vector(128) for (j = 0; j < end; j++) { x[j] = norm_temp2 * z[j]; } } // end of main iter inv pow meth timer_stop(T_bench); }/*end acc data*/ //--------------------------------------------------------------------- // End of timed section //--------------------------------------------------------------------- t = timer_read(T_bench); printf(" Benchmark completed\n"); epsilon = 1.0e-10; if (Class != 'U') { err = fabs(zeta - zeta_verify_value) / zeta_verify_value; if (err <= epsilon) { verified = true; printf(" VERIFICATION SUCCESSFUL\n"); printf(" Zeta is %20.13E\n", zeta); printf(" Error is %20.13E\n", err); } else { verified = false; printf(" VERIFICATION FAILED\n"); printf(" Zeta %20.13E\n", zeta); printf(" The correct zeta is %20.13E\n", zeta_verify_value); } } else { verified = false; printf(" Problem size unknown\n"); printf(" NO VERIFICATION PERFORMED\n"); } if (t != 0.0) { mflops = (double)(2*NITER*NA) * (3.0+(double)(NONZER*(NONZER+1)) + 25.0*(5.0+(double)(NONZER*(NONZER+1))) + 3.0) / t / 1000000.0; } else { mflops = 0.0; } print_results("CG", Class, NA, 0, 0, NITER, t, mflops, " floating point", verified, NPBVERSION, COMPILETIME, CS1, CS2, CS3, CS4, CS5, CS6, CS7); //--------------------------------------------------------------------- // More timers //--------------------------------------------------------------------- if (timeron) { tmax = timer_read(T_bench); if (tmax == 0.0) tmax = 1.0; printf(" SECTION Time (secs)\n"); for (i = 0; i < T_last; i++) { t = timer_read(i); if (i == T_init) { printf(" %8s:%9.3f\n", t_names[i], t); } else { printf(" %8s:%9.3f (%6.2f%%)\n", t_names[i], t, t*100.0/tmax); if (i == T_conj_grad) { t = tmax - t; printf(" --> %8s:%9.3f (%6.2f%%)\n", "rest", t, t*100.0/tmax); } } } } acc_shutdown(acc_device_default); printf("conj calls=%d, loop iter = %d. \n", conj_calls, loop_iter); return 0; } //--------------------------------------------------------------------- // Floaging point arrays here are named as in NPB1 spec discussion of // CG algorithm //--------------------------------------------------------------------- static void conj_grad(int colidx[], int rowstr[], double x[], double z[], double a[], double p[], double q[], double r[], double *rnorm) { int j, k,tmp1,tmp2,tmp3; int end; int cgit, cgitmax = 25; double d, sum, rho, rho0, alpha, beta; double sum_array[NA+2]; conj_calls ++; rho = 0.0; unsigned int num_gangs = 0; #pragma acc data present(colidx[0:nz], \ rowstr[0:na+1], \ x[0:na+2],z[0:na+2], \ a[0:nz],p[0:na+2], \ q[0:na+2],r[0:na+2]) { //--------------------------------------------------------------------- // Initialize the CG algorithm: //--------------------------------------------------------------------- #pragma acc kernels loop gang((naa+127)/128) vector(128) independent for (j = 0; j < naa; j++) { q[j] = 0.0; z[j] = 0.0; r[j] = x[j]; p[j] = r[j]; } //--------------------------------------------------------------------- // rho = r.r // Now, obtain the norm of r: First, sum squares of r elements locally... //--------------------------------------------------------------------- //num_gangs = (lastcol - firstcol + 1)/128; #pragma acc parallel loop gang worker vector num_gangs((lastcol-firstcol+1+127)/128) \ num_workers(4) vector_length(32) reduction(+:rho) for (j = 0; j < lastcol - firstcol + 1; j++) { rho = rho + r[j]*r[j]; } //--------------------------------------------------------------------- //----> // The conj grad iteration loop //----> //--------------------------------------------------------------------- //#pragma acc kernels loop private(cgit,j,k) for (cgit = 1; cgit <= cgitmax; cgit++) { //#pragma acc update host(p[0:NA+2]) //--------------------------------------------------------------------- // q = A.p // The partition submatrix-vector multiply: use workspace w //--------------------------------------------------------------------- // // NOTE: this version of the multiply is actually (slightly: maybe %5) // faster on the sp2 on 16 nodes than is the unrolled-by-2 version // below. On the Cray t3d, the reverse is true, i.e., the // unrolled-by-two version is some 10% faster. // The unrolled-by-8 version below is significantly faster // on the Cray t3d - overall speed of code is 1.5 times faster. /* for (j = 0; j < lastrow - firstrow + 1; j++) { sum_array[j]=0.0; } for (j = 0; j < lastrow - firstrow + 1; j++) { tmp1=rowstr[j]; tmp2=rowstr[j+1]; for (k = tmp1; k < tmp2; k++) { tmp3=colidx[k]; sum_array[j] = sum_array[j] + a[k]*p[tmp3]; } q[j] = sum_array[j]; } */ loop_iter ++; //num_gangs = (lastrow - firstrow + 1)/128; end = lastrow - firstrow + 1; #pragma acc parallel num_gangs(end) num_workers(4) vector_length(32) { #pragma acc loop gang for (j = 0; j < end; j++) { tmp1 = rowstr[j]; tmp2 = rowstr[j+1]; sum = 0.0; #pragma acc loop worker vector reduction(+:sum) for (k = tmp1; k < tmp2; k++) { tmp3 = colidx[k]; sum = sum + a[k]*p[tmp3]; } q[j] = sum; } } //--------------------------------------------------------------------- // Obtain p.q //--------------------------------------------------------------------- d = 0.0; end = lastcol - firstcol + 1; #pragma acc parallel num_gangs((end+127)/128) num_workers(4) vector_length(32) { #pragma acc loop gang worker vector reduction(+:d) for (j = 0; j < end; j++) { d = d + p[j]*q[j]; } } //--------------------------------------------------------------------- // Obtain alpha = rho / (p.q) //--------------------------------------------------------------------- alpha = rho / d; //--------------------------------------------------------------------- // Save a temporary of rho //--------------------------------------------------------------------- rho0 = rho; //--------------------------------------------------------------------- // Obtain z = z + alpha*p // and r = r - alpha*q //--------------------------------------------------------------------- rho = 0.0; #pragma acc kernels loop gang((end+1023)/1024) vector(1024) independent for (j = 0; j < end; j++) { z[j] = z[j] + alpha*p[j]; r[j] = r[j] - alpha*q[j]; } //--------------------------------------------------------------------- // rho = r.r // Now, obtain the norm of r: First, sum squares of r elements locally... //--------------------------------------------------------------------- #pragma acc parallel num_gangs((end+127)/128) num_workers(4) vector_length(32) { #pragma acc loop gang worker vector reduction(+:rho) for (j = 0; j < end; j++) { rho = rho + r[j]*r[j]; } } //--------------------------------------------------------------------- // Obtain beta: //--------------------------------------------------------------------- beta = rho / rho0; //--------------------------------------------------------------------- // p = r + beta*p //--------------------------------------------------------------------- #pragma acc kernels loop gang((end+127)/128) vector(128) independent for (j = 0; j < end; j++) { p[j] = r[j] + beta*p[j]; } } // end of do cgit=1,cgitmax //--------------------------------------------------------------------- // Compute residual norm explicitly: ||r|| = ||x - A.z|| // First, form A.z // The partition submatrix-vector multiply //--------------------------------------------------------------------- end = lastrow - firstrow + 1; //num_gangs = end/128; #pragma acc parallel loop gang num_gangs(end) \ num_workers(4) vector_length(32) for (j = 0; j < end; j++) { tmp1=rowstr[j]; tmp2=rowstr[j+1]; d = 0.0; #pragma acc loop worker vector reduction(+:d) for (k = tmp1; k < tmp2; k++) { tmp3=colidx[k]; d = d + a[k]*z[tmp3]; } r[j] = d; } //--------------------------------------------------------------------- // At this point, r contains A.z //--------------------------------------------------------------------- sum = 0.0; //num_gangs = (lastcol-firstcol+1)/128; #pragma acc parallel loop gang worker vector \ num_gangs((lastcol-firstcol+1+127)/128) \ num_workers(4) vector_length(32) \ reduction(+:sum) for (j = 0; j < lastcol-firstcol+1; j++) { d = x[j] - r[j]; sum = sum + d*d; } }/*end acc data*/ *rnorm = sqrt(sum); } //--------------------------------------------------------------------- // generate the test problem for benchmark 6 // makea generates a sparse matrix with a // prescribed sparsity distribution // // parameter type usage // // input // // n i number of cols/rows of matrix // nz i nonzeros as declared array size // rcond r*8 condition number // shift r*8 main diagonal shift // // output // // a r*8 array for nonzeros // colidx i col indices // rowstr i row pointers // // workspace // // iv, arow, acol i // aelt r*8 //--------------------------------------------------------------------- static void makea(int n, int nz, double a[], int colidx[], int rowstr[], int firstrow, int lastrow, int firstcol, int lastcol, int arow[], int acol[][NONZER+1], double aelt[][NONZER+1], int iv[]) { int iouter, ivelt, nzv, nn1; int ivc[NONZER+1]; double vc[NONZER+1]; //--------------------------------------------------------------------- // nonzer is approximately (int(sqrt(nnza /n))); //--------------------------------------------------------------------- //--------------------------------------------------------------------- // nn1 is the smallest power of two not less than n //--------------------------------------------------------------------- nn1 = 1; do { nn1 = 2 * nn1; } while (nn1 < n); //--------------------------------------------------------------------- // Generate nonzero positions and save for the use in sparse. //--------------------------------------------------------------------- for (iouter = 0; iouter < n; iouter++) { nzv = NONZER; sprnvc(n, nzv, nn1, vc, ivc); vecset(n, vc, ivc, &nzv, iouter+1, 0.5); arow[iouter] = nzv; for (ivelt = 0; ivelt < nzv; ivelt++) { acol[iouter][ivelt] = ivc[ivelt] - 1; aelt[iouter][ivelt] = vc[ivelt]; } } //--------------------------------------------------------------------- // ... make the sparse matrix from list of elements with duplicates // (iv is used as workspace) //--------------------------------------------------------------------- sparse(a, colidx, rowstr, n, nz, NONZER, arow, acol, aelt, firstrow, lastrow, iv, RCOND, SHIFT); } //--------------------------------------------------------------------- // rows range from firstrow to lastrow // the rowstr pointers are defined for nrows = lastrow-firstrow+1 values //--------------------------------------------------------------------- static void sparse(double a[], int colidx[], int rowstr[], int n, int nz, int nozer, int arow[], int acol[][NONZER+1], double aelt[][NONZER+1], int firstrow, int lastrow, int nzloc[], double rcond, double shift) { int nrows; //--------------------------------------------------- // generate a sparse matrix from a list of // [col, row, element] tri //--------------------------------------------------- int i, j, j1, j2, nza, k, kk, nzrow, jcol; double size, scale, ratio, va; logical cont40; //--------------------------------------------------------------------- // how many rows of result //--------------------------------------------------------------------- nrows = lastrow - firstrow + 1; //--------------------------------------------------------------------- // ...count the number of triples in each row //--------------------------------------------------------------------- for (j = 0; j < nrows+1; j++) { rowstr[j] = 0; } for (i = 0; i < n; i++) { for (nza = 0; nza < arow[i]; nza++) { j = acol[i][nza] + 1; rowstr[j] = rowstr[j] + arow[i]; } } rowstr[0] = 0; for (j = 1; j < nrows+1; j++) { rowstr[j] = rowstr[j] + rowstr[j-1]; } nza = rowstr[nrows] - 1; //--------------------------------------------------------------------- // ... rowstr(j) now is the location of the first nonzero // of row j of a //--------------------------------------------------------------------- if (nza > nz) { printf("Space for matrix elements exceeded in sparse\n"); printf("nza, nzmax = %d, %d\n", nza, nz); exit(EXIT_FAILURE); } //--------------------------------------------------------------------- // ... preload data pages //--------------------------------------------------------------------- for (j = 0; j < nrows; j++) { for (k = rowstr[j]; k < rowstr[j+1]; k++) { a[k] = 0.0; colidx[k] = -1; } nzloc[j] = 0; } //--------------------------------------------------------------------- // ... generate actual values by summing duplicates //--------------------------------------------------------------------- size = 1.0; ratio = pow(rcond, (1.0 / (double)(n))); for (i = 0; i < n; i++) { for (nza = 0; nza < arow[i]; nza++) { j = acol[i][nza]; scale = size * aelt[i][nza]; for (nzrow = 0; nzrow < arow[i]; nzrow++) { jcol = acol[i][nzrow]; va = aelt[i][nzrow] * scale; //-------------------------------------------------------------------- // ... add the identity * rcond to the generated matrix to bound // the smallest eigenvalue from below by rcond //-------------------------------------------------------------------- if (jcol == j && j == i) { va = va + rcond - shift; } cont40 = false; for (k = rowstr[j]; k < rowstr[j+1]; k++) { if (colidx[k] > jcol) { //---------------------------------------------------------------- // ... insert colidx here orderly //---------------------------------------------------------------- for (kk = rowstr[j+1]-2; kk >= k; kk--) { if (colidx[kk] > -1) { a[kk+1] = a[kk]; colidx[kk+1] = colidx[kk]; } } colidx[k] = jcol; a[k] = 0.0; cont40 = true; break; } else if (colidx[k] == -1) { colidx[k] = jcol; cont40 = true; break; } else if (colidx[k] == jcol) { //-------------------------------------------------------------- // ... mark the duplicated entry //-------------------------------------------------------------- nzloc[j] = nzloc[j] + 1; cont40 = true; break; } } if (cont40 == false) { printf("internal error in sparse: i=%d\n", i); exit(EXIT_FAILURE); } a[k] = a[k] + va; } } size = size * ratio; } //--------------------------------------------------------------------- // ... remove empty entries and generate final results //--------------------------------------------------------------------- for (j = 1; j < nrows; j++) { nzloc[j] = nzloc[j] + nzloc[j-1]; } for (j = 0; j < nrows; j++) { if (j > 0) { j1 = rowstr[j] - nzloc[j-1]; } else { j1 = 0; } j2 = rowstr[j+1] - nzloc[j]; nza = rowstr[j]; for (k = j1; k < j2; k++) { a[k] = a[nza]; colidx[k] = colidx[nza]; nza = nza + 1; } } for (j = 1; j < nrows+1; j++) { rowstr[j] = rowstr[j] - nzloc[j-1]; } nza = rowstr[nrows] - 1; } //--------------------------------------------------------------------- // generate a sparse n-vector (v, iv) // having nzv nonzeros // // mark(i) is set to 1 if position i is nonzero. // mark is all zero on entry and is reset to all zero before exit // this corrects a performance bug found by John G. Lewis, caused by // reinitialization of mark on every one of the n calls to sprnvc //--------------------------------------------------------------------- static void sprnvc(int n, int nz, int nn1, double v[], int iv[]) { int nzv, ii, i; double vecelt, vecloc; nzv = 0; while (nzv < nz) { vecelt = randlc(&tran, amult); //--------------------------------------------------------------------- // generate an integer between 1 and n in a portable manner //--------------------------------------------------------------------- vecloc = randlc(&tran, amult); i = icnvrt(vecloc, nn1) + 1; if (i > n) continue; //--------------------------------------------------------------------- // was this integer generated already? //--------------------------------------------------------------------- logical was_gen = false; for (ii = 0; ii < nzv; ii++) { if (iv[ii] == i) { was_gen = true; break; } } if (was_gen) continue; v[nzv] = vecelt; iv[nzv] = i; nzv = nzv + 1; } } //--------------------------------------------------------------------- // scale a double precision number x in (0,1) by a power of 2 and chop it //--------------------------------------------------------------------- static int icnvrt(double x, int ipwr2) { return (int)(ipwr2 * x); } //--------------------------------------------------------------------- // set ith element of sparse vector (v, iv) with // nzv nonzeros to val //--------------------------------------------------------------------- static void vecset(int n, double v[], int iv[], int *nzv, int i, double val) { int k; logical set; set = false; for (k = 0; k < *nzv; k++) { if (iv[k] == i) { v[k] = val; set = true; } } if (set == false) { v[*nzv] = val; iv[*nzv] = i; *nzv = *nzv + 1; } }
./openacc-vv/wait_if.cpp
#include "acc_testsuite.h" #ifndef T1 //T1:parallel,wait,async,if,V:2.7-3.3 int test1(){ int err = 0; srand(SEED); real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; real_t *f = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0.0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = rand() / (real_t)(RAND_MAX / 10); f[x] = 0.0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n], e[0:n]) create(c[0:n], f[0:n]) { #pragma acc parallel async(1) { #pragma acc loop for (int x = 0; x < n; ++x){ c[x] = a[x] + b[x]; } } #pragma acc parallel async(2) { #pragma acc loop for (int x = 0; x < n; ++x){ f[x] = d[x] + e[x]; } } #pragma acc update host(c[0:n], f[0:n]) wait(1, 2) if(true) } for (int x = 0; x < n; ++x){ if (abs(c[x] - (a[x] + b[x])) > PRECISION){ err++; } if (abs(f[x] - (d[x] + e[x])) > PRECISION){ err++; } } return err; } #endif #ifndef T2 //T2:parallel,wait,async,if,V:2.7-3.3 int test2(){ int err = 0; srand(SEED); real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; real_t *f = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0.0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = rand() / (real_t)(RAND_MAX / 10); f[x] = 0.0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n], e[0:n]) create(c[0:n], f[0:n]) { #pragma acc parallel async(1) { #pragma acc loop for (int x = 0; x < n; ++x){ c[x] = a[x] + b[x]; } } #pragma acc parallel async(2) { #pragma acc loop for (int x = 0; x < n; ++x){ f[x] = d[x] + e[x]; } } #pragma acc update host(c[0:n], f[0:n]) wait(1) if(true) #pragma acc update host(c[0:n], f[0:n]) wait(2) if(true) } for (int x = 0; x < n; ++x){ if (abs(c[x] - (a[x] + b[x])) > PRECISION){ err++; } if (abs(f[x] - (d[x] + e[x])) > PRECISION){ err++; } } return err; } #endif #ifndef T3 //T3:parallel,wait,async,if,V:2.7-3.3 int test3(){ int err = 0; srand(time(NULL)); real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; real_t *f = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0.0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = rand() / (real_t)(RAND_MAX / 10); f[x] = 0.0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n], e[0:n]) create(c[0:n], f[0:n]) { #pragma acc parallel async(1) { #pragma acc loop for (int x = 0; x < n; ++x){ c[x] = a[x] + b[x]; } } #pragma acc parallel async(2) { #pragma acc loop for (int x = 0; x < n; ++x){ f[x] = d[x] + e[x]; } } #pragma acc update host(c[0:n], f[0:n]) wait(1, 2) if(false) } for (int x = 0; x < n; ++x){ if (c[x] > PRECISION){ err++; } if (f[x] > PRECISION){ err++; } } return err; } #endif #ifndef T4 //T4:parallel,wait,async,if,V:2.7-3.3 int test4(){ int err = 0; srand(time(NULL)); real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; real_t *f = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0.0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = rand() / (real_t)(RAND_MAX / 10); f[x] = 0.0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n], e[0:n]) create(c[0:n], f[0:n]) { #pragma acc parallel async(1) { #pragma acc loop for (int x = 0; x < n; ++x){ c[x] = a[x] + b[x]; } } #pragma acc parallel async(2) { #pragma acc loop for (int x = 0; x < n; ++x){ f[x] = d[x] + e[x]; } } #pragma acc update host(c[0:n], f[0:n]) wait(1) if(false) #pragma acc update host(c[0:n], f[0:n]) wait(2) if(false) } for (int x = 0; x < n; ++x){ if (c[x] > PRECISION){ err++; } if (f[x] > PRECISION){ err++; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed += test1(); } if (failed){ failcode += (1 << 0); } #endif #ifndef T2 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed += test2(); } if (failed){ failcode += (1 << 1); } #endif #ifndef T3 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed += test3(); } if (failed){ failcode += (1 << 2); } #endif #ifndef T4 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed += test4(); } if (failed){ failcode += (1 << 3); } #endif return failcode; }
./openacc-vv/atomic_expr_minus_x.cpp
#include "acc_testsuite.h" bool possible_result(real_t * remaining_combinations, int length, real_t current_value, real_t test_value){ if (length == 0){ if (fabs(current_value - test_value) > PRECISION){ return true; } else { return false; } } real_t * passed = new real_t[(length - 1)]; for (int x = 0; x < length; ++x){ for (int y = 0; y < x; ++y){ passed[y] = remaining_combinations[y]; } for (int y = x + 1; y < length; ++y){ passed[y - 1] = remaining_combinations[y]; } if (possible_result(passed, length - 1, remaining_combinations[x] - current_value, test_value)){ delete[] passed; return true; } } delete[] passed; return false; } #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = new real_t[n]; real_t *totals = new real_t[((n/10) + 1)]; int indexer = 0; real_t * passed = new real_t[10]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); } for (int x = 0; x < (n/10) + 1; ++x){ totals[x] = 0; } #pragma acc data copyin(a[0:n]) copy(totals[0:(n/10) + 1]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc atomic totals[x%((int) (n/10) + 1)] = a[x] - totals[x%((int) (n/10) + 1)]; } } } for (int x = 0; x < (n/10) + 1; ++x){ indexer = x; while (indexer < n){ passed[indexer/((int) (n/10) + 1)] = a[indexer]; indexer += (n/10) + 1; } if (!(possible_result(passed, 10, 0, totals[x]))){ err += 1; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/parallel_reduction.F90
#ifndef T1 !T1:parallel,reduction,V:1.0-2.7 LOGICAL FUNCTION test1() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x !Iterators REAL(8),DIMENSION(LOOPCOUNT):: a !Data REAL(8) :: results = 0 INTEGER :: errors = 0 !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(a) !$acc data copyin(a(1:LOOPCOUNT)) !$acc parallel reduction(+:results) !$acc loop DO x = 1, LOOPCOUNT results = results + a(x) END DO !$acc end parallel !$acc end data DO x = 1, LOOPCOUNT results = results - a(x) END DO IF (abs(results) .gt. PRECISION) THEN errors = errors + 1 END IF IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
./openacc-vv/atomic_plus_equals.c
#include "acc_testsuite.h" #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = (real_t *)malloc(n * sizeof(real_t)); real_t *b = (real_t *)malloc(n * sizeof(real_t)); real_t *totals = (real_t *)malloc((n/10 + 1) * sizeof(real_t)); real_t *totals_comparison = (real_t *)malloc((n/10 + 1) * sizeof(real_t)); for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); } for (int x = 0; x < n/10 + 1; ++x){ totals[x] = 0; totals_comparison[x] = 0; } #pragma acc data copyin(a[0:n], b[0:n]) copy(totals[0:n/10 + 1]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc atomic totals[x%(n/10 + 1)] += a[x] * b[x]; } } } for (int x = 0; x < n; ++x){ totals_comparison[x%(n/10 + 1)] += a[x] * b[x]; } for (int x = 0; x < n/10 + 1; ++x){ if (fabs(totals_comparison[x] - totals[x]) > (n/10 + 1) * PRECISION){ err += 1; break; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/atomic_update_min_expr_list_x_end.F90
#ifndef T1 !T1:construct-independent,atomic,V:2.0-2.7 LOGICAL FUNCTION test1() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x, y !Iterators REAL(8),DIMENSION(LOOPCOUNT):: a, b !Data REAL(8),DIMENSION(LOOPCOUNT/10 + 1):: totals, totals_comparison INTEGER :: errors = 0 !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(a) CALL RANDOM_NUMBER(b) totals = 1 totals_comparison = 1 !$acc data copyin(a(1:LOOPCOUNT), b(1:LOOPCOUNT)) copy(totals(1:(LOOPCOUNT/10 + 1))) !$acc parallel !$acc loop DO x = 1, LOOPCOUNT !$acc atomic update totals(MOD(x, LOOPCOUNT/10 + 1) + 1) = min(a(x), b(x), totals(MOD(x, LOOPCOUNT/10 + 1) + 1)) !$acc end atomic END DO !$acc end parallel !$acc end data DO x = 1, LOOPCOUNT totals_comparison(MOD(x, LOOPCOUNT/10 + 1) + 1) = min(totals_comparison(MOD(x, LOOPCOUNT/10 + 1) + 1), a(x), b(x)) END DO DO x = 1, LOOPCOUNT/10 + 1 IF (totals_comparison(x) .NE. totals(x)) THEN errors = errors + 1 WRITE(*, *) totals_comparison(x) END IF END DO IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
./openacc-vv/exit_data_copyout_reference_counts.F90
#ifndef T1 !T1:data,executable-data,devonly,construct-independent,V:2.0-2.7 LOGICAL FUNCTION test1() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x !Iterators REAL(8),DIMENSION(LOOPCOUNT):: a, b, c !Data INTEGER :: errors = 0 INTEGER,DIMENSION(1):: devtest devtest(1) = 1 !$acc enter data copyin(devtest(1:1)) !$acc parallel present(devtest(1:1)) devtest(1) = 0 !$acc end parallel !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(a) CALL RANDOM_NUMBER(b) c = 0 IF (devtest(1) .eq. 1) THEN !$acc enter data copyin(a(1:LOOPCOUNT), b(1:LOOPCOUNT), c(1:LOOPCOUNT)) !$acc data copyin(c(1:LOOPCOUNT)) !$acc parallel !$acc loop DO x = 1, LOOPCOUNT c(x) = c(x) + a(x) + b(x) END DO !$acc end parallel !$acc exit data delete(a(1:LOOPCOUNT), b(1:LOOPCOUNT)) copyout(c(1:LOOPCOUNT)) !$acc end data DO x = 1, LOOPCOUNT IF (abs(c(x)) .gt. PRECISION) THEN errors = errors + 1 EXIT END IF END DO END IF IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif #ifndef T2 !T2:data,executable-data,devonly,construct-independent,V:2.0-2.7 LOGICAL FUNCTION test2() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x !Iterators REAL(8),DIMENSION(LOOPCOUNT):: a, b, c !Data INTEGER :: errors = 0 INTEGER,DIMENSION(1):: devtest devtest(1) = 1 !$acc enter data copyin(devtest(1:1)) !$acc parallel present(devtest(1:1)) devtest(1) = 0 !$acc end parallel !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(a) CALL RANDOM_NUMBER(b) c = 0 !$acc enter data copyin(a(1:LOOPCOUNT), b(1:LOOPCOUNT), c(1:LOOPCOUNT)) !$acc data copyin(c(1:LOOPCOUNT)) !$acc parallel !$acc loop DO x = 1, LOOPCOUNT c(x) = c(x) + a(x) + b(x) END DO !$acc end parallel !$acc end data !$acc exit data copyout(c(1:LOOPCOUNT)) delete(a(1:LOOPCOUNT), b(1:LOOPCOUNT)) DO x = 1, LOOPCOUNT IF (abs(c(x) - (a(x) + b(x))) .gt. PRECISION) THEN errors = errors + 2 EXIT END IF END DO IF (errors .eq. 0) THEN test2 = .FALSE. ELSE test2 = .TRUE. END IF END #endif #ifndef T3 !T3:data,executable-data,devonly,construct-independent,V:2.0-2.7 LOGICAL FUNCTION test3() IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: x !Iterators REAL(8),DIMENSION(LOOPCOUNT):: a, b, c !Data INTEGER :: errors = 0 INTEGER,DIMENSION(1):: devtest devtest(1) = 1 !$acc enter data copyin(devtest(1:1)) !$acc parallel present(devtest(1:1)) devtest(1) = 0 !$acc end parallel !Initilization SEEDDIM(1) = 1 # ifdef SEED SEEDDIM(1) = SEED # endif CALL RANDOM_SEED(PUT=SEEDDIM) CALL RANDOM_NUMBER(a) CALL RANDOM_NUMBER(b) c = 0 !$acc enter data copyin(a(1:LOOPCOUNT), b(1:LOOPCOUNT), c(1:LOOPCOUNT)) !$acc enter data copyin(c(1:LOOPCOUNT)) !$acc parallel !$acc loop DO x = 1, LOOPCOUNT c(x) = c(x) + a(x) + b(x) END DO !$acc end parallel !$acc exit data delete(c(1:LOOPCOUNT)) !$acc exit data delete(a(1:LOOPCOUNT), b(1:LOOPCOUNT)) copyout(c(1:LOOPCOUNT)) DO x = 1, LOOPCOUNT IF (abs(c(x) - (a(x) + b(x))) .gt. PRECISION) THEN errors = errors + 4 EXIT END IF END DO IF (errors .eq. 0) THEN test3 = .FALSE. ELSE test3 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif #ifndef T2 LOGICAL :: test2 #endif #ifndef T3 LOGICAL :: test3 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif #ifndef T2 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test2() END DO IF (failed) THEN failcode = failcode + 2 ** 1 failed = .FALSE. END IF #endif #ifndef T3 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test3() END DO IF (failed) THEN failcode = failcode + 2 ** 2 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
./openacc-vv/atomic_update_preincrement.c
#include "acc_testsuite.h" #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = (real_t *)malloc(n * sizeof(real_t)); real_t *b = (real_t *)malloc(n * sizeof(real_t)); int *distribution = (int *)malloc(10 * sizeof(int)); int *distribution_comparison = (int *)malloc(10 * sizeof(int)); for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); } for (int x = 0; x < 10; ++x){ distribution[x] = 0; distribution_comparison[x] = 0; } #pragma acc data copyin(a[0:n], b[0:n]) copy(distribution[0:10]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc loop for (int y = 0; y < n; ++y){ #pragma acc atomic update ++distribution[(int) (a[x]*b[y]/10)]; } } } } for (int x = 0; x < n; ++x){ for (int y = 0; y < n; ++y){ distribution_comparison[(int) (a[x]*b[y]/10)]++; } } for (int x = 0; x < 10; ++x){ if (distribution_comparison[x] != distribution[x]){ err += 1; break; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./SPECaccel/benchspec/ACCEL/557.pcsp/src/error.c
//-------------------------------------------------------------------------// // // // This benchmark is a serial C version of the NPB SP code. This C // // version is developed by the Center for Manycore Programming at Seoul // // National University and derived from the serial Fortran versions in // // "NPB3.3-SER" developed by NAS. // // // // Permission to use, copy, distribute and modify this software for any // // purpose with or without fee is hereby granted. This software is // // provided "as is" without express or implied warranty. // // // // Information on NPB 3.3, including the technical report, the original // // specifications, source code, results and information on how to submit // // new results, is available at: // // // // http://www.nas.nasa.gov/Software/NPB/ // // // // Send comments or suggestions for this C version to cmp@aces.snu.ac.kr // // // // Center for Manycore Programming // // School of Computer Science and Engineering // // Seoul National University // // Seoul 151-744, Korea // // // // E-mail: cmp@aces.snu.ac.kr // // // //-------------------------------------------------------------------------// //-------------------------------------------------------------------------// // Authors: Sangmin Seo, Jungwon Kim, Jun Lee, Jeongho Nah, Gangwon Jo, // // and Jaejin Lee // //-------------------------------------------------------------------------// #include "header.h" #include <math.h> //--------------------------------------------------------------------- // this function computes the norm of the difference between the // computed solution and the exact solution //--------------------------------------------------------------------- void error_norm(double rms[5]) { int i, j, k, m, d; double xi, eta, zeta, u_exact[5], add; for (m = 0; m < 5; m++) { rms[m] = 0.0; } for (k = 0; k <= grid_points[2]-1; k++) { zeta = (double)k * dnzm1; for (j = 0; j <= grid_points[1]-1; j++) { eta = (double)j * dnym1; for (i = 0; i <= grid_points[0]-1; i++) { xi = (double)i * dnxm1; exact_solution(xi, eta, zeta, u_exact); for (m = 0; m < 5; m++) { add = u[m][k][j][i]-u_exact[m]; rms[m] = rms[m] + add*add; } } } } for (m = 0; m < 5; m++) { for (d = 0; d < 3; d++) { rms[m] = rms[m] / (double)(grid_points[d]-2); } rms[m] = sqrt(rms[m]); } } void rhs_norm(double rms[5]) { int i, j, k, d, m; double add; double rms0, rms1,rms2,rms3,rms4; rms0=0.0; rms1=0.0; rms2=0.0; rms3=0.0; rms4=0.0; #pragma omp target map(tofrom: rms0,rms1,rms2,rms3,rms4) //present(rhs) #ifdef SPEC_USE_INNER_SIMD #pragma omp teams distribute parallel for collapse(2) private(i) reduction(+:rms0,rms1,rms2,rms3,rms4) #else #pragma omp teams distribute parallel for simd collapse(3) reduction(+:rms0,rms1,rms2,rms3,rms4) private(add) #endif for (k = 1; k <= nz2; k++) { for (j = 1; j <= ny2; j++) { #ifdef SPEC_USE_INNER_SIMD #pragma omp simd reduction(+:rms0,rms1,rms2,rms3,rms4) private(add) #endif for (i = 1; i <= nx2; i++) { add = rhs[0][k][j][i]; rms0 = rms0 + add*add; add = rhs[1][k][j][i]; rms1 = rms1 + add*add; add = rhs[2][k][j][i]; rms2 = rms2 + add*add; add = rhs[3][k][j][i]; rms3 = rms3 + add*add; add = rhs[4][k][j][i]; rms4 = rms4 + add*add; } } } rms[0]=rms0; rms[1]=rms1; rms[2]=rms2; rms[3]=rms3; rms[4]=rms4; for (m = 0; m < 5; m++) { for (d = 0; d < 3; d++) { rms[m] = rms[m] / (double)(grid_points[d]-2); } rms[m] = sqrt(rms[m]); } }
./openacc-vv/serial_loop_reduction_bitxor_loop.c
#include "acc_testsuite.h" #ifndef T1 //T1:serial,loop,reduction,combined-constructs,V:2.6-3.2 int test1(){ int err = 0; srand(SEED); unsigned int * a = (unsigned int *)malloc(10 * n * sizeof(unsigned int)); unsigned int * b = (unsigned int *)malloc(10 * n * sizeof(unsigned int)); unsigned int * b_copy = (unsigned int *)malloc(10 * n * sizeof(unsigned int)); unsigned int * c = (unsigned int *)malloc(10 * sizeof(unsigned int)); unsigned int temp = 0; for (int x = 0; x < 10*n; ++x){ b[x] = (unsigned int) rand() / (real_t)(RAND_MAX / 1000); b_copy[x] = b[x]; a[x] = (unsigned int) rand() / (real_t)(RAND_MAX / 1000); } for (int x = 0; x < 10; ++x){ c[x] = 0; } #pragma acc data copyin(a[0:10*n]) copy(b[0:10*n], c[0:10]) { #pragma acc serial loop private(temp) for (int x = 0; x < 10; ++x){ temp = 0; #pragma acc loop worker reduction(^:temp) for (int y = 0; y < n; ++y){ temp = temp ^ a[x * n + y]; } c[x] = temp; #pragma acc loop worker for (int y = 0; y < n; ++y){ b[x * n + y] = b[x * n + y] + c[x]; } } } for (int x = 0; x < 10; ++x){ temp = 0; for (int y = 0; y < n; ++y){ temp = temp ^ a[x * n + y]; } if (temp != c[x]){ err += 1; } for (int y = 0; y < n; ++y){ if (b[x * n + y] != b_copy[x * n + y] + c[x]){ err += 1; } } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/serial_loop_seq.c
#include "acc_testsuite.h" #ifndef T1 //T1:serial,loop,combined-constructs,V:2.6-2.7 int test1(){ int err = 0; srand(SEED); real_t * a = (real_t *)malloc(n * sizeof(real_t)); real_t * b = (real_t *)malloc(n * sizeof(real_t)); real_t temp = 0.0; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = 0.0; } #pragma acc data copyin(a[0:n]) copy(b[0:n]) { #pragma acc serial loop seq for (int x = 1; x < n; ++x){ b[x] = b[x-1] + a[x]; } } for (int x = 1; x < n; ++x){ temp += a[x]; if (fabs(b[x] - temp) > PRECISION){ err = 1; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/serial_loop_reduction_bitor_general.cpp
#include "acc_testsuite.h" #ifndef T1 //T1:serial,loop,reduction,combined-constructs,V:2.6-2.7 int test1(){ int err = 0; srand(SEED); unsigned int * a = (unsigned int *)malloc(n * sizeof(unsigned int)); unsigned int b = 0; unsigned int host_b; real_t false_margin = pow(exp(1), log(.5)/n); unsigned int temp = 1; for (int x = 0; x < n; ++x){ for (int y = 0; y < 16; ++y){ if (rand() / (real_t) RAND_MAX > false_margin){ for (int z = 0; z < y; ++z){ temp *= 2; } a[x] += temp; temp = 1; } } } #pragma acc data copyin(a[0:n]) { #pragma acc serial loop reduction(|:b) for (int x = 0; x < n; ++x){ b = b | a[x]; } } host_b = a[0]; for (int x = 1; x < n; ++x){ host_b = host_b | a[x]; } if (b != host_b){ err = 1; } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./PhysiCell_GPU/sample_projects/virus_macrophage/config/PhysiCell_settings.xml
<?xml version="1.0" encoding="UTF-8"?> <!-- /* ############################################################################### # If you use PhysiCell in your project, please cite PhysiCell and the version # # number, such as below: # # # # We implemented and solved the model using PhysiCell (Version x.y.z) [1]. # # # # [1] A Ghaffarizadeh, R Heiland, SH Friedman, SM Mumenthaler, and P Macklin, # # PhysiCell: an Open Source Physics-Based Cell Simulator for Multicellu- # # lar Systems, PLoS Comput. Biol. 14(2): e1005991, 2018 # # DOI: 10.1371/journal.pcbi.1005991 # # # # See VERSION.txt or call get_PhysiCell_version() to get the current version # # x.y.z. Call display_citations() to get detailed information on all cite-# # able software used in your PhysiCell application. # # # # Because PhysiCell extensively uses BioFVM, we suggest you also cite BioFVM # # as below: # # # # We implemented and solved the model using PhysiCell (Version x.y.z) [1], # # with BioFVM [2] to solve the transport equations. # # # # [1] A Ghaffarizadeh, R Heiland, SH Friedman, SM Mumenthaler, and P Macklin, # # PhysiCell: an Open Source Physics-Based Cell Simulator for Multicellu- # # lar Systems, PLoS Comput. Biol. 14(2): e1005991, 2018 # # DOI: 10.1371/journal.pcbi.1005991 # # # # [2] A Ghaffarizadeh, SH Friedman, and P Macklin, BioFVM: an efficient para- # # llelized diffusive transport solver for 3-D biological simulations, # # Bioinformatics 32(8): 1256-8, 2016. DOI: 10.1093/bioinformatics/btv730 # # # ############################################################################### # # # BSD 3-Clause License (see https://opensource.org/licenses/BSD-3-Clause) # # # # Copyright (c) 2015-2019, Paul Macklin and the PhysiCell Project # # All rights reserved. # # # # 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 copyright holder nor the names of its # # contributors may be used to endorse or promote products derived from this # # software without specific prior written permission. # # # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 THE COPYRIGHT HOLDER OR 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. # # # ############################################################################### */ --> <!-- <user_details /> --> <PhysiCell_settings version="devel-version"> <domain> <x_min>-500</x_min> <x_max>500</x_max> <y_min>-500</y_min> <y_max>500</y_max> <z_min>-10</z_min> <z_max>10</z_max> <dx>20</dx> <dy>20</dy> <dz>20</dz> <use_2D>true</use_2D> </domain> <overall> <max_time units="min">1440</max_time> <!-- 5 days * 24 h * 60 min --> <time_units>min</time_units> <space_units>micron</space_units> </overall> <parallel> <omp_num_threads>4</omp_num_threads> </parallel> <save> <folder>output</folder> <!-- use . for root --> <full_data> <interval units="min">60</interval> <enable>true</enable> </full_data> <SVG> <interval units="min">10</interval> <enable>true</enable> </SVG> <legacy_data> <enable>false</enable> </legacy_data> </save> <microenvironment_setup> <variable name="virus" units="particles/micron^3" ID="0"> <physical_parameter_set> <diffusion_coefficient units="micron^2/min">1000</diffusion_coefficient> <decay_rate units="1/min">0</decay_rate> </physical_parameter_set> <initial_condition units="particles/micron^3">0</initial_condition> <Dirichlet_boundary_condition units="particles/micron^3" enabled="false">0</Dirichlet_boundary_condition> </variable> <options> <calculate_gradients>true</calculate_gradients> <track_internalized_substrates_in_each_agent>true</track_internalized_substrates_in_each_agent> <!-- not yet supported --> <initial_condition type="matlab" enabled="false"> <filename>./config/initial.mat</filename> </initial_condition> <!-- not yet supported --> <dirichlet_nodes type="matlab" enabled="false"> <filename>./config/dirichlet.mat</filename> </dirichlet_nodes> </options> </microenvironment_setup> <user_parameters> <random_seed type="int" units="dimensionless" hidden="true">0</random_seed> <!-- example parameters from the template --> <!-- virus properties --> <viral_replication_rate type="double" units="viral particles/min" description="rate at which infected cells create new viral particles">0.4167</viral_replication_rate> <min_virion_count type="double" units="none" description="minimum number of virions to start replication">1</min_virion_count> <burst_virion_count type="double" units="none" description="max virion load, where infected cells lyse to release viral particles">100</burst_virion_count> <viral_internalization_rate type="double" units="1/min" description="internalization rate for viral particles">10</viral_internalization_rate> <viral_diffusion_coefficient type="double" units="micron^2/min" description="diffusion coefficient for viral particles">1e3</viral_diffusion_coefficient> <number_of_infected_cells type="int" units="none" description="initial number of infected cells">1</number_of_infected_cells> <!-- virus diffusion coefficient ~ 1e3 micron^2/min : https://bionumbers.hms.harvard.edu/bionumber.aspx?id=105894 --> <!-- macrophage properties --> <number_of_macrophages type="int" units="none" description="initial number of macrophages">50</number_of_macrophages> <min_virion_detection_threshold type="double" units="none" description="minimal viral load for microphages to detect as infected">1</min_virion_detection_threshold> <virus_digestion_rate type="double" units="1/min" description="rate that macrophages digest internalized viral particles">0.01</virus_digestion_rate> <macrophage_persistence_time type="double" units="min">15</macrophage_persistence_time> <macrophage_migration_speed type="double" units="micron/min">1</macrophage_migration_speed> <macrophage_migration_bias type="double" units="dimensionless">0.1</macrophage_migration_bias> <macrophage_relative_adhesion type="double" units="dimensionless">0.05</macrophage_relative_adhesion> </user_parameters> </PhysiCell_settings>
./SPECaccel/benchspec/ACCEL/551.ppalm/data/test/input/ENVPAR
&envpar run_identifier = 'acc_small', host = 'unknown', write_binary = 'false', tasks_per_node = 1, maximum_cpu_time_allowed = 999999., revision = 'Rev: ', local_dvrserver_running = .FALSE. /
./openacc-vv/atomic_structured_divided_equals_assign.cpp
#include "acc_testsuite.h" bool is_possible(real_t* a, real_t* b, real_t* c, int length, real_t prev){ if (length == 0){ return true; } real_t *passed_a = new real_t[(length - 1)]; real_t *passed_b = new real_t[(length - 1)]; real_t *passed_c = new real_t[(length - 1)]; for (int x = 0; x < length; ++x){ if (fabs(c[x] - (prev / (a[x] + b[x]))) < PRECISION){ for (int y = 0; y < x; ++y){ passed_a[y] = a[y]; passed_b[y] = b[y]; passed_c[y] = c[y]; } for (int y = x + 1; y < length; ++y){ passed_a[y - 1] = a[y]; passed_b[y - 1] = b[y]; passed_c[y - 1] = c[y]; } if (is_possible(passed_a, passed_b, passed_c, length - 1, c[x])){ delete[] passed_a; delete[] passed_b; delete[] passed_c; return true; } } } delete[] passed_a; delete[] passed_b; delete[] passed_c; return false; } #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *totals = new real_t[(n/10 + 1)]; real_t *totals_comparison = new real_t[(n/10 + 1)]; real_t *temp_a = new real_t[10]; real_t *temp_b = new real_t[10]; real_t *temp_c = new real_t[10]; int temp_iterator; int ab_iterator; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0; } for (int x = 0; x < n/10 + 1; ++x){ totals[x] = 1; totals_comparison[x] = 1; } #pragma acc data copyin(a[0:n], b[0:n]) copy(totals[0:n/10 + 1]) copyout(c[0:n]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc atomic capture { totals[x/10] /= (a[x] + b[x]); c[x] = totals[x/10]; } } } } for (int x = 0; x < n; ++x){ totals_comparison[x/10] /= a[x] + b[x]; } for (int x = 0; x < 10; ++x){ if (fabs(totals_comparison[x] - totals[x]) > PRECISION){ err += 1; break; } } for (int x = 0; x < n; x = x + 10){ temp_iterator = 0; for (ab_iterator = x; ab_iterator < n && ab_iterator < x + 10; ab_iterator+= 1){ temp_a[temp_iterator] = a[ab_iterator]; temp_b[temp_iterator] = b[ab_iterator]; temp_c[temp_iterator] = c[ab_iterator]; temp_iterator++; } if (!(is_possible(temp_a, temp_b, temp_c, temp_iterator, 1))){ err++; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/atomic_capture_rshift_equals.cpp
#include "acc_testsuite.h" bool is_possible(unsigned int a, unsigned int* b, int length, unsigned int prev){ if (length == 0){ return true; } unsigned int passed_a = 0; unsigned int *passed_b = (unsigned int *)malloc((length - 1) * sizeof(unsigned int)); for (int x = 0; x < length; ++x){ if ((b[x] == prev>>1 && ((a>>x)%2)==1) || ((a>>x)%2==0 && b[x] == prev)){ for (int y = 0; y < x; ++y){ if ((a>>y)%2 == 1){ passed_a += 1<<y; } passed_b[y] = b[y]; } for (int y = x + 1; y < length; ++y){ if ((a>>y) % 2 == 1){ passed_a += 1<<(y - 1); } passed_b[y - 1] = b[y]; } if (is_possible(passed_a, passed_b, length - 1, b[x])){ delete[] passed_b; return true; } } } delete[] passed_b; return false; } #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); unsigned int *a = (unsigned int *)malloc(n * sizeof(int)); unsigned int *b = (unsigned int *)malloc(n * sizeof(int)); unsigned int *c = (unsigned int *)malloc(7 * n * sizeof(int)); unsigned int passed = 1<<8; for (int x = 0; x < n; ++x){ a[x] = 1<<8; for (int y = 0; y < 7; ++y){ if ((rand()/(real_t) (RAND_MAX)) > .5){ b[x] += 1<<y; } } } #pragma acc data copyin(b[0:n]) copy(a[0:n]) copyout(c[0:7*n]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc loop for (int y = 0; y < 7; ++y){ c[x * 7 + y] = a[x]; if ((b[x]>>y)%2 == 1){ #pragma acc atomic capture c[x * 7 + y] = a[x] >>= 1; } } } } } for (int x = 0; x < n; ++x){ for (int y = 0; y < 7; ++y){ if ((b[x]>>y)%2 == 1){ a[x] <<= 1; } } if (a[x] != 1<<8){ err += 1; } } for (int x = 0; x < n; ++x){ if (!is_possible(b[x], &(c[x * 7]), 7, passed)){ err++; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./PhysiCell_GPU/unit_tests/substrate_internalization/config/PhysiCell_settings.xml
<?xml version="1.0" encoding="UTF-8"?> <!-- /* ############################################################################### # If you use PhysiCell in your project, please cite PhysiCell and the version # # number, such as below: # # # # We implemented and solved the model using PhysiCell (Version x.y.z) [1]. # # # # [1] A Ghaffarizadeh, R Heiland, SH Friedman, SM Mumenthaler, and P Macklin, # # PhysiCell: an Open Source Physics-Based Cell Simulator for Multicellu- # # lar Systems, PLoS Comput. Biol. 14(2): e1005991, 2018 # # DOI: 10.1371/journal.pcbi.1005991 # # # # See VERSION.txt or call get_PhysiCell_version() to get the current version # # x.y.z. Call display_citations() to get detailed information on all cite-# # able software used in your PhysiCell application. # # # # Because PhysiCell extensively uses BioFVM, we suggest you also cite BioFVM # # as below: # # # # We implemented and solved the model using PhysiCell (Version x.y.z) [1], # # with BioFVM [2] to solve the transport equations. # # # # [1] A Ghaffarizadeh, R Heiland, SH Friedman, SM Mumenthaler, and P Macklin, # # PhysiCell: an Open Source Physics-Based Cell Simulator for Multicellu- # # lar Systems, PLoS Comput. Biol. 14(2): e1005991, 2018 # # DOI: 10.1371/journal.pcbi.1005991 # # # # [2] A Ghaffarizadeh, SH Friedman, and P Macklin, BioFVM: an efficient para- # # llelized diffusive transport solver for 3-D biological simulations, # # Bioinformatics 32(8): 1256-8, 2016. DOI: 10.1093/bioinformatics/btv730 # # # ############################################################################### # # # BSD 3-Clause License (see https://opensource.org/licenses/BSD-3-Clause) # # # # Copyright (c) 2015-2018, Paul Macklin and the PhysiCell Project # # All rights reserved. # # # # 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 copyright holder nor the names of its # # contributors may be used to endorse or promote products derived from this # # software without specific prior written permission. # # # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 THE COPYRIGHT HOLDER OR 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. # # # ############################################################################### */ --> <!-- <user_details /> --> <PhysiCell_settings version="devel-version"> <domain> <x_min>-500</x_min> <x_max>500</x_max> <y_min>-500</y_min> <y_max>500</y_max> <z_min>-10</z_min> <z_max>10</z_max> <dx>20</dx> <dy>20</dy> <dz>20</dz> <use_2D>true</use_2D> </domain> <overall> <max_time units="min">1440</max_time> <!-- 5 days * 24 h * 60 min --> <time_units>min</time_units> <space_units>micron</space_units> </overall> <parallel> <omp_num_threads>4</omp_num_threads> </parallel> <save> <folder>output</folder> <!-- use . for root --> <full_data> <interval units="min">60</interval> <enable>true</enable> </full_data> <SVG> <interval units="min">6</interval> <enable>true</enable> </SVG> <legacy_data> <enable>false</enable> </legacy_data> </save> <user_parameters> <random_seed type="int" units="dimensionless">0</random_seed> <!-- example parameters from the template --> <!-- motile cell type parameters --> <motile_cell_persistence_time type="double" units="min">15</motile_cell_persistence_time> <motile_cell_migration_speed type="double" units="micron/min">0.5</motile_cell_migration_speed> <motile_cell_relative_adhesion type="double" units="dimensionless">0.05</motile_cell_relative_adhesion> <motile_cell_apoptosis_rate type="double" units="1/min">0.0</motile_cell_apoptosis_rate> <motile_cell_relative_cycle_entry_rate type="double" units="dimensionless">0.1</motile_cell_relative_cycle_entry_rate> </user_parameters> </PhysiCell_settings>
./SPEChpc/benchspec/HPC/628.pot3d_s/src/hdf5/H5HFiblock.c
/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Copyright by The HDF Group. * * Copyright by the Board of Trustees of the University of Illinois. * * All rights reserved. * * * * This file is part of HDF5. The full HDF5 copyright notice, including * * terms governing use, modification, and redistribution, is contained in * * the COPYING file, which can be found at the root of the source code * * distribution tree, or in https://support.hdfgroup.org/ftp/HDF5/releases. * * If you do not have access to either file, you may request a copy from * * help@hdfgroup.org. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */ /*------------------------------------------------------------------------- * * Created: H5HFiblock.c * Apr 10 2006 * Quincey Koziol <koziol@ncsa.uiuc.edu> * * Purpose: Indirect block routines for fractal heaps. * *------------------------------------------------------------------------- */ /****************/ /* Module Setup */ /****************/ #include "H5HFmodule.h" /* This source code file is part of the H5HF module */ /***********/ /* Headers */ /***********/ #include "H5private.h" /* Generic Functions */ #include "H5Eprivate.h" /* Error handling */ #include "H5Fprivate.h" /* File access */ #include "H5HFpkg.h" /* Fractal heaps */ #include "H5MFprivate.h" /* File memory management */ #include "H5VMprivate.h" /* Vectors and arrays */ /****************/ /* Local Macros */ /****************/ /******************/ /* Local Typedefs */ /******************/ /********************/ /* Package Typedefs */ /********************/ /********************/ /* Local Prototypes */ /********************/ static herr_t H5HF__iblock_pin(H5HF_indirect_t *iblock); static herr_t H5HF__iblock_unpin(H5HF_indirect_t *iblock); static herr_t H5HF__man_iblock_root_halve(H5HF_indirect_t *root_iblock); static herr_t H5HF__man_iblock_root_revert(H5HF_indirect_t *root_iblock); /*********************/ /* Package Variables */ /*********************/ /* Declare a free list to manage the H5HF_indirect_t struct */ H5FL_DEFINE(H5HF_indirect_t); /* Declare a free list to manage the H5HF_indirect_ent_t sequence information */ H5FL_SEQ_DEFINE(H5HF_indirect_ent_t); /* Declare a free list to manage the H5HF_indirect_filt_ent_t sequence information */ H5FL_SEQ_DEFINE(H5HF_indirect_filt_ent_t); /* Declare a free list to manage the H5HF_indirect_t * sequence information */ H5FL_SEQ_DEFINE(H5HF_indirect_ptr_t); /*****************************/ /* Library Private Variables */ /*****************************/ /*******************/ /* Local Variables */ /*******************/ /*------------------------------------------------------------------------- * Function: H5HF__iblock_pin * * Purpose: Pin an indirect block in memory * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@hdfgroup.org * Aug 17 2006 * *------------------------------------------------------------------------- */ static herr_t H5HF__iblock_pin(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_STATIC /* Sanity checks */ HDassert(iblock); /* Mark block as un-evictable */ if(H5AC_pin_protected_entry(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTPIN, FAIL, "unable to pin fractal heap indirect block") /* If this indirect block has a parent, update it's child iblock pointer */ if(iblock->parent) { H5HF_indirect_t *par_iblock = iblock->parent; /* Parent indirect block */ unsigned indir_idx; /* Index in parent's child iblock pointer array */ /* Sanity check */ HDassert(par_iblock->child_iblocks); HDassert(iblock->par_entry >= (iblock->hdr->man_dtable.max_direct_rows * iblock->hdr->man_dtable.cparam.width)); /* Compute index in parent's child iblock pointer array */ indir_idx = iblock->par_entry - (iblock->hdr->man_dtable.max_direct_rows * iblock->hdr->man_dtable.cparam.width); /* Set pointer to pinned indirect block in parent */ HDassert(par_iblock->child_iblocks[indir_idx] == NULL); par_iblock->child_iblocks[indir_idx] = iblock; } /* end if */ else { /* Check for pinning the root indirect block */ if(iblock->block_off == 0) { /* Sanity check - shouldn't be recursively pinning root indirect block */ HDassert(0 == (iblock->hdr->root_iblock_flags & H5HF_ROOT_IBLOCK_PINNED)); /* Check if we should set the root iblock pointer */ if(0 == iblock->hdr->root_iblock_flags) { HDassert(NULL == iblock->hdr->root_iblock); iblock->hdr->root_iblock = iblock; } /* end if */ /* Indicate that the root indirect block is pinned */ iblock->hdr->root_iblock_flags |= H5HF_ROOT_IBLOCK_PINNED; } /* end if */ } /* end if */ done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__iblock_pin() */ /*------------------------------------------------------------------------- * Function: H5HF__iblock_unpin * * Purpose: Unpin an indirect block in the metadata cache * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@hdfgroup.org * Aug 17 2006 * *------------------------------------------------------------------------- */ static herr_t H5HF__iblock_unpin(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_STATIC /* Sanity check */ HDassert(iblock); /* Mark block as evictable again */ if(H5AC_unpin_entry(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPIN, FAIL, "unable to unpin fractal heap indirect block") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__iblock_unpin() */ /*------------------------------------------------------------------------- * Function: H5HF_iblock_incr * * Purpose: Increment reference count on shared indirect block * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Mar 27 2006 * *------------------------------------------------------------------------- */ herr_t H5HF_iblock_incr(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_NOAPI_NOINIT /* Sanity checks */ HDassert(iblock); HDassert(iblock->block_off == 0 || iblock->parent); /* Mark block as un-evictable when a child block is depending on it */ if(iblock->rc == 0) if(H5HF__iblock_pin(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTPIN, FAIL, "unable to pin fractal heap indirect block") /* Increment reference count on shared indirect block */ iblock->rc++; done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF_iblock_incr() */ /*------------------------------------------------------------------------- * Function: H5HF__iblock_decr * * Purpose: Decrement reference count on shared indirect block * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Mar 27 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__iblock_decr(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* Sanity check */ HDassert(iblock); /* Decrement reference count on shared indirect block */ iblock->rc--; /* Check for last reference to block */ if(iblock->rc == 0) { /* If this indirect block has a parent, reset it's child iblock pointer */ if(iblock->parent) { H5HF_indirect_t *par_iblock = iblock->parent; /* Parent indirect block */ unsigned indir_idx; /* Index in parent's child iblock pointer array */ /* Sanity check */ HDassert(par_iblock->child_iblocks); HDassert(iblock->par_entry >= (iblock->hdr->man_dtable.max_direct_rows * iblock->hdr->man_dtable.cparam.width)); /* Compute index in parent's child iblock pointer array */ indir_idx = iblock->par_entry - (iblock->hdr->man_dtable.max_direct_rows * iblock->hdr->man_dtable.cparam.width); /* Reset pointer to pinned child indirect block in parent */ HDassert(par_iblock->child_iblocks[indir_idx]); par_iblock->child_iblocks[indir_idx] = NULL; } /* end if */ else { /* Check for root indirect block */ if(iblock->block_off == 0) { /* Sanity check - shouldn't be recursively unpinning root indirect block */ HDassert(iblock->hdr->root_iblock_flags & H5HF_ROOT_IBLOCK_PINNED); /* Check if we should reset the root iblock pointer */ if(H5HF_ROOT_IBLOCK_PINNED == iblock->hdr->root_iblock_flags) { HDassert(NULL != iblock->hdr->root_iblock); iblock->hdr->root_iblock = NULL; } /* end if */ /* Indicate that the root indirect block is unpinned */ iblock->hdr->root_iblock_flags &= (unsigned)(~(H5HF_ROOT_IBLOCK_PINNED)); } /* end if */ } /* end else */ /* Check if the block is still in the cache */ if(!iblock->removed_from_cache) { /* Unpin the indirect block, making it evictable again */ if(H5HF__iblock_unpin(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPIN, FAIL, "unable to unpin fractal heap indirect block") } /* end if */ else { /* Destroy the indirect block */ if(H5HF_man_iblock_dest(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to destroy fractal heap indirect block") } /* end else */ } /* end if */ done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__iblock_decr() */ /*------------------------------------------------------------------------- * Function: H5HF_iblock_dirty * * Purpose: Mark indirect block as dirty * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Mar 21 2006 * *------------------------------------------------------------------------- */ herr_t H5HF_iblock_dirty(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_NOAPI_NOINIT /* Sanity check */ HDassert(iblock); /* Mark indirect block as dirty in cache */ if(H5AC_mark_entry_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTMARKDIRTY, FAIL, "unable to mark fractal heap indirect block as dirty") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF_iblock_dirty() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_root_create * * Purpose: Create root indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * May 2 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_root_create(H5HF_hdr_t *hdr, size_t min_dblock_size) { H5HF_indirect_t *iblock; /* Pointer to indirect block */ haddr_t iblock_addr; /* Indirect block's address */ hsize_t acc_dblock_free; /* Accumulated free space in direct blocks */ hbool_t have_direct_block; /* Flag to indicate a direct block already exists */ hbool_t did_protect; /* Whether we protected the indirect block or not */ unsigned nrows; /* Number of rows for root indirect block */ unsigned u; /* Local index variable */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* Check for allocating entire root indirect block initially */ if(hdr->man_dtable.cparam.start_root_rows == 0) nrows = hdr->man_dtable.max_root_rows; else { unsigned rows_needed; /* Number of rows needed to get to direct block size */ unsigned block_row_off; /* Row offset from larger block sizes */ nrows = hdr->man_dtable.cparam.start_root_rows; block_row_off = H5VM_log2_of2((uint32_t)min_dblock_size) - H5VM_log2_of2((uint32_t)hdr->man_dtable.cparam.start_block_size); if(block_row_off > 0) block_row_off++; /* Account for the pair of initial rows of the initial block size */ rows_needed = 1 + block_row_off; if(nrows < rows_needed) nrows = rows_needed; } /* end else */ /* Allocate root indirect block */ if(H5HF__man_iblock_create(hdr, NULL, 0, nrows, hdr->man_dtable.max_root_rows, &iblock_addr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTALLOC, FAIL, "can't allocate fractal heap indirect block") /* Move current direct block (used as root) into new indirect block */ /* Lock new indirect block */ if(NULL == (iblock = H5HF__man_iblock_protect(hdr, iblock_addr, nrows, NULL, 0, FALSE, H5AC__NO_FLAGS_SET, &did_protect))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, FAIL, "unable to protect fractal heap indirect block") /* Check if there's already a direct block as root) */ have_direct_block = H5F_addr_defined(hdr->man_dtable.table_addr); if(have_direct_block) { H5HF_direct_t *dblock; /* Pointer to direct block to query */ /* Lock first (root) direct block */ if(NULL == (dblock = H5HF__man_dblock_protect(hdr, hdr->man_dtable.table_addr, hdr->man_dtable.cparam.start_block_size, NULL, 0, H5AC__NO_FLAGS_SET))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, FAIL, "unable to protect fractal heap direct block") /* Attach direct block to new root indirect block */ dblock->parent = iblock; dblock->par_entry = 0; /* Destroy flush dependency between direct block and header */ if(H5AC_destroy_flush_dependency(dblock->fd_parent, dblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNDEPEND, FAIL, "unable to destroy flush dependency") dblock->fd_parent = NULL; /* Create flush dependency between direct block and new root indirect block */ if(H5AC_create_flush_dependency(iblock, dblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEPEND, FAIL, "unable to create flush dependency") dblock->fd_parent = iblock; if(H5HF_man_iblock_attach(iblock, 0, hdr->man_dtable.table_addr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTATTACH, FAIL, "can't attach root direct block to parent indirect block") /* Check for I/O filters on this heap */ if(hdr->filter_len > 0) { /* Set the pipeline filter information from the header */ iblock->filt_ents[0].size = hdr->pline_root_direct_size; iblock->filt_ents[0].filter_mask = hdr->pline_root_direct_filter_mask; /* Reset the header's pipeline information */ hdr->pline_root_direct_size = 0; hdr->pline_root_direct_filter_mask = 0; } /* end if */ /* Scan free space sections to set any 'parent' pointers to new indirect block */ if(H5HF__space_create_root(hdr, iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSET, FAIL, "can't set free space section info to new root indirect block") /* Unlock first (previously the root) direct block */ if(H5AC_unprotect(hdr->f, H5AC_FHEAP_DBLOCK, hdr->man_dtable.table_addr, dblock, H5AC__NO_FLAGS_SET) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap direct block") dblock = NULL; } /* end if */ /* Start iterator at correct location */ if(H5HF_hdr_start_iter(hdr, iblock, (hsize_t)(have_direct_block ? hdr->man_dtable.cparam.start_block_size : 0), have_direct_block) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTINIT, FAIL, "can't initialize block iterator") /* Check for skipping over direct blocks, in order to get to large enough block */ if(min_dblock_size > hdr->man_dtable.cparam.start_block_size) /* Add skipped blocks to heap's free space */ if(H5HF__hdr_skip_blocks(hdr, iblock, have_direct_block, ((nrows - 1) * hdr->man_dtable.cparam.width) - have_direct_block) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't add skipped blocks to heap's free space") /* Mark indirect block as modified */ if(H5HF_iblock_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDIRTY, FAIL, "can't mark indirect block as dirty") /* Unprotect root indirect block (it's pinned by the iterator though) */ if(H5HF__man_iblock_unprotect(iblock, H5AC__DIRTIED_FLAG, did_protect) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap indirect block") iblock = NULL; /* Point heap header at new indirect block */ hdr->man_dtable.curr_root_rows = nrows; hdr->man_dtable.table_addr = iblock_addr; /* Compute free space in direct blocks referenced from entries in root indirect block */ acc_dblock_free = 0; for(u = 0; u < nrows; u++) acc_dblock_free += hdr->man_dtable.row_tot_dblock_free[u] * hdr->man_dtable.cparam.width; /* Account for potential initial direct block */ if(have_direct_block) acc_dblock_free -= hdr->man_dtable.row_tot_dblock_free[0]; /* Extend heap to cover new root indirect block */ if(H5HF_hdr_adjust_heap(hdr, hdr->man_dtable.row_block_off[nrows], (hssize_t)acc_dblock_free) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTEXTEND, FAIL, "can't increase space to cover root direct block") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_root_create() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_root_double * * Purpose: Double size of root indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Apr 17 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_root_double(H5HF_hdr_t *hdr, size_t min_dblock_size) { H5HF_indirect_t *iblock; /* Pointer to root indirect block */ haddr_t new_addr; /* New address of indirect block */ hsize_t acc_dblock_free; /* Accumulated free space in direct blocks */ hsize_t next_size; /* The previous value of the "next size" for the new block iterator */ hsize_t old_iblock_size; /* Old size of indirect block */ unsigned next_row; /* The next row to allocate block in */ unsigned next_entry; /* The previous value of the "next entry" for the new block iterator */ unsigned new_next_entry = 0;/* The new value of the "next entry" for the new block iterator */ unsigned min_nrows = 0; /* Min. # of direct rows */ unsigned old_nrows; /* Old # of rows */ unsigned new_nrows; /* New # of rows */ hbool_t skip_direct_rows = FALSE; /* Whether we are skipping direct rows */ size_t u; /* Local index variable */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* Get "new block" iterator information */ if(H5HF_man_iter_curr(&hdr->next_block, &next_row, NULL, &next_entry, &iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTGET, FAIL, "unable to retrieve current block iterator location") next_size = hdr->man_dtable.row_block_size[next_row]; /* Make certain the iterator is at the root indirect block */ HDassert(iblock->parent == NULL); HDassert(iblock->block_off == 0); /* Keep this for later */ old_nrows = iblock->nrows; /* Check for skipping over direct block rows */ if(iblock->nrows < hdr->man_dtable.max_direct_rows && min_dblock_size > next_size) { /* Sanity check */ HDassert(min_dblock_size > hdr->man_dtable.cparam.start_block_size); /* Set flag */ skip_direct_rows = TRUE; /* Make certain we allocate at least the required row for the block requested */ min_nrows = 1 + H5HF_dtable_size_to_row(&hdr->man_dtable, min_dblock_size); /* Set the information for the next block, of the appropriate size */ new_next_entry = (min_nrows - 1) * hdr->man_dtable.cparam.width; } /* end if */ /* Compute new # of rows in indirect block */ new_nrows = MAX(min_nrows, MIN(2 * iblock->nrows, iblock->max_rows)); /* Check if the indirect block is NOT currently allocated in temp. file space */ /* (temp. file space does not need to be freed) */ if(!H5F_IS_TMP_ADDR(hdr->f, iblock->addr)) /* Free previous indirect block disk space */ if(H5MF_xfree(hdr->f, H5FD_MEM_FHEAP_IBLOCK, iblock->addr, (hsize_t)iblock->size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to free fractal heap indirect block file space") /* Compute size of buffer needed for new indirect block */ iblock->nrows = new_nrows; old_iblock_size = iblock->size; iblock->size = H5HF_MAN_INDIRECT_SIZE(hdr, iblock->nrows); /* Allocate [temporary] space for the new indirect block on disk */ if(H5F_USE_TMP_SPACE(hdr->f)) { if(HADDR_UNDEF == (new_addr = H5MF_alloc_tmp(hdr->f, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end if */ else { if(HADDR_UNDEF == (new_addr = H5MF_alloc(hdr->f, H5FD_MEM_FHEAP_IBLOCK, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end else */ /* Resize pinned indirect block in the cache, if its changed size */ if(old_iblock_size != iblock->size) { if(H5AC_resize_entry(iblock, (size_t)iblock->size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTRESIZE, FAIL, "unable to resize fractal heap indirect block") } /* end if */ /* Move object in cache, if it actually was relocated */ if(H5F_addr_ne(iblock->addr, new_addr)) { if(H5AC_move_entry(hdr->f, H5AC_FHEAP_IBLOCK, iblock->addr, new_addr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTMOVE, FAIL, "unable to move fractal heap root indirect block") iblock->addr = new_addr; } /* end if */ /* Re-allocate child block entry array */ if(NULL == (iblock->ents = H5FL_SEQ_REALLOC(H5HF_indirect_ent_t, iblock->ents, (size_t)(iblock->nrows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "memory allocation failed for direct entries") /* Check for skipping over rows and add free section for skipped rows */ if(skip_direct_rows) /* Add skipped blocks to heap's free space */ if(H5HF__hdr_skip_blocks(hdr, iblock, next_entry, (new_next_entry - next_entry)) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't add skipped blocks to heap's free space") /* Initialize new direct block entries in rows added */ acc_dblock_free = 0; for(u = (old_nrows * hdr->man_dtable.cparam.width); u < (iblock->nrows * hdr->man_dtable.cparam.width); u++) { unsigned row = (unsigned)(u / hdr->man_dtable.cparam.width); /* Row for current entry */ iblock->ents[u].addr = HADDR_UNDEF; acc_dblock_free += hdr->man_dtable.row_tot_dblock_free[row]; } /* end for */ /* Check for needing to re-allocate filtered entry array */ if(hdr->filter_len > 0 && old_nrows < hdr->man_dtable.max_direct_rows) { unsigned dir_rows; /* Number of direct rows in this indirect block */ /* Compute the number of direct rows for this indirect block */ dir_rows = MIN(iblock->nrows, hdr->man_dtable.max_direct_rows); HDassert(dir_rows > old_nrows); /* Re-allocate filtered direct block entry array */ if(NULL == (iblock->filt_ents = H5FL_SEQ_REALLOC(H5HF_indirect_filt_ent_t, iblock->filt_ents, (size_t)(dir_rows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "memory allocation failed for filtered direct entries") /* Initialize new entries allocated */ for(u = (old_nrows * hdr->man_dtable.cparam.width); u < (dir_rows * hdr->man_dtable.cparam.width); u++) { iblock->filt_ents[u].size = 0; iblock->filt_ents[u].filter_mask = 0; } /* end for */ } /* end if */ /* Check for needing to re-allocate child iblock pointer array */ if(iblock->nrows > hdr->man_dtable.max_direct_rows) { unsigned indir_rows; /* Number of indirect rows in this indirect block */ unsigned old_indir_rows; /* Previous number of indirect rows in this indirect block */ /* Compute the number of direct rows for this indirect block */ indir_rows = iblock->nrows - hdr->man_dtable.max_direct_rows; /* Re-allocate child indirect block array */ if(NULL == (iblock->child_iblocks = H5FL_SEQ_REALLOC(H5HF_indirect_ptr_t, iblock->child_iblocks, (size_t)(indir_rows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "memory allocation failed for filtered direct entries") /* Compute the previous # of indirect rows in this block */ if(old_nrows < hdr->man_dtable.max_direct_rows) old_indir_rows = 0; else old_indir_rows = old_nrows - hdr->man_dtable.max_direct_rows; /* Initialize new entries allocated */ for(u = (old_indir_rows * hdr->man_dtable.cparam.width); u < (indir_rows * hdr->man_dtable.cparam.width); u++) iblock->child_iblocks[u] = NULL; } /* end if */ /* Mark indirect block as dirty */ if(H5HF_iblock_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDIRTY, FAIL, "can't mark indirect block as dirty") /* Update other shared header info */ hdr->man_dtable.curr_root_rows = new_nrows; hdr->man_dtable.table_addr = new_addr; /* Extend heap to cover new root indirect block */ if(H5HF_hdr_adjust_heap(hdr, 2 * hdr->man_dtable.row_block_off[new_nrows - 1], (hssize_t)acc_dblock_free) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTEXTEND, FAIL, "can't increase space to cover root direct block") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_root_double() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_root_halve * * Purpose: Halve size of root indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Jun 12 2006 * *------------------------------------------------------------------------- */ static herr_t H5HF__man_iblock_root_halve(H5HF_indirect_t *iblock) { H5HF_hdr_t *hdr = iblock->hdr; /* Pointer to heap header */ haddr_t new_addr; /* New address of indirect block */ hsize_t acc_dblock_free; /* Accumulated free space in direct blocks */ hsize_t old_size; /* Old size of indirect block */ unsigned max_child_row; /* Row for max. child entry */ unsigned old_nrows; /* Old # of rows */ unsigned new_nrows; /* New # of rows */ unsigned u; /* Local index variable */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_STATIC /* Sanity check */ HDassert(iblock); HDassert(iblock->parent == NULL); HDassert(hdr); /* Compute maximum row used by child of indirect block */ max_child_row = iblock->max_child / hdr->man_dtable.cparam.width; /* Compute new # of rows in root indirect block */ new_nrows = (unsigned)1 << (1 + H5VM_log2_gen((uint64_t)max_child_row)); /* Check if the indirect block is NOT currently allocated in temp. file space */ /* (temp. file space does not need to be freed) */ if(!H5F_IS_TMP_ADDR(hdr->f, iblock->addr)) /* Free previous indirect block disk space */ if(H5MF_xfree(hdr->f, H5FD_MEM_FHEAP_IBLOCK, iblock->addr, (hsize_t)iblock->size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to free fractal heap indirect block file space") /* Compute free space in rows to delete */ acc_dblock_free = 0; for(u = new_nrows; u < iblock->nrows; u++) acc_dblock_free += hdr->man_dtable.row_tot_dblock_free[u] * hdr->man_dtable.cparam.width; /* Compute size of buffer needed for new indirect block */ old_nrows = iblock->nrows; iblock->nrows = new_nrows; old_size = iblock->size; iblock->size = H5HF_MAN_INDIRECT_SIZE(hdr, iblock->nrows); /* Allocate [temporary] space for the new indirect block on disk */ if(H5F_USE_TMP_SPACE(hdr->f)) { if(HADDR_UNDEF == (new_addr = H5MF_alloc_tmp(hdr->f, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end if */ else { if(HADDR_UNDEF == (new_addr = H5MF_alloc(hdr->f, H5FD_MEM_FHEAP_IBLOCK, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end else */ /* Resize pinned indirect block in the cache, if it has changed size */ if(old_size != iblock->size) { if(H5AC_resize_entry(iblock, (size_t)iblock->size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTRESIZE, FAIL, "unable to resize fractal heap indirect block") } /* end if */ /* Move object in cache, if it actually was relocated */ if(H5F_addr_ne(iblock->addr, new_addr)) { if(H5AC_move_entry(hdr->f, H5AC_FHEAP_IBLOCK, iblock->addr, new_addr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSPLIT, FAIL, "unable to move fractal heap root indirect block") iblock->addr = new_addr; } /* end if */ /* Re-allocate child block entry array */ if(NULL == (iblock->ents = H5FL_SEQ_REALLOC(H5HF_indirect_ent_t, iblock->ents, (size_t)(iblock->nrows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for direct entries") /* Check for needing to re-allocate filtered entry array */ if(hdr->filter_len > 0 && new_nrows < hdr->man_dtable.max_direct_rows) { /* Re-allocate filtered direct block entry array */ if(NULL == (iblock->filt_ents = H5FL_SEQ_REALLOC(H5HF_indirect_filt_ent_t, iblock->filt_ents, (size_t)(iblock->nrows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "memory allocation failed for filtered direct entries") } /* end if */ /* Check for needing to re-allocate child iblock pointer array */ if(old_nrows > hdr->man_dtable.max_direct_rows) { /* Check for shrinking away child iblock pointer array */ if(iblock->nrows > hdr->man_dtable.max_direct_rows) { unsigned indir_rows; /* Number of indirect rows in this indirect block */ /* Compute the number of direct rows for this indirect block */ indir_rows = iblock->nrows - hdr->man_dtable.max_direct_rows; /* Re-allocate child indirect block array */ if(NULL == (iblock->child_iblocks = H5FL_SEQ_REALLOC(H5HF_indirect_ptr_t, iblock->child_iblocks, (size_t)(indir_rows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_HEAP, H5E_NOSPACE, FAIL, "memory allocation failed for filtered direct entries") } /* end if */ else iblock->child_iblocks = (H5HF_indirect_ptr_t *)H5FL_SEQ_FREE(H5HF_indirect_ptr_t, iblock->child_iblocks); } /* end if */ /* Mark indirect block as dirty */ if(H5HF_iblock_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDIRTY, FAIL, "can't mark indirect block as dirty") /* Update other shared header info */ hdr->man_dtable.curr_root_rows = new_nrows; hdr->man_dtable.table_addr = new_addr; /* Shrink heap to only cover new root indirect block */ if(H5HF_hdr_adjust_heap(hdr, 2 * hdr->man_dtable.row_block_off[new_nrows - 1], -(hssize_t)acc_dblock_free) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSHRINK, FAIL, "can't reduce space to cover root direct block") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_root_halve() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_root_revert * * Purpose: Revert root indirect block back to root direct block * * Note: Any sections left pointing to the old root indirect block * will be cleaned up by the free space manager * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * May 31 2006 * *------------------------------------------------------------------------- */ static herr_t H5HF__man_iblock_root_revert(H5HF_indirect_t *root_iblock) { H5HF_hdr_t *hdr; /* Pointer to heap's header */ H5HF_direct_t *dblock = NULL; /* Pointer to new root indirect block */ haddr_t dblock_addr; /* Direct block's address in the file */ size_t dblock_size; /* Direct block's size */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_STATIC /* * Check arguments. */ HDassert(root_iblock); /* Set up local convenience variables */ hdr = root_iblock->hdr; dblock_addr = root_iblock->ents[0].addr; dblock_size = hdr->man_dtable.cparam.start_block_size; /* Get pointer to last direct block */ if(NULL == (dblock = H5HF__man_dblock_protect(hdr, dblock_addr, dblock_size, root_iblock, 0, H5AC__NO_FLAGS_SET))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, FAIL, "unable to protect fractal heap direct block") HDassert(dblock->parent == root_iblock); HDassert(dblock->par_entry == 0); /* Check for I/O filters on this heap */ if(hdr->filter_len > 0) { /* Set the header's pipeline information from the indirect block */ hdr->pline_root_direct_size = root_iblock->filt_ents[0].size; hdr->pline_root_direct_filter_mask = root_iblock->filt_ents[0].filter_mask; } /* end if */ /* Destroy flush dependency between old root iblock and new root direct block */ if(H5AC_destroy_flush_dependency(dblock->fd_parent, dblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNDEPEND, FAIL, "unable to destroy flush dependency") dblock->fd_parent = NULL; /* Detach direct block from parent */ if(H5HF__man_iblock_detach(dblock->parent, 0) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTATTACH, FAIL, "can't detach direct block from parent indirect block") dblock->parent = NULL; dblock->par_entry = 0; /* Create flush dependency between header and new root direct block */ if(H5AC_create_flush_dependency(hdr, dblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEPEND, FAIL, "unable to create flush dependency") dblock->fd_parent = hdr; /* Point root at direct block */ hdr->man_dtable.curr_root_rows = 0; hdr->man_dtable.table_addr = dblock_addr; /* Reset 'next block' iterator */ if(H5HF_hdr_reset_iter(hdr, (hsize_t)dblock_size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTRELEASE, FAIL, "can't reset block iterator") /* Extend heap to just cover first direct block */ if(H5HF_hdr_adjust_heap(hdr, (hsize_t)hdr->man_dtable.cparam.start_block_size, (hssize_t)hdr->man_dtable.row_tot_dblock_free[0]) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTEXTEND, FAIL, "can't increase space to cover root direct block") /* Scan free space sections to reset any 'parent' pointers */ if(H5HF__space_revert_root(hdr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTRESET, FAIL, "can't reset free space section info") done: if(dblock && H5AC_unprotect(hdr->f, H5AC_FHEAP_DBLOCK, dblock_addr, dblock, H5AC__NO_FLAGS_SET) < 0) HDONE_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap direct block") FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_root_revert() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_alloc_row * * Purpose: Allocate a "single" section for an object, out of a * "row" section. * * Note: Creates necessary direct & indirect blocks * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * July 6 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_alloc_row(H5HF_hdr_t *hdr, H5HF_free_section_t **sec_node) { H5HF_indirect_t *iblock = NULL; /* Pointer to indirect block */ H5HF_free_section_t *old_sec_node = *sec_node; /* Pointer to old indirect section node */ unsigned dblock_entry; /* Entry for direct block */ hbool_t iblock_held = FALSE; /* Flag to indicate that indirect block is held */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(hdr); HDassert(sec_node && old_sec_node); HDassert(old_sec_node->u.row.row < hdr->man_dtable.max_direct_rows); /* Check for serialized row section, or serialized / deleted indirect * section under it. */ if(old_sec_node->sect_info.state == H5FS_SECT_SERIALIZED || (H5FS_SECT_SERIALIZED == old_sec_node->u.row.under->sect_info.state) || (TRUE == old_sec_node->u.row.under->u.indirect.u.iblock->removed_from_cache)) /* Revive row and / or indirect section */ if(H5HF__sect_row_revive(hdr, old_sec_node) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTREVIVE, FAIL, "can't revive indirect section") /* Get a pointer to the indirect block covering the section */ if(NULL == (iblock = H5HF_sect_row_get_iblock(old_sec_node))) HGOTO_ERROR(H5E_HEAP, H5E_CANTGET, FAIL, "can't retrieve indirect block for row section") /* Hold indirect block in memory, until direct block can point to it */ if(H5HF_iblock_incr(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTINC, FAIL, "can't increment reference count on shared indirect block") iblock_held = TRUE; /* Reduce (& possibly re-add) 'row' section */ if(H5HF__sect_row_reduce(hdr, old_sec_node, &dblock_entry) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSHRINK, FAIL, "can't reduce row section node") /* Create direct block & single section */ if(H5HF__man_dblock_create(hdr, iblock, dblock_entry, NULL, sec_node) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTALLOC, FAIL, "can't allocate fractal heap direct block") done: /* Release hold on indirect block */ if(iblock_held) if(H5HF__iblock_decr(iblock) < 0) HDONE_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't decrement reference count on shared indirect block") FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_alloc_row() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_create * * Purpose: Allocate & initialize a managed indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Mar 6 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_create(H5HF_hdr_t *hdr, H5HF_indirect_t *par_iblock, unsigned par_entry, unsigned nrows, unsigned max_rows, haddr_t *addr_p) { H5HF_indirect_t *iblock = NULL; /* Pointer to indirect block */ size_t u; /* Local index variable */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(hdr); HDassert(nrows > 0); HDassert(addr_p); /* * Allocate file and memory data structures. */ if(NULL == (iblock = H5FL_MALLOC(H5HF_indirect_t))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for fractal heap indirect block") /* Reset the metadata cache info for the heap header */ HDmemset(&iblock->cache_info, 0, sizeof(H5AC_info_t)); /* Share common heap information */ iblock->hdr = hdr; if(H5HF_hdr_incr(hdr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTINC, FAIL, "can't increment reference count on shared heap header") /* Set info for indirect block */ iblock->rc = 0; iblock->nrows = nrows; iblock->max_rows = max_rows; iblock->removed_from_cache = FALSE; /* Compute size of buffer needed for indirect block */ iblock->size = H5HF_MAN_INDIRECT_SIZE(hdr, iblock->nrows); /* Allocate child block entry array */ if(NULL == (iblock->ents = H5FL_SEQ_MALLOC(H5HF_indirect_ent_t, (size_t)(iblock->nrows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for block entries") /* Initialize indirect block entry tables */ for(u = 0; u < (iblock->nrows * hdr->man_dtable.cparam.width); u++) iblock->ents[u].addr = HADDR_UNDEF; /* Check for I/O filters to apply to this heap */ if(hdr->filter_len > 0) { unsigned dir_rows; /* Number of direct rows in this indirect block */ /* Compute the number of direct rows for this indirect block */ dir_rows = MIN(iblock->nrows, hdr->man_dtable.max_direct_rows); /* Allocate & initialize indirect block filtered entry array */ if(NULL == (iblock->filt_ents = H5FL_SEQ_CALLOC(H5HF_indirect_filt_ent_t, (size_t)(dir_rows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for block entries") } /* end if */ else iblock->filt_ents = NULL; /* Check if we have any indirect block children */ if(iblock->nrows > hdr->man_dtable.max_direct_rows) { unsigned indir_rows; /* Number of indirect rows in this indirect block */ /* Compute the number of indirect rows for this indirect block */ indir_rows = iblock->nrows - hdr->man_dtable.max_direct_rows; /* Allocate & initialize child indirect block pointer array */ if(NULL == (iblock->child_iblocks = H5FL_SEQ_CALLOC(H5HF_indirect_ptr_t, (size_t)(indir_rows * hdr->man_dtable.cparam.width)))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for block entries") } /* end if */ else iblock->child_iblocks = NULL; /* Allocate [temporary] space for the indirect block on disk */ if(H5F_USE_TMP_SPACE(hdr->f)) { if(HADDR_UNDEF == (*addr_p = H5MF_alloc_tmp(hdr->f, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end if */ else { if(HADDR_UNDEF == (*addr_p = H5MF_alloc(hdr->f, H5FD_MEM_FHEAP_IBLOCK, (hsize_t)iblock->size))) HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "file allocation failed for fractal heap indirect block") } /* end else */ iblock->addr = *addr_p; /* Attach to parent indirect block, if there is one */ iblock->parent = par_iblock; iblock->par_entry = par_entry; if(iblock->parent) { /* Attach new block to parent */ if(H5HF_man_iblock_attach(iblock->parent, par_entry, *addr_p) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTATTACH, FAIL, "can't attach indirect block to parent indirect block") /* Compute the indirect block's offset in the heap's address space */ /* (based on parent's block offset) */ iblock->block_off = par_iblock->block_off; iblock->block_off += hdr->man_dtable.row_block_off[par_entry / hdr->man_dtable.cparam.width]; iblock->block_off += hdr->man_dtable.row_block_size[par_entry / hdr->man_dtable.cparam.width] * (par_entry % hdr->man_dtable.cparam.width); /* Set indirect block parent as flush dependency parent */ iblock->fd_parent = par_iblock; } /* end if */ else { iblock->block_off = 0; /* Must be the root indirect block... */ /* Set heap header as flush dependency parent */ iblock->fd_parent = hdr; } /* end else */ /* Update indirect block's statistics */ iblock->nchildren = 0; iblock->max_child = 0; /* Cache the new indirect block */ if(H5AC_insert_entry(hdr->f, H5AC_FHEAP_IBLOCK, *addr_p, iblock, H5AC__NO_FLAGS_SET) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTINIT, FAIL, "can't add fractal heap indirect block to cache") done: if(ret_value < 0) if(iblock) if(H5HF_man_iblock_dest(iblock) < 0) HDONE_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to destroy fractal heap indirect block") FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_create() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_protect * * Purpose: Convenience wrapper around H5AC_protect on an indirect block * * Return: Pointer to indirect block on success, NULL on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Apr 17 2006 * *------------------------------------------------------------------------- */ H5HF_indirect_t * H5HF__man_iblock_protect(H5HF_hdr_t *hdr, haddr_t iblock_addr, unsigned iblock_nrows, H5HF_indirect_t *par_iblock, unsigned par_entry, hbool_t must_protect, unsigned flags, hbool_t *did_protect) { H5HF_parent_t par_info; /* Parent info for loading block */ H5HF_indirect_t *iblock = NULL; /* Indirect block from cache */ hbool_t should_protect = FALSE; /* Whether we should protect the indirect block or not */ H5HF_indirect_t *ret_value = NULL; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(hdr); HDassert(H5F_addr_defined(iblock_addr)); HDassert(iblock_nrows > 0); HDassert(did_protect); /* only H5AC__READ_ONLY_FLAG may appear in flags */ HDassert((flags & (unsigned)(~H5AC__READ_ONLY_FLAG)) == 0); /* Check if we are allowed to use existing pinned iblock pointer */ if(!must_protect) { /* Check for this block already being pinned */ if(par_iblock) { unsigned indir_idx; /* Index in parent's child iblock pointer array */ /* Sanity check */ HDassert(par_iblock->child_iblocks); HDassert(par_entry >= (hdr->man_dtable.max_direct_rows * hdr->man_dtable.cparam.width)); /* Compute index in parent's child iblock pointer array */ indir_idx = par_entry - (hdr->man_dtable.max_direct_rows * hdr->man_dtable.cparam.width); /* Check for pointer to pinned indirect block in parent */ if(par_iblock->child_iblocks[indir_idx]) iblock = par_iblock->child_iblocks[indir_idx]; else should_protect = TRUE; } /* end if */ else { /* Check for root indirect block */ if(H5F_addr_eq(iblock_addr, hdr->man_dtable.table_addr)) { /* Check for valid pointer to pinned indirect block in root */ if(H5HF_ROOT_IBLOCK_PINNED == hdr->root_iblock_flags) { /* Sanity check */ HDassert(NULL != hdr->root_iblock); /* Return the pointer to the pinned root indirect block */ iblock = hdr->root_iblock; } /* end if */ else { /* Sanity check */ HDassert(NULL == hdr->root_iblock); should_protect = TRUE; } /* end else */ } /* end if */ else should_protect = TRUE; } /* end else */ } /* end if */ /* Check for protecting indirect block */ if(must_protect || should_protect) { H5HF_iblock_cache_ud_t cache_udata; /* User-data for callback */ /* Set up parent info */ par_info.hdr = hdr; par_info.iblock = par_iblock; par_info.entry = par_entry; /* Set up user data for protect call */ cache_udata.f = hdr->f; cache_udata.par_info = &par_info; cache_udata.nrows = &iblock_nrows; /* Protect the indirect block */ if(NULL == (iblock = (H5HF_indirect_t *)H5AC_protect(hdr->f, H5AC_FHEAP_IBLOCK, iblock_addr, &cache_udata, flags))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, NULL, "unable to protect fractal heap indirect block") /* Set the indirect block's address */ iblock->addr = iblock_addr; /* Check for root indirect block */ if(iblock->block_off == 0) { /* Sanity check - shouldn't be recursively protecting root indirect block */ HDassert(0 == (hdr->root_iblock_flags & H5HF_ROOT_IBLOCK_PROTECTED)); /* Check if we should set the root iblock pointer */ if(0 == hdr->root_iblock_flags) { HDassert(NULL == hdr->root_iblock); hdr->root_iblock = iblock; } /* end if */ /* Indicate that the root indirect block is protected */ hdr->root_iblock_flags |= H5HF_ROOT_IBLOCK_PROTECTED; } /* end if */ /* Indicate that the indirect block was protected */ *did_protect = TRUE; } /* end if */ else /* Indicate that the indirect block was _not_ protected */ *did_protect = FALSE; /* Set the return value */ ret_value = iblock; done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_protect() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_unprotect * * Purpose: Convenience wrapper around H5AC_unprotect on an indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@hdfgroup.org * Aug 17 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_unprotect(H5HF_indirect_t *iblock, unsigned cache_flags, hbool_t did_protect) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(iblock); /* Check if we previously protected this indirect block */ /* (as opposed to using an existing pointer to a pinned child indirect block) */ if(did_protect) { /* Check for root indirect block */ if(iblock->block_off == 0) { /* Sanity check - shouldn't be recursively unprotecting root indirect block */ HDassert(iblock->hdr->root_iblock_flags & H5HF_ROOT_IBLOCK_PROTECTED); /* Check if we should reset the root iblock pointer */ if(H5HF_ROOT_IBLOCK_PROTECTED == iblock->hdr->root_iblock_flags) { HDassert(NULL != iblock->hdr->root_iblock); iblock->hdr->root_iblock = NULL; } /* end if */ /* Indicate that the root indirect block is unprotected */ iblock->hdr->root_iblock_flags &= (unsigned)(~(H5HF_ROOT_IBLOCK_PROTECTED)); } /* end if */ /* Unprotect the indirect block */ if(H5AC_unprotect(iblock->hdr->f, H5AC_FHEAP_IBLOCK, iblock->addr, iblock, cache_flags) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap indirect block") } /* end if */ done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_unprotect() */ /*------------------------------------------------------------------------- * Function: H5HF_man_iblock_attach * * Purpose: Attach a child block (direct or indirect) to an indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * May 30 2006 * *------------------------------------------------------------------------- */ herr_t H5HF_man_iblock_attach(H5HF_indirect_t *iblock, unsigned entry, haddr_t child_addr) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_NOAPI_NOINIT /* * Check arguments. */ HDassert(iblock); HDassert(H5F_addr_defined(child_addr)); HDassert(!H5F_addr_defined(iblock->ents[entry].addr)); /* Increment the reference count on this indirect block */ if(H5HF_iblock_incr(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTINC, FAIL, "can't increment reference count on shared indirect block") /* Point at the child block */ iblock->ents[entry].addr = child_addr; /* Check for I/O filters on this heap */ if(iblock->hdr->filter_len > 0) { unsigned row; /* Row for entry */ /* Sanity check */ HDassert(iblock->filt_ents); /* Compute row for entry */ row = entry / iblock->hdr->man_dtable.cparam.width; /* If this is a direct block, set its initial size */ if(row < iblock->hdr->man_dtable.max_direct_rows) iblock->filt_ents[entry].size = iblock->hdr->man_dtable.row_block_size[row]; } /* end if */ /* Check for max. entry used */ if(entry > iblock->max_child) iblock->max_child = entry; /* Increment the # of child blocks */ iblock->nchildren++; /* Mark indirect block as modified */ if(H5HF_iblock_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDIRTY, FAIL, "can't mark indirect block as dirty") done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF_man_iblock_attach() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_detach * * Purpose: Detach a child block (direct or indirect) from an indirect block * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * May 31 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_detach(H5HF_indirect_t *iblock, unsigned entry) { H5HF_hdr_t *hdr; /* Fractal heap header */ H5HF_indirect_t *del_iblock = NULL; /* Pointer to protected indirect block, when deleting */ unsigned row; /* Row for entry */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(iblock); HDassert(iblock->nchildren); /* Set up convenience variables */ hdr = iblock->hdr; /* Reset address of entry */ iblock->ents[entry].addr = HADDR_UNDEF; /* Compute row for entry */ row = entry / hdr->man_dtable.cparam.width; /* Check for I/O filters on this heap */ if(hdr->filter_len > 0) { /* Sanity check */ HDassert(iblock->filt_ents); /* If this is a direct block, reset its initial size */ if(row < hdr->man_dtable.max_direct_rows) { iblock->filt_ents[entry].size = 0; iblock->filt_ents[entry].filter_mask = 0; } /* end if */ } /* end if */ /* Check for indirect block being detached */ if(row >= hdr->man_dtable.max_direct_rows) { unsigned indir_idx; /* Index in parent's child iblock pointer array */ /* Sanity check */ HDassert(iblock->child_iblocks); /* Compute index in child iblock pointer array */ indir_idx = entry - (hdr->man_dtable.max_direct_rows * hdr->man_dtable.cparam.width); /* Sanity check */ HDassert(iblock->child_iblocks[indir_idx]); /* Reset pointer to child indirect block in parent */ iblock->child_iblocks[indir_idx] = NULL; } /* end if */ /* Decrement the # of child blocks */ /* (If the number of children drop to 0, the indirect block will be * removed from the heap when its ref. count drops to zero and the * metadata cache calls the indirect block destructor) */ iblock->nchildren--; /* Reduce the max. entry used, if necessary */ if(entry == iblock->max_child) { if(iblock->nchildren > 0) while(!H5F_addr_defined(iblock->ents[iblock->max_child].addr)) iblock->max_child--; else iblock->max_child = 0; } /* end if */ /* If this is the root indirect block handle some special cases */ if(iblock->block_off == 0) { /* If the number of children drops to 1, and that child is the first * direct block in the heap, convert the heap back to using a root * direct block */ if(iblock->nchildren == 1 && H5F_addr_defined(iblock->ents[0].addr)) if(H5HF__man_iblock_root_revert(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSHRINK, FAIL, "can't convert root indirect block back to root direct block") /* If the indirect block wasn't removed already (by reverting it) */ if(!iblock->removed_from_cache) { /* Check for reducing size of root indirect block */ if(iblock->nchildren > 0 && hdr->man_dtable.cparam.start_root_rows != 0 && entry > iblock->max_child) { unsigned max_child_row; /* Row for max. child entry */ /* Compute information needed for determining whether to reduce size of root indirect block */ max_child_row = iblock->max_child / hdr->man_dtable.cparam.width; /* Check if the root indirect block should be reduced */ if(iblock->nrows > 1 && max_child_row <= (iblock->nrows / 2)) if(H5HF__man_iblock_root_halve(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSHRINK, FAIL, "can't reduce size of root indirect block") } /* end if */ } /* end if */ } /* end if */ /* If the indirect block wasn't removed already (by reverting it) */ if(!iblock->removed_from_cache) { /* Mark indirect block as modified */ if(H5HF_iblock_dirty(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDIRTY, FAIL, "can't mark indirect block as dirty") /* Check for last child being removed from indirect block */ if(iblock->nchildren == 0) { hbool_t did_protect = FALSE; /* Whether the indirect block was protected */ /* If this indirect block's refcount is >1, then it's being deleted * from the fractal heap (since its nchildren == 0), but is still * referred to from free space sections in the heap (refcount >1). * Its space in the file needs to be freed now, and it also needs * to be removed from the metadata cache now, in case the space in * the file is reused by another piece of metadata that is inserted * into the cache before the indirect block's entry is evicted * (having two entries at the same address would be an error, from * the cache's perspective). */ /* Lock indirect block for deletion */ if(NULL == (del_iblock = H5HF__man_iblock_protect(hdr, iblock->addr, iblock->nrows, iblock->parent, iblock->par_entry, TRUE, H5AC__NO_FLAGS_SET, &did_protect))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, FAIL, "unable to protect fractal heap indirect block") HDassert(did_protect == TRUE); /* Check for deleting root indirect block (and no root direct block) */ if(iblock->block_off == 0 && hdr->man_dtable.curr_root_rows > 0) /* Reset header information back to "empty heap" state */ if(H5HF__hdr_empty(hdr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTSHRINK, FAIL, "can't make heap empty") /* Detach from parent indirect block */ if(iblock->parent) { /* Destroy flush dependency between indirect block and parent */ if(H5AC_destroy_flush_dependency(iblock->fd_parent, iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNDEPEND, FAIL, "unable to destroy flush dependency") iblock->fd_parent = NULL; /* Detach from parent indirect block */ if(H5HF__man_iblock_detach(iblock->parent, iblock->par_entry) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTATTACH, FAIL, "can't detach from parent indirect block") iblock->parent = NULL; iblock->par_entry = 0; } /* end if */ } /* end if */ } /* end if */ /* Decrement the reference count on this indirect block if we're not deleting it */ /* (should be after iblock needs to be modified, so that potential 'unpin' * on this indirect block doesn't invalidate the 'iblock' variable, if it's * not being deleted) */ if(H5HF__iblock_decr(iblock) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't decrement reference count on shared indirect block") iblock = NULL; /* Delete indirect block from cache, if appropriate */ if(del_iblock) { unsigned cache_flags = H5AC__NO_FLAGS_SET; /* Flags for unprotect */ hbool_t took_ownership = FALSE; /* Flag to indicate that block ownership has transitioned */ /* If the refcount is still >0, unpin the block and take ownership * from the cache, otherwise let the cache destroy it. */ if(del_iblock->rc > 0) { cache_flags |= (H5AC__DELETED_FLAG | H5AC__TAKE_OWNERSHIP_FLAG); cache_flags |= H5AC__UNPIN_ENTRY_FLAG; took_ownership = TRUE; } /* end if */ else { /* Entry should be removed from the cache */ cache_flags |= H5AC__DELETED_FLAG; /* If the indirect block is in real file space, tell * the cache to free its file space as well. */ if(!H5F_IS_TMP_ADDR(hdr->f, del_iblock->addr)) cache_flags |= H5AC__FREE_FILE_SPACE_FLAG; } /* end else */ /* Unprotect the indirect block, with appropriate flags */ if(H5HF__man_iblock_unprotect(del_iblock, cache_flags, TRUE) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap indirect block") /* if took ownership, free file space & mark block as removed from cache */ if(took_ownership) { /* Free indirect block disk space, if it's in real space */ if(!H5F_IS_TMP_ADDR(hdr->f, del_iblock->addr)) if(H5MF_xfree(hdr->f, H5FD_MEM_FHEAP_IBLOCK, del_iblock->addr, (hsize_t)del_iblock->size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to free fractal heap indirect block file space") del_iblock->addr = HADDR_UNDEF; /* Mark block as removed from the cache */ del_iblock->removed_from_cache = TRUE; } /* end if */ } /* end if */ done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_detach() */ /*------------------------------------------------------------------------- * Function: H5HF_man_iblock_entry_addr * * Purpose: Retrieve the address of an indirect block's child * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * July 10 2006 * *------------------------------------------------------------------------- */ herr_t H5HF_man_iblock_entry_addr(H5HF_indirect_t *iblock, unsigned entry, haddr_t *child_addr) { FUNC_ENTER_NOAPI_NOINIT_NOERR /* * Check arguments. */ HDassert(iblock); HDassert(child_addr); /* Retrieve address of entry */ *child_addr = iblock->ents[entry].addr; FUNC_LEAVE_NOAPI(SUCCEED) } /* end H5HF_man_iblock_entry_addr() */ /*------------------------------------------------------------------------- * Function: H5HF__man_iblock_delete * * Purpose: Delete a managed indirect block * * Note: This routine does _not_ modify any indirect block that points * to this indirect block, it is assumed that the whole heap is * being deleted in a top-down fashion. * * Return: SUCCEED/FAIL * * Programmer: Quincey Koziol * koziol@hdfgroup.org * Aug 7 2006 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_delete(H5HF_hdr_t *hdr, haddr_t iblock_addr, unsigned iblock_nrows, H5HF_indirect_t *par_iblock, unsigned par_entry) { H5HF_indirect_t *iblock; /* Pointer to indirect block */ unsigned row, col; /* Current row & column in indirect block */ unsigned entry; /* Current entry in row */ unsigned cache_flags = H5AC__NO_FLAGS_SET; /* Flags for unprotecting indirect block */ hbool_t did_protect; /* Whether we protected the indirect block or not */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(hdr); HDassert(H5F_addr_defined(iblock_addr)); HDassert(iblock_nrows > 0); /* Lock indirect block */ if(NULL == (iblock = H5HF__man_iblock_protect(hdr, iblock_addr, iblock_nrows, par_iblock, par_entry, TRUE, H5AC__NO_FLAGS_SET, &did_protect))) HGOTO_ERROR(H5E_HEAP, H5E_CANTPROTECT, FAIL, "unable to protect fractal heap indirect block") HDassert(iblock->nchildren > 0); HDassert(did_protect == TRUE); /* Iterate over rows in this indirect block */ entry = 0; for(row = 0; row < iblock->nrows; row++) { /* Iterate over entries in this row */ for(col = 0; col < hdr->man_dtable.cparam.width; col++, entry++) { /* Check for child entry at this position */ if(H5F_addr_defined(iblock->ents[entry].addr)) { /* Are we in a direct or indirect block row */ if(row < hdr->man_dtable.max_direct_rows) { hsize_t dblock_size; /* Size of direct block on disk */ /* Check for I/O filters on this heap */ if(hdr->filter_len > 0) dblock_size = iblock->filt_ents[entry].size; else dblock_size = hdr->man_dtable.row_block_size[row]; /* Delete child direct block */ if(H5HF__man_dblock_delete(hdr->f, iblock->ents[entry].addr, dblock_size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to release fractal heap child direct block") } /* end if */ else { hsize_t row_block_size; /* The size of blocks in this row */ unsigned child_nrows; /* Number of rows in new indirect block */ /* Get the row's block size */ row_block_size = (hsize_t)hdr->man_dtable.row_block_size[row]; /* Compute # of rows in next child indirect block to use */ child_nrows = H5HF_dtable_size_to_rows(&hdr->man_dtable, row_block_size); /* Delete child indirect block */ if(H5HF__man_iblock_delete(hdr, iblock->ents[entry].addr, child_nrows, iblock, entry) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTFREE, FAIL, "unable to release fractal heap child indirect block") } /* end else */ } /* end if */ } /* end for */ } /* end row */ #ifndef NDEBUG { unsigned iblock_status = 0; /* Indirect block's status in the metadata cache */ /* Check the indirect block's status in the metadata cache */ if(H5AC_get_entry_status(hdr->f, iblock_addr, &iblock_status) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTGET, FAIL, "unable to check metadata cache status for indirect block") /* Check if indirect block is pinned */ HDassert(!(iblock_status & H5AC_ES__IS_PINNED)); } #endif /* NDEBUG */ /* Indicate that the indirect block should be deleted */ cache_flags |= H5AC__DIRTIED_FLAG | H5AC__DELETED_FLAG; /* If the indirect block is in real file space, tell * the cache to free its file space as well. */ if (!H5F_IS_TMP_ADDR(hdr->f, iblock_addr)) cache_flags |= H5AC__FREE_FILE_SPACE_FLAG; done: /* Unprotect the indirect block, with appropriate flags */ if(iblock && H5HF__man_iblock_unprotect(iblock, cache_flags, did_protect) < 0) HDONE_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap indirect block") FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_delete() */ /*------------------------------------------------------------------------- * * Function: H5HF__man_iblock_size * * Purpose: Gather storage used for the indirect block in fractal heap * * Return: non-negative on success, negative on error * * Programmer: Vailin Choi * July 12 2007 *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_size(H5F_t *f, H5HF_hdr_t *hdr, haddr_t iblock_addr, unsigned nrows, H5HF_indirect_t *par_iblock, unsigned par_entry, hsize_t *heap_size) { H5HF_indirect_t *iblock = NULL; /* Pointer to indirect block */ hbool_t did_protect; /* Whether we protected the indirect block or not */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(f); HDassert(hdr); HDassert(H5F_addr_defined(iblock_addr)); HDassert(heap_size); /* Protect the indirect block */ if(NULL == (iblock = H5HF__man_iblock_protect(hdr, iblock_addr, nrows, par_iblock, par_entry, FALSE, H5AC__READ_ONLY_FLAG, &did_protect))) HGOTO_ERROR(H5E_HEAP, H5E_CANTLOAD, FAIL, "unable to load fractal heap indirect block") /* Accumulate size of this indirect block */ *heap_size += iblock->size; /* Indirect entries in this indirect block */ if(iblock->nrows > hdr->man_dtable.max_direct_rows) { unsigned first_row_bits; /* Number of bits used bit addresses in first row */ unsigned num_indirect_rows; /* Number of rows of blocks in each indirect block */ unsigned entry; /* Current entry in row */ size_t u; /* Local index variable */ entry = hdr->man_dtable.max_direct_rows * hdr->man_dtable.cparam.width; first_row_bits = H5VM_log2_of2((uint32_t)hdr->man_dtable.cparam.start_block_size) + H5VM_log2_of2(hdr->man_dtable.cparam.width); num_indirect_rows = (H5VM_log2_gen(hdr->man_dtable.row_block_size[hdr->man_dtable.max_direct_rows]) - first_row_bits) + 1; for(u = hdr->man_dtable.max_direct_rows; u < iblock->nrows; u++, num_indirect_rows++) { size_t v; /* Local index variable */ for(v = 0; v < hdr->man_dtable.cparam.width; v++, entry++) if(H5F_addr_defined(iblock->ents[entry].addr)) if(H5HF__man_iblock_size(f, hdr, iblock->ents[entry].addr, num_indirect_rows, iblock, entry, heap_size) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTLOAD, FAIL, "unable to get fractal heap storage info for indirect block") } /* end for */ } /* end if */ done: /* Release the indirect block */ if(iblock && H5HF__man_iblock_unprotect(iblock, H5AC__NO_FLAGS_SET, did_protect) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTUNPROTECT, FAIL, "unable to release fractal heap indirect block") iblock = NULL; FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_size() */ /*------------------------------------------------------------------------- * * Function: H5HF_man_iblock_parent_info * * Purpose: Determine the parent block's offset and entry location * (within it's parent) of an indirect block, given its offset * within the heap. * * Return: Non-negative on success / Negative on failure * * Programmer: Quincey Koziol * koziol@lbl.gov * Jan 14 2018 * *------------------------------------------------------------------------- */ herr_t H5HF__man_iblock_parent_info(const H5HF_hdr_t *hdr, hsize_t block_off, hsize_t *ret_par_block_off, unsigned *ret_entry) { hsize_t par_block_off; /* Offset of parent within heap */ hsize_t prev_par_block_off; /* Offset of previous parent block within heap */ unsigned row, col; /* Row & column for block */ unsigned prev_row = 0, prev_col = 0;/* Previous row & column for block */ herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_PACKAGE /* * Check arguments. */ HDassert(hdr); HDassert(block_off > 0); HDassert(ret_entry); /* Look up row & column for object */ if(H5HF_dtable_lookup(&hdr->man_dtable, block_off, &row, &col) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTCOMPUTE, FAIL, "can't compute row & column of block") /* Sanity check - first lookup must be an indirect block */ HDassert(row >= hdr->man_dtable.max_direct_rows); /* Traverse down, until a direct block at the offset is found, then * use previous (i.e. parent's) offset, row, and column. */ prev_par_block_off = par_block_off = 0; while(row >= hdr->man_dtable.max_direct_rows) { /* Retain previous parent block offset */ prev_par_block_off = par_block_off; /* Compute the new parent indirect block's offset in the heap's address space */ /* (based on previous block offset) */ par_block_off += hdr->man_dtable.row_block_off[row]; par_block_off += hdr->man_dtable.row_block_size[row] * col; /* Preserve current row & column */ prev_row = row; prev_col = col; /* Look up row & column in new indirect block for object */ if(H5HF_dtable_lookup(&hdr->man_dtable, (block_off - par_block_off), &row, &col) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTCOMPUTE, FAIL, "can't compute row & column of block") } /* end while */ /* Sanity check */ HDassert(row == 0); HDassert(col == 0); /* Set return parameters */ *ret_par_block_off = prev_par_block_off; *ret_entry = (prev_row * hdr->man_dtable.cparam.width) + prev_col; done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF__man_iblock_par_info() */ /*------------------------------------------------------------------------- * Function: H5HF_man_iblock_dest * * Purpose: Destroys a fractal heap indirect block in memory. * * Return: Non-negative on success/Negative on failure * * Programmer: Quincey Koziol * koziol@ncsa.uiuc.edu * Mar 6 2006 * *------------------------------------------------------------------------- */ herr_t H5HF_man_iblock_dest(H5HF_indirect_t *iblock) { herr_t ret_value = SUCCEED; /* Return value */ FUNC_ENTER_NOAPI_NOINIT /* * Check arguments. */ HDassert(iblock); HDassert(iblock->rc == 0); /* Decrement reference count on shared info */ HDassert(iblock->hdr); if(H5HF_hdr_decr(iblock->hdr) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't decrement reference count on shared heap header") if(iblock->parent) if(H5HF__iblock_decr(iblock->parent) < 0) HGOTO_ERROR(H5E_HEAP, H5E_CANTDEC, FAIL, "can't decrement reference count on shared indirect block") /* Release entry tables */ if(iblock->ents) iblock->ents = H5FL_SEQ_FREE(H5HF_indirect_ent_t, iblock->ents); if(iblock->filt_ents) iblock->filt_ents = H5FL_SEQ_FREE(H5HF_indirect_filt_ent_t, iblock->filt_ents); if(iblock->child_iblocks) iblock->child_iblocks = H5FL_SEQ_FREE(H5HF_indirect_ptr_t, iblock->child_iblocks); /* Free fractal heap indirect block info */ iblock = H5FL_FREE(H5HF_indirect_t, iblock); done: FUNC_LEAVE_NOAPI(ret_value) } /* end H5HF_man_iblock_dest() */
./openacc-vv/serial_copy.c
#include "acc_testsuite.h" #ifndef T1 //T1:serial,data,data-region,V:2.6-2.7 int test1(){ int err = 0; srand(SEED); real_t * a = (real_t *)malloc(n * sizeof(real_t)); real_t * a_host = (real_t *)malloc(n * sizeof(real_t)); for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); a_host[x] = a[x]; } #pragma acc serial copy(a[0:n]) { #pragma acc loop for (int x = 0; x < n; ++x){ a[x] = 2 * a[x]; } } for (int x = 0; x < n; ++x){ if (fabs(a[x] - (2 * a_host[x])) > PRECISION){ err = 1; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/acc_async_test.cpp
#include "acc_testsuite.h" #ifndef T1 //T1:async,runtime,construct-independent,V:2.0-2.7 int test1(){ int err = 0; real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = 0; } #pragma acc enter data copyin(a[0:n], b[0:n]) create(c[0:n]) async(1) #pragma acc enter data copyin(d[0:n]) create(e[0:n]) async(2) #pragma acc parallel present(a[0:n], b[0:n], c[0:n]) async(1) { #pragma acc loop for (int x = 0; x < n; ++x){ c[x] = a[x] + b[x]; } } #pragma acc parallel present(c[0:n], d[0:n], e[0:n]) async(1) wait(2) { #pragma acc loop for (int x = 0; x < n; ++x){ e[x] = c[x] + d[x]; } } #pragma acc exit data copyout(e[0:n]) async(1) while (!acc_async_test(1)); for (int x = 0; x < n; ++x){ if (fabs(e[x] - (a[x] + b[x] + d[x])) > PRECISION){ err += 1; } } return err; } #endif #ifndef T2 //T2:async,runtime,construct-independent,V:1.0-2.7 int test2(){ int err = 0; real_t *a = new real_t[n]; real_t *b = new real_t[n]; real_t *c = new real_t[n]; real_t *d = new real_t[n]; real_t *e = new real_t[n]; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = 0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n]) create(c[0:n]) copyout(e[0:n]) { #pragma acc parallel present(a[0:n], b[0:n], c[0:n]) async(1) { #pragma acc loop for (int x = 0; x < n; ++x) { c[x] = a[x] + b[x]; } } #pragma acc parallel present(c[0:n], d[0:n], e[0:n]) async(1) { #pragma acc loop for (int x = 0; x < n; ++x) { e[x] = c[x] + d[x]; } } while (!acc_async_test(1)); } for (int x = 0; x < n; ++x) { if (fabs(e[x] - (a[x] + b[x] + d[x])) > PRECISION) { err += 1; } } return err; } #endif #ifndef T3 //T3:async,runtime,construct-independent,V:2.5-2.7 int test3() { int err = 0; real_t* a = new real_t[n]; real_t* b = new real_t[n]; real_t* c = new real_t[n]; real_t* d = new real_t[n]; real_t* e = new real_t[n]; int async_val = acc_get_default_async(); for (int x = 0; x < n; ++x) { a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0; d[x] = rand() / (real_t)(RAND_MAX / 10); e[x] = 0; } #pragma acc data copyin(a[0:n], b[0:n], d[0:n]) create(c[0:n]) copyout(e[0:n]) { #pragma acc parallel present(a[0:n], b[0:n], c[0:n]) async { #pragma acc loop for (int x = 0; x < n; ++x) { c[x] = a[x] + b[x]; } } #pragma acc parallel present(c[0:n], d[0:n], e[0:n]) async { #pragma acc loop for (int x = 0; x < n; ++x) { e[x] = c[x] + d[x]; } } while (!acc_async_test(async_val)); } for (int x = 0; x < n; ++x) { if (fabs(e[x] - (a[x] + b[x] + d[x])) > PRECISION) { err += 1; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif #ifndef T2 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test2(); } if (failed != 0){ failcode = failcode + (1 << 1); } #endif #ifndef T3 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x) { failed = failed + test3(); } if (failed != 0) { failcode = failcode + (1 << 2); } #endif return failcode; }
./openacc-vv/kernels_loop_vector_blocking.c
#include "acc_testsuite.h" #ifndef T1 //T1:kernels,loop,V:1.0-2.7 int test1(){ int err = 0; srand(SEED); real_t * a = (real_t *)malloc(n * sizeof(real_t)); real_t * b = (real_t *)malloc(n * sizeof(real_t)); real_t * c = (real_t *)malloc(n * sizeof(real_t)); real_t multiplyer = 1; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); c[x] = 0.0; } #pragma acc data copyin(a[0:n], b[0:n]) copyout(c[0:n]) { #pragma acc kernels { #pragma acc loop vector for (int x = 0; x < n; ++x){ c[x] = (a[x] + b[x]) * multiplyer; } multiplyer += 1; #pragma acc loop vector for (int x = 0; x < n; ++x){ c[x] += (a[x] + b[x]) * multiplyer; } } } for (int x = 0; x < n; ++x){ if (fabs(c[x] - 3 * (a[x] + b[x])) > PRECISION){ err + 1; break; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/atomic_structured_predecrement_assign.c
#include "acc_testsuite.h" #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = (real_t *)malloc(n * sizeof(real_t)); real_t *b = (real_t *)malloc(n * sizeof(real_t)); int *c = (int *)malloc(n * sizeof(int)); int *distribution = (int *)malloc(10 * sizeof(int)); int *distribution_comparison = (int *)malloc(10 * sizeof(int)); bool found = false; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); b[x] = rand() / (real_t)(RAND_MAX / 10); } for (int x = 0; x < 10; ++x){ distribution[x] = 0; distribution_comparison[x] = 0; } #pragma acc data copyin(a[0:n], b[0:n]) copy(distribution[0:10]) copyout(c[0:n]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc atomic capture { --distribution[(int) (a[x]*b[x]/10)]; c[x] = distribution[(int) (a[x]*b[x]/10)]; } } } } for (int x = 0; x < n; ++x){ distribution_comparison[(int) (a[x]*b[x]/10)]--; } for (int x = 0; x < 10; ++x){ if (distribution_comparison[x] != distribution[x]){ err += 1; break; } } for (int x = 0; x < 10; ++x){ for (int y = 0; y > distribution[x]; --y){ for (int z = 0; z < n; ++z){ if (c[z] == y - 1 && x == (int) (a[z] * b[z] / 10)){ found = true; break; } } if (!found){ err++; } found = false; } } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./openacc-vv/atomic_capture_expr_minus_x.c
#include "acc_testsuite.h" bool is_possible(real_t* a, real_t* b, int length, real_t prev){ if (length == 0){ return true; } real_t *passed_a = (real_t *)malloc((length - 1) * sizeof(real_t)); real_t *passed_b = (real_t *)malloc((length - 1) * sizeof(real_t)); for (int x = 0; x < length; ++x){ if (fabs(b[x] - (a[x] - prev)) < PRECISION){ for (int y = 0; y < x; ++y){ passed_a[y] = a[y]; passed_b[y] = b[y]; } for (int y = x + 1; y < length; ++y){ passed_a[y - 1] = a[y]; passed_b[y - 1] = b[y]; } if (is_possible(passed_a, passed_b, length - 1, b[x])){ free(passed_a); free(passed_b); return true; } } } free(passed_a); free(passed_b); return false; } bool possible_result(real_t * remaining_combinations, int length, real_t current_value, real_t test_value){ if (length == 0){ if (fabs(current_value - test_value) > PRECISION){ return true; } else { return false; } } real_t * passed = (real_t *)malloc((length - 1) * sizeof(real_t)); for (int x = 0; x < length; ++x){ for (int y = 0; y < x; ++y){ passed[y] = remaining_combinations[y]; } for (int y = x + 1; y < length; ++y){ passed[y - 1] = remaining_combinations[y]; } if (possible_result(passed, length - 1, remaining_combinations[x] - current_value, test_value)){ free(passed); return true; } } free(passed); return false; } #ifndef T1 //T1:atomic,construct-independent,V:2.0-2.7 int test1(){ int err = 0; srand(SEED); real_t *a = (real_t *)malloc(n * sizeof(real_t)); real_t *b = (real_t *)malloc(n * sizeof(real_t)); real_t *totals = (real_t *)malloc(((n/10) + 1) * sizeof(real_t)); int indexer = 0; real_t * passed = (real_t *)malloc(10 * sizeof(real_t)); real_t *passed_a = (real_t *)malloc(10 * sizeof(real_t)); real_t *passed_b = (real_t *)malloc(10 * sizeof(real_t)); int passed_indexer; int absolute_indexer; for (int x = 0; x < n; ++x){ a[x] = rand() / (real_t)(RAND_MAX / 10); } for (int x = 0; x < (n/10) + 1; ++x){ totals[x] = 0; } #pragma acc data copyin(a[0:n]) copy(totals[0:(n/10) + 1]) copyout(b[0:n]) { #pragma acc parallel { #pragma acc loop for (int x = 0; x < n; ++x){ #pragma acc atomic capture b[x] = totals[x%((int) (n/10) + 1)] = a[x] - totals[x%((int) (n/10) + 1)]; } } } for (int x = 0; x < (n/10) + 1; ++x){ indexer = x; while (indexer < n){ passed[indexer/((int) (n/10) + 1)] = a[indexer]; indexer += (n/10) + 1; } if (!(possible_result(passed, 10, 0, totals[x]))){ err += 1; } } for (int x = 0; x < n/10 + 1; ++x){ for (passed_indexer = 0, absolute_indexer = x; absolute_indexer < n; passed_indexer++, absolute_indexer += n/10 + 1){ passed_a[passed_indexer] = a[absolute_indexer]; passed_b[passed_indexer] = b[absolute_indexer]; } if (!is_possible(passed_a, passed_b, passed_indexer, 0)){ err += 1; } break; } return err; } #endif int main(){ int failcode = 0; int failed; #ifndef T1 failed = 0; for (int x = 0; x < NUM_TEST_CALLS; ++x){ failed = failed + test1(); } if (failed != 0){ failcode = failcode + (1 << 0); } #endif return failcode; }
./SPECaccel/benchspec/ACCEL/351.palm/src/cpu_statistics.F90
SUBROUTINE cpu_statistics !--------------------------------------------------------------------------------! ! This file is part of PALM. ! ! PALM is free software: you can redistribute it and/or modify it under the terms ! of the GNU General Public License as published by the Free Software Foundation, ! either version 3 of the License, or (at your option) any later version. ! ! PALM is distributed in the hope that 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 ! PALM. If not, see <http://www.gnu.org/licenses/>. ! ! Copyright 1997-2012 Leibniz University Hannover !--------------------------------------------------------------------------------! ! ! Current revisions: ! ----------------- ! ! ! Former revisions: ! ----------------- ! $Id: cpu_statistics.f90 1112 2013-03-09 00:34:37Z raasch $ ! ! 1111 2013-03-08 23:54:10Z raasch ! output of grid point numbers and average CPU time per grid point and timestep ! ! 1092 2013-02-02 11:24:22Z raasch ! unused variables removed ! ! 1036 2012-10-22 13:43:42Z raasch ! code put under GPL (PALM 3.9) ! ! 1015 2012-09-27 09:23:24Z raasch ! output of accelerator board information ! ! 683 2011-02-09 14:25:15Z raasch ! output of handling of ghostpoint exchange ! ! 622 2010-12-10 08:08:13Z raasch ! output of handling of collective operations ! ! 222 2009-01-12 16:04:16Z letzel ! Bugfix for nonparallel execution ! ! 197 2008-09-16 15:29:03Z raasch ! Format adjustments in order to allow CPU# > 999, ! data are collected from PE0 in an ordered sequence which seems to avoid ! hanging of processes on SGI-ICE ! ! 82 2007-04-16 15:40:52Z raasch ! Preprocessor directives for old systems removed ! ! RCS Log replace by Id keyword, revision history cleaned up ! ! Revision 1.13 2006/04/26 12:10:51 raasch ! Output of number of threads per task, max = min in case of 1 PE ! ! Revision 1.1 1997/07/24 11:11:11 raasch ! Initial revision ! ! ! Description: ! ------------ ! Analysis and output of the cpu-times measured. All PE results are collected ! on PE0 in order to calculate the mean cpu-time over all PEs and other ! statistics. The output is sorted according to the amount of cpu-time consumed ! and output on PE0. !------------------------------------------------------------------------------! USE control_parameters USE cpulog USE indices, ONLY: nx, ny, nz USE pegrid IMPLICIT NONE INTEGER :: i, ii(1), iii, sender REAL :: average_cputime REAL, SAVE :: norm = 1.0 REAL, DIMENSION(:), ALLOCATABLE :: pe_max, pe_min, pe_rms, sum REAL, DIMENSION(:,:), ALLOCATABLE :: pe_log_points ! !-- Compute cpu-times in seconds log_point%mtime = log_point%mtime / norm log_point%sum = log_point%sum / norm log_point%vector = log_point%vector / norm WHERE ( log_point%counts /= 0 ) log_point%mean = log_point%sum / log_point%counts END WHERE ! !-- Collect cpu-times from all PEs and calculate statistics IF ( myid == 0 ) THEN ! !-- Allocate and initialize temporary arrays needed for statistics ALLOCATE( pe_max( SIZE( log_point ) ), pe_min( SIZE( log_point ) ), & pe_rms( SIZE( log_point ) ), & pe_log_points( SIZE( log_point ), 0:numprocs-1 ) ) pe_min = log_point%sum pe_max = log_point%sum ! need to be set in case of 1 PE pe_rms = 0.0 #if defined( __parallel ) ! !-- Receive data from all PEs DO i = 1, numprocs-1 CALL MPI_RECV( pe_max(1), SIZE( log_point ), MPI_REAL, & i, i, comm2d, status, ierr ) sender = status(MPI_SOURCE) pe_log_points(:,sender) = pe_max ENDDO pe_log_points(:,0) = log_point%sum ! Results from PE0 ! !-- Calculate mean of all PEs, store it on log_point%sum !-- and find minimum and maximum DO iii = 1, SIZE( log_point ) DO i = 1, numprocs-1 log_point(iii)%sum = log_point(iii)%sum + pe_log_points(iii,i) pe_min(iii) = MIN( pe_min(iii), pe_log_points(iii,i) ) pe_max(iii) = MAX( pe_max(iii), pe_log_points(iii,i) ) ENDDO log_point(iii)%sum = log_point(iii)%sum / numprocs ! !-- Calculate rms DO i = 0, numprocs-1 pe_rms(iii) = pe_rms(iii) + ( & pe_log_points(iii,i) - log_point(iii)%sum & )**2 ENDDO pe_rms(iii) = SQRT( pe_rms(iii) / numprocs ) ENDDO ELSE ! !-- Send data to PE0 (pe_max is used as temporary storage to send !-- the data in order to avoid sending the data type log) ALLOCATE( pe_max( SIZE( log_point ) ) ) pe_max = log_point%sum CALL MPI_SEND( pe_max(1), SIZE( log_point ), MPI_REAL, 0, myid, comm2d, & ierr ) #endif ENDIF ! !-- Write cpu-times IF ( myid == 0 ) THEN ! !-- Re-store sums ALLOCATE( sum( SIZE( log_point ) ) ) WHERE ( log_point%counts /= 0 ) sum = log_point%sum ELSEWHERE sum = -1.0 ENDWHERE ! !-- Get total time in order to calculate CPU-time per gridpoint and timestep IF ( nr_timesteps_this_run /= 0 ) THEN average_cputime = log_point(1)%sum / REAL( (nx+1) * (ny+1) * nz ) / & REAL( nr_timesteps_this_run ) * 1E6 ! in micro-sec ELSE average_cputime = -1.0 ENDIF ! !-- Write cpu-times sorted by size CALL check_open( 18 ) #if defined( __parallel ) WRITE ( 18, 100 ) TRIM( run_description_header ), & numprocs * threads_per_task, pdims(1), pdims(2), & threads_per_task, nx+1, ny+1, nz, nr_timesteps_this_run, & average_cputime IF ( num_acc_per_node /= 0 ) WRITE ( 18, 108 ) num_acc_per_node WRITE ( 18, 110 ) #else WRITE ( 18, 100 ) TRIM( run_description_header ), & numprocs * threads_per_task, 1, 1, & threads_per_task, nx+1, ny+1, nz, nr_timesteps_this_run, & average_cputime IF ( num_acc_per_node /= 0 ) WRITE ( 18, 109 ) num_acc_per_node WRITE ( 18, 110 ) #endif DO ii = MAXLOC( sum ) i = ii(1) IF ( sum(i) /= -1.0 ) THEN WRITE ( 18, 102 ) & log_point(i)%place, log_point(i)%sum, & log_point(i)%sum / log_point(1)%sum * 100.0, & log_point(i)%counts, pe_min(i), pe_max(i), pe_rms(i) sum(i) = -1.0 ELSE EXIT ENDIF ENDDO ENDIF ! !-- The same procedure again for the individual measurements. ! !-- Compute cpu-times in seconds log_point_s%mtime = log_point_s%mtime / norm log_point_s%sum = log_point_s%sum / norm log_point_s%vector = log_point_s%vector / norm WHERE ( log_point_s%counts /= 0 ) log_point_s%mean = log_point_s%sum / log_point_s%counts END WHERE ! !-- Collect cpu-times from all PEs and calculate statistics #if defined( __parallel ) ! !-- Set barrier in order to avoid that PE0 receives log_point_s-data !-- while still busy with receiving log_point-data (see above) CALL MPI_BARRIER( comm2d, ierr ) #endif IF ( myid == 0 ) THEN ! !-- Initialize temporary arrays needed for statistics pe_min = log_point_s%sum pe_max = log_point_s%sum ! need to be set in case of 1 PE pe_rms = 0.0 #if defined( __parallel ) ! !-- Receive data from all PEs DO i = 1, numprocs-1 CALL MPI_RECV( pe_max(1), SIZE( log_point ), MPI_REAL, & MPI_ANY_SOURCE, MPI_ANY_TAG, comm2d, status, ierr ) sender = status(MPI_SOURCE) pe_log_points(:,sender) = pe_max ENDDO pe_log_points(:,0) = log_point_s%sum ! Results from PE0 ! !-- Calculate mean of all PEs, store it on log_point_s%sum !-- and find minimum and maximum DO iii = 1, SIZE( log_point ) DO i = 1, numprocs-1 log_point_s(iii)%sum = log_point_s(iii)%sum + pe_log_points(iii,i) pe_min(iii) = MIN( pe_min(iii), pe_log_points(iii,i) ) pe_max(iii) = MAX( pe_max(iii), pe_log_points(iii,i) ) ENDDO log_point_s(iii)%sum = log_point_s(iii)%sum / numprocs ! !-- Calculate rms DO i = 0, numprocs-1 pe_rms(iii) = pe_rms(iii) + ( & pe_log_points(iii,i) - log_point_s(iii)%sum & )**2 ENDDO pe_rms(iii) = SQRT( pe_rms(iii) / numprocs ) ENDDO ELSE ! !-- Send data to PE0 (pe_max is used as temporary storage to send !-- the data in order to avoid sending the data type log) pe_max = log_point_s%sum CALL MPI_SEND( pe_max(1), SIZE( log_point ), MPI_REAL, 0, 0, comm2d, & ierr ) #endif ENDIF ! !-- Write cpu-times IF ( myid == 0 ) THEN ! !-- Re-store sums WHERE ( log_point_s%counts /= 0 ) sum = log_point_s%sum ELSEWHERE sum = -1.0 ENDWHERE ! !-- Write cpu-times sorted by size WRITE ( 18, 101 ) DO ii = MAXLOC( sum ) i = ii(1) IF ( sum(i) /= -1.0 ) THEN WRITE ( 18, 102 ) & log_point_s(i)%place, log_point_s(i)%sum, & log_point_s(i)%sum / log_point(1)%sum * 100.0, & log_point_s(i)%counts, pe_min(i), pe_max(i), pe_rms(i) sum(i) = -1.0 ELSE EXIT ENDIF ENDDO ! !-- Output of handling of MPI operations IF ( collective_wait ) THEN WRITE ( 18, 103 ) ELSE WRITE ( 18, 104 ) ENDIF IF ( synchronous_exchange ) THEN WRITE ( 18, 105 ) ELSE WRITE ( 18, 106 ) ENDIF ! !-- Empty lines in order to create a gap to the results of the model !-- continuation runs WRITE ( 18, 107 ) ! !-- Unit 18 is not needed anymore CALL close_file( 18 ) ENDIF 100 FORMAT (A/11('-')//'CPU measures for ',I5,' PEs (',I5,'(x) * ',I5,'(y', & &') tasks *',I5,' threads):'// & 'gridpoints (x/y/z): ',20X,I5,' * ',I5,' * ',I5/ & 'nr of timesteps: ',22X,I6/ & 'cpu time per grid point and timestep: ',5X,F8.5,' * 10**-6 s') 101 FORMAT (/'special measures:'/ & &'-----------------------------------------------------------', & &'--------------------') 102 FORMAT (A20,2X,F9.3,2X,F7.2,1X,I7,3(1X,F9.3)) 103 FORMAT (/'Barriers are set in front of collective operations') 104 FORMAT (/'No barriers are set in front of collective operations') 105 FORMAT (/'Exchange of ghostpoints via MPI_SENDRCV') 106 FORMAT (/'Exchange of ghostpoints via MPI_ISEND/MPI_IRECV') 107 FORMAT (//) 108 FORMAT ('Accelerator boards per node: ',14X,I2) 109 FORMAT ('Accelerator boards: ',23X,I2) 110 FORMAT ('----------------------------------------------------------', & &'------------'//& &'place: mean counts min ', & &' max rms'/ & &' sec. % sec. ', & &' sec. sec.'/ & &'-----------------------------------------------------------', & &'-------------------') END SUBROUTINE cpu_statistics
./openacc-vv/set_device_type_num_nvidia.F90
#ifndef T1 !T1:runtime,construct-independent,internal-control-values,set,nonvalidating,V:2.5-2.7 LOGICAL FUNCTION test1() USE OPENACC IMPLICIT NONE INCLUDE "acc_testsuite.Fh" INTEGER :: device_num INTEGER :: device_type INTEGER :: errors = 0 device_type = acc_get_device_type() device_num = acc_get_device_num(device_type) !$acc set device_type(nvidia) device_num(device_num) IF (errors .eq. 0) THEN test1 = .FALSE. ELSE test1 = .TRUE. END IF END #endif PROGRAM main IMPLICIT NONE INTEGER :: failcode, testrun LOGICAL :: failed INCLUDE "acc_testsuite.Fh" #ifndef T1 LOGICAL :: test1 #endif failed = .FALSE. failcode = 0 #ifndef T1 DO testrun = 1, NUM_TEST_CALLS failed = failed .or. test1() END DO IF (failed) THEN failcode = failcode + 2 ** 0 failed = .FALSE. END IF #endif CALL EXIT (failcode) END PROGRAM
End of preview. Expand in Data Studio

Dataset Card for "llm4vv-test-split"

More Information needed

Downloads last month
6