Merging the new branch

This commit is contained in:
Aditya Ujeniya 2024-07-08 09:52:11 +02:00
parent 28fec03be9
commit d5053b96ea
433 changed files with 5948 additions and 4848038 deletions

View File

@ -1,3 +1,3 @@
CompileFlags: # Tweak the parse settings
Add: [-I/usr/local/include] # treat all files as C++, enable more warnings
CompileFlags:
Add: [-I/usr/local/include, -I/opt/homebrew/include, -D_MPI]
Compiler: clang

View File

@ -1,48 +0,0 @@
# C source skeleton
## Build
1. Configure the toolchain and additional options in `config.mk`:
```
# Supported: GCC, CLANG, ICC
TAG ?= GCC
ENABLE_OPENMP ?= false
OPTIONS += -DARRAY_ALIGNMENT=64
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER
```
The verbosity options enable detailed output about affinity settings, allocation sizes and timer resolution.
2. Build with:
```
make
```
You can build multiple toolchains in the same directory, but notice that the Makefile is only acting on the one currently set.
Intermediate build results are located in the `<TOOLCHAIN>` directory.
To output the executed commands use:
```
make Q=
```
3. Clean up with:
```
make clean
```
to clean intermediate build results.
```
make distclean
```
to clean intermediate build results and binary.
4. (Optional) Generate assembler:
```
make asm
```
The assembler files will also be located in the `<TOOLCHAIN>` directory.

View File

@ -1,47 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name canal # name of flow setup
bcLeft 3 # flags for boundary conditions
bcRight 3 # 1 = no-slip 3 = outflow
bcBottom 1 # 2 = free-slip 4 = periodic
bcTop 1 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 100.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 4.0 # domain size in y-direction
imax 200 # number of interior cells in x-direction
jmax 50 # number of interior cells in y-direction
# Time Data:
# ---------
te 100.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.00001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
#===============================================================================

View File

@ -1,47 +0,0 @@
#==============================================================================
# Driven Cavity
#==============================================================================
# Problem specific Data:
# ---------------------
name dcavity # name of flow setup
bcLeft 1 # flags for boundary conditions
bcRight 1 # 1 = no-slip 3 = outflow
bcBottom 1 # 2 = free-slip 4 = periodic
bcTop 1 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 500.0 # Reynolds number
u_init 0.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 1.0 # domain size in x-direction
ylength 1.0 # domain size in y-direction
imax 100 # number of interior cells in x-direction
jmax 100 # number of interior cells in y-direction
# Time Data:
# ---------
te 25.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 1000 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.5
omg 1.7 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
#===============================================================================

View File

@ -1,16 +0,0 @@
CC = mpicc
GCC = cc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
#OPENMP = -Xpreprocessor -fopenmp #required on Macos with homebrew libomp
LIBS = # -lomp
endif
VERSION = --version
CFLAGS = -Ofast -std=c99 $(OPENMP)
#CFLAGS = -Ofast -fnt-store=aggressive -std=c99 $(OPENMP) #AMD CLANG
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE# -DDEBUG
INCLUDES = -I/usr/local/include

View File

@ -1,14 +0,0 @@
CC = mpiicc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -qopenmp
endif
VERSION = --version
CFLAGS = -O3 -xHost -qopt-zmm-usage=high -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

File diff suppressed because it is too large Load Diff

View File

@ -1,61 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifdef __linux__
#ifdef _OPENMP
#include <pthread.h>
#include <sched.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <unistd.h>
#define MAX_NUM_THREADS 128
#define gettid() syscall(SYS_gettid)
static int getProcessorID(cpu_set_t* cpu_set)
{
int processorId;
for (processorId = 0; processorId < MAX_NUM_THREADS; processorId++) {
if (CPU_ISSET(processorId, cpu_set)) {
break;
}
}
return processorId;
}
int affinity_getProcessorId()
{
cpu_set_t cpu_set;
CPU_ZERO(&cpu_set);
sched_getaffinity(gettid(), sizeof(cpu_set_t), &cpu_set);
return getProcessorID(&cpu_set);
}
void affinity_pinThread(int processorId)
{
cpu_set_t cpuset;
pthread_t thread;
thread = pthread_self();
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
pthread_setaffinity_np(thread, sizeof(cpu_set_t), &cpuset);
}
void affinity_pinProcess(int processorId)
{
cpu_set_t cpuset;
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
sched_setaffinity(0, sizeof(cpu_set_t), &cpuset);
}
#endif /*_OPENMP*/
#endif /*__linux__*/

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef AFFINITY_H
#define AFFINITY_H
extern int affinity_getProcessorId();
extern void affinity_pinProcess(int);
extern void affinity_pinThread(int);
#endif /*AFFINITY_H*/

View File

@ -1,35 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
void* allocate(int alignment, size_t bytesize)
{
int errorCode;
void* ptr;
errorCode = posix_memalign(&ptr, alignment, bytesize);
if (errorCode) {
if (errorCode == EINVAL) {
fprintf(stderr, "Error: Alignment parameter is not a power of two\n");
exit(EXIT_FAILURE);
}
if (errorCode == ENOMEM) {
fprintf(stderr, "Error: Insufficient memory to fulfill the request\n");
exit(EXIT_FAILURE);
}
}
if (ptr == NULL) {
fprintf(stderr, "Error: posix_memalign failed!\n");
exit(EXIT_FAILURE);
}
return ptr;
}

View File

@ -1,13 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __ALLOCATE_H_
#define __ALLOCATE_H_
#include <stdlib.h>
extern void* allocate(int alignment, size_t bytesize);
#endif

View File

@ -1,90 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <float.h>
#include <limits.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include "parameter.h"
#include "progress.h"
#include "solver.h"
#include "timing.h"
enum VARIANT { SOR = 1, RB, RBA };
int main (int argc, char** argv)
{
int rank;
int variant = RB;
double S, E;
Parameter params;
Solver solver;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
initParameter(&params);
if (argc < 2) {
printf("Usage: %s <configFile>\n", argv[0]);
exit(EXIT_SUCCESS);
}
readParameter(&params, argv[1]);
if (argc == 3)
{
variant = atoi(argv[2]);
}
if (rank == 0) {
printParameter(&params);
}
initSolver(&solver, &params);
initProgress(solver.te);
void (*solver_generic[])(solver) = {solve, solveRB, solveRBA};
double tau = solver.tau;
double te = solver.te;
double t = 0.0;
S = getTimeStamp();
while (t <= te)
{
if (tau > 0.0) {
computeTimestep(&solver);
}
setBoundaryConditions(&solver);
setSpecialBoundaryCondition(&solver);
computeFG(&solver);
computeRHS(&solver);
(*solver_generic[variant - 1])(&solver);
adaptUV(&solver);
t += solver.dt;
#ifdef VERBOSE
if (rank == 0) {
printf("TIME %f , TIMESTEP %f\n", t, solver.dt);
}
#else
printProgress(t);
#endif
}
E = getTimeStamp();
stopProgress();
if (rank == 0) {
printf("Solution took %.2fs\n", E - S);
}
collectResult(&solver);
MPI_Finalize();
return EXIT_SUCCESS;
}

View File

@ -1,115 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "parameter.h"
#include "util.h"
#define MAXLINE 4096
void initParameter(Parameter* param)
{
param->xlength = 1.0;
param->ylength = 1.0;
param->imax = 100;
param->jmax = 100;
param->itermax = 1000;
param->eps = 0.0001;
param->omg = 1.7;
param->re = 100.0;
param->gamma = 0.9;
param->tau = 0.5;
param->rho = 0.99;
}
void readParameter(Parameter* param, const char* filename)
{
FILE* fp = fopen(filename, "r");
char line[MAXLINE];
int i;
if (!fp) {
fprintf(stderr, "Could not open parameter file: %s\n", filename);
exit(EXIT_FAILURE);
}
while (!feof(fp)) {
line[0] = '\0';
fgets(line, MAXLINE, fp);
for (i = 0; line[i] != '\0' && line[i] != '#'; i++)
;
line[i] = '\0';
char* tok = strtok(line, " ");
char* val = strtok(NULL, " ");
#define PARSE_PARAM(p, f) \
if (strncmp(tok, #p, sizeof(#p) / sizeof(#p[0]) - 1) == 0) { \
param->p = f(val); \
}
#define PARSE_STRING(p) PARSE_PARAM(p, strdup)
#define PARSE_INT(p) PARSE_PARAM(p, atoi)
#define PARSE_REAL(p) PARSE_PARAM(p, atof)
if (tok != NULL && val != NULL) {
PARSE_REAL(xlength);
PARSE_REAL(ylength);
PARSE_INT(imax);
PARSE_INT(jmax);
PARSE_INT(itermax);
PARSE_REAL(eps);
PARSE_REAL(omg);
PARSE_REAL(re);
PARSE_REAL(tau);
PARSE_REAL(gamma);
PARSE_REAL(dt);
PARSE_REAL(te);
PARSE_REAL(gx);
PARSE_REAL(gy);
PARSE_STRING(name);
PARSE_INT(bcLeft);
PARSE_INT(bcRight);
PARSE_INT(bcBottom);
PARSE_INT(bcTop);
PARSE_REAL(u_init);
PARSE_REAL(v_init);
PARSE_REAL(p_init);
PARSE_REAL(rho);
}
}
fclose(fp);
}
void printParameter(Parameter* param)
{
printf("Parameters for %s\n", param->name);
printf("Boundary conditions Left:%d Right:%d Bottom:%d Top:%d\n",
param->bcLeft,
param->bcRight,
param->bcBottom,
param->bcTop);
printf("\tReynolds number: %.2f\n", param->re);
printf("\tInit arrays: U:%.2f V:%.2f P:%.2f\n",
param->u_init,
param->v_init,
param->p_init);
printf("Geometry data:\n");
printf("\tDomain box size (x, y): %.2f, %.2f\n", param->xlength, param->ylength);
printf("\tCells (x, y): %d, %d\n", param->imax, param->jmax);
printf("Timestep parameters:\n");
printf("\tDefault stepsize: %.2f, Final time %.2f\n", param->dt, param->te);
printf("\tTau factor: %.2f\n", param->tau);
printf("Iterative solver parameters:\n");
printf("\tMax iterations: %d\n", param->itermax);
printf("\tepsilon (stopping tolerance) : %f\n", param->eps);
printf("\tgamma (stopping tolerance) : %f\n", param->gamma);
printf("\tomega (SOR relaxation): %f\n", param->omg);
}

View File

@ -1,26 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __PARAMETER_H_
#define __PARAMETER_H_
typedef struct {
double xlength, ylength;
int imax, jmax;
int itermax;
double eps, omg, rho;
double re, tau, gamma;
double te, dt;
double gx, gy;
char* name;
int bcLeft, bcRight, bcBottom, bcTop;
double u_init, v_init, p_init;
} Parameter;
void initParameter(Parameter*);
void readParameter(Parameter*, const char*);
void printParameter(Parameter*);
#endif

View File

@ -1,60 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <math.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "progress.h"
static double _end;
static int _current;
static int _rank = -1;
void initProgress(double end)
{
MPI_Comm_rank(MPI_COMM_WORLD, &_rank);
_end = end;
_current = 0;
if (_rank == 0) {
printf("[ ]");
fflush(stdout);
}
}
void printProgress(double current)
{
if (_rank == 0) {
int new = (int)rint((current / _end) * 10.0);
if (new > _current) {
char progress[11];
_current = new;
progress[0] = 0;
for (int i = 0; i < 10; i++) {
if (i < _current) {
sprintf(progress + strlen(progress), "#");
} else {
sprintf(progress + strlen(progress), " ");
}
}
printf("\r[%s]", progress);
}
fflush(stdout);
}
}
void stopProgress()
{
if (_rank == 0) {
printf("\n");
fflush(stdout);
}
}

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __PROGRESS_H_
#define __PROGRESS_H_
extern void initProgress(double);
extern void printProgress(double);
extern void stopProgress();
#endif

View File

@ -1,850 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <float.h>
#include <math.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "allocate.h"
#include "parameter.h"
#include "solver.h"
#include "util.h"
#define P(i, j) p[(j) * (imax + 2) + (i)]
#define F(i, j) f[(j) * (imax + 2) + (i)]
#define G(i, j) g[(j) * (imax + 2) + (i)]
#define U(i, j) u[(j) * (imax + 2) + (i)]
#define V(i, j) v[(j) * (imax + 2) + (i)]
#define RHS(i, j) rhs[(j) * (imax + 2) + (i)]
static int sizeOfRank(int rank, int size, int N)
{
return N / size + ((N % size > rank) ? 1 : 0);
}
static void print(Solver* solver, double* grid)
{
int imax = solver->imax;
for (int i = 0; i < solver->size; i++) {
if (i == solver->rank) {
printf("### RANK %d "
"#######################################################\n",
solver->rank);
for (int j = 0; j < solver->jmaxLocal + 2; j++) {
printf("%02d: ", j);
for (int i = 0; i < solver->imax + 2; i++) {
printf("%12.8f ", grid[j * (imax + 2) + i]);
}
printf("\n");
}
fflush(stdout);
}
MPI_Barrier(MPI_COMM_WORLD);
}
}
static void exchange(Solver* solver, double* grid)
{
MPI_Request requests[4] = { MPI_REQUEST_NULL,
MPI_REQUEST_NULL,
MPI_REQUEST_NULL,
MPI_REQUEST_NULL };
/* exchange ghost cells with top neighbor */
if (solver->rank + 1 < solver->size) {
int top = solver->rank + 1;
double* src = grid + (solver->jmaxLocal) * (solver->imax + 2) + 1;
double* dst = grid + (solver->jmaxLocal + 1) * (solver->imax + 2) + 1;
MPI_Isend(src, solver->imax, MPI_DOUBLE, top, 1, MPI_COMM_WORLD, &requests[0]);
MPI_Irecv(dst, solver->imax, MPI_DOUBLE, top, 2, MPI_COMM_WORLD, &requests[1]);
}
/* exchange ghost cells with bottom neighbor */
if (solver->rank > 0) {
int bottom = solver->rank - 1;
double* src = grid + (solver->imax + 2) + 1;
double* dst = grid + 1;
MPI_Isend(src, solver->imax, MPI_DOUBLE, bottom, 2, MPI_COMM_WORLD, &requests[2]);
MPI_Irecv(dst, solver->imax, MPI_DOUBLE, bottom, 1, MPI_COMM_WORLD, &requests[3]);
}
MPI_Waitall(4, requests, MPI_STATUSES_IGNORE);
}
static void shift(Solver* solver)
{
MPI_Request requests[2] = { MPI_REQUEST_NULL, MPI_REQUEST_NULL };
double* g = solver->g;
/* shift G */
/* receive ghost cells from bottom neighbor */
if (solver->rank > 0) {
int bottom = solver->rank - 1;
MPI_Irecv(g + 1,
solver->imax,
MPI_DOUBLE,
bottom,
0,
MPI_COMM_WORLD,
&requests[0]);
}
if (solver->rank + 1 < solver->size) {
int top = solver->rank + 1;
double* buf = g + (solver->jmaxLocal) * (solver->imax + 2) + 1;
/* send ghost cells to top neighbor */
MPI_Isend(buf, solver->imax, MPI_DOUBLE, top, 0, MPI_COMM_WORLD, &requests[1]);
}
MPI_Waitall(2, requests, MPI_STATUSES_IGNORE);
}
void collectResult(Solver* solver)
{
double* Pall = NULL;
double* Uall = NULL;
double* Vall = NULL;
int *rcvCounts, *displs;
if (solver->rank == 0) {
Pall = allocate(64, (solver->imax + 2) * (solver->jmax + 2) * sizeof(double));
Uall = allocate(64, (solver->imax + 2) * (solver->jmax + 2) * sizeof(double));
Vall = allocate(64, (solver->imax + 2) * (solver->jmax + 2) * sizeof(double));
rcvCounts = (int*)malloc(solver->size * sizeof(int));
displs = (int*)malloc(solver->size * sizeof(int));
rcvCounts[0] = solver->jmaxLocal * (solver->imax + 2);
displs[0] = 0;
int cursor = rcvCounts[0];
for (int i = 1; i < solver->size; i++) {
rcvCounts[i] = sizeOfRank(i, solver->size, solver->jmax) * (solver->imax + 2);
displs[i] = cursor;
cursor += rcvCounts[i];
}
}
int cnt = solver->jmaxLocal * (solver->imax + 2);
double* sendbuffer = solver->p + (solver->imax + 2);
MPI_Gatherv(sendbuffer,
cnt,
MPI_DOUBLE,
Pall,
rcvCounts,
displs,
MPI_DOUBLE,
0,
MPI_COMM_WORLD);
sendbuffer = solver->u + (solver->imax + 2);
MPI_Gatherv(sendbuffer,
cnt,
MPI_DOUBLE,
Uall,
rcvCounts,
displs,
MPI_DOUBLE,
0,
MPI_COMM_WORLD);
sendbuffer = solver->v + (solver->imax + 2);
MPI_Gatherv(sendbuffer,
cnt,
MPI_DOUBLE,
Vall,
rcvCounts,
displs,
MPI_DOUBLE,
0,
MPI_COMM_WORLD);
if (solver->rank == 0) {
writeResult(solver, Pall, Uall, Vall);
}
}
static void printConfig(Solver* solver)
{
if (solver->rank == 0) {
printf("Parameters for #%s#\n", solver->problem);
printf("Boundary conditions Left:%d Right:%d Bottom:%d Top:%d\n",
solver->bcLeft,
solver->bcRight,
solver->bcBottom,
solver->bcTop);
printf("\tReynolds number: %.2f\n", solver->re);
printf("\tGx Gy: %.2f %.2f\n", solver->gx, solver->gy);
printf("Geometry data:\n");
printf("\tDomain box size (x, y): %.2f, %.2f\n",
solver->xlength,
solver->ylength);
printf("\tCells (x, y): %d, %d\n", solver->imax, solver->jmax);
printf("Timestep parameters:\n");
printf("\tDefault stepsize: %.2f, Final time %.2f\n", solver->dt, solver->te);
printf("\tdt bound: %.6f\n", solver->dtBound);
printf("\tTau factor: %.2f\n", solver->tau);
printf("Iterative solver parameters:\n");
printf("\tMax iterations: %d\n", solver->itermax);
printf("\tepsilon (stopping tolerance) : %f\n", solver->eps);
printf("\tgamma factor: %f\n", solver->gamma);
printf("\tomega (SOR relaxation): %f\n", solver->omega);
printf("Communication parameters:\n");
}
for (int i = 0; i < solver->size; i++) {
if (i == solver->rank) {
printf("\tRank %d of %d\n", solver->rank, solver->size);
printf("\tLocal domain size: %dx%d\n", solver->imax, solver->jmaxLocal);
fflush(stdout);
}
}
}
void initSolver(Solver* solver, Parameter* params)
{
MPI_Comm_rank(MPI_COMM_WORLD, &(solver->rank));
MPI_Comm_size(MPI_COMM_WORLD, &(solver->size));
solver->problem = params->name;
solver->bcLeft = params->bcLeft;
solver->bcRight = params->bcRight;
solver->bcBottom = params->bcBottom;
solver->bcTop = params->bcTop;
solver->imax = params->imax;
solver->jmax = params->jmax;
solver->jmaxLocal = sizeOfRank(solver->rank, solver->size, solver->jmax);
solver->xlength = params->xlength;
solver->ylength = params->ylength;
solver->dx = params->xlength / params->imax;
solver->dy = params->ylength / params->jmax;
solver->eps = params->eps;
solver->omega = params->omg;
solver->itermax = params->itermax;
solver->re = params->re;
solver->gx = params->gx;
solver->gy = params->gy;
solver->dt = params->dt;
solver->te = params->te;
solver->tau = params->tau;
solver->gamma = params->gamma;
solver->rho = params->rho;
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
size_t bytesize = (imax + 2) * (jmaxLocal + 2) * sizeof(double);
solver->u = allocate(64, bytesize);
solver->v = allocate(64, bytesize);
solver->p = allocate(64, bytesize);
solver->rhs = allocate(64, bytesize);
solver->f = allocate(64, bytesize);
solver->g = allocate(64, bytesize);
for (int i = 0; i < (imax + 2) * (jmaxLocal + 2); i++) {
solver->u[i] = params->u_init;
solver->v[i] = params->v_init;
solver->p[i] = params->p_init;
solver->rhs[i] = 0.0;
solver->f[i] = 0.0;
solver->g[i] = 0.0;
}
double dx = solver->dx;
double dy = solver->dy;
double inv_sqr_sum = 1.0 / (dx * dx) + 1.0 / (dy * dy);
solver->dtBound = 0.5 * solver->re * 1.0 / inv_sqr_sum;
#ifdef VERBOSE
printConfig(solver);
#endif
}
void computeRHS(Solver* solver)
{
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
double idx = 1.0 / solver->dx;
double idy = 1.0 / solver->dy;
double idt = 1.0 / solver->dt;
double* rhs = solver->rhs;
double* f = solver->f;
double* g = solver->g;
shift(solver);
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = 1; i < imax + 1; i++) {
RHS(i, j) = ((F(i, j) - F(i - 1, j)) * idx + (G(i, j) - G(i, j - 1)) * idy) *
idt;
}
}
}
void solve(Solver* solver)
{
int imax = solver->imax;
int jmax = solver->jmax;
int jmaxLocal = solver->jmaxLocal;
double eps = solver->eps;
int itermax = solver->itermax;
double dx2 = solver->dx * solver->dx;
double dy2 = solver->dy * solver->dy;
double idx2 = 1.0 / dx2;
double idy2 = 1.0 / dy2;
double factor = solver->omega * 0.5 * (dx2 * dy2) / (dx2 + dy2);
double* p = solver->p;
double* rhs = solver->rhs;
double epssq = eps * eps;
int it = 0;
double res = 1.0;
while ((res >= epssq) && (it < itermax)) {
res = 0.0;
exchange(solver, p);
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = 1; i < imax + 1; i++) {
double r = RHS(i, j) -
((P(i + 1, j) - 2.0 * P(i, j) + P(i - 1, j)) * idx2 +
(P(i, j + 1) - 2.0 * P(i, j) + P(i, j - 1)) * idy2);
P(i, j) -= (factor * r);
res += (r * r);
}
}
if (solver->rank == 0) {
for (int i = 1; i < imax + 1; i++) {
P(i, 0) = P(i, 1);
}
}
if (solver->rank == (solver->size - 1)) {
for (int i = 1; i < imax + 1; i++) {
P(i, jmaxLocal + 1) = P(i, jmaxLocal);
}
}
for (int j = 1; j < jmaxLocal + 1; j++) {
P(0, j) = P(1, j);
P(imax + 1, j) = P(imax, j);
}
MPI_Allreduce(MPI_IN_PLACE, &res, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
res = res / (double)(imax * jmax);
#ifdef DEBUG
if (solver->rank == 0) {
printf("%d Residuum: %e\n", it, res);
}
#endif
it++;
}
#ifdef VERBOSE
if (solver->rank == 0) {
printf("Solver took %d iterations to reach %f\n", it, sqrt(res));
}
#endif
}
void solveRB(Solver* solver)
{
int imax = solver->imax;
int jmax = solver->jmax;
int jmaxLocal = solver->jmaxLocal;
double eps = solver->eps;
int itermax = solver->itermax;
double dx2 = solver->dx * solver->dx;
double dy2 = solver->dy * solver->dy;
double idx2 = 1.0 / dx2;
double idy2 = 1.0 / dy2;
double factor = solver->omega * 0.5 * (dx2 * dy2) / (dx2 + dy2);
double* p = solver->p;
double* rhs = solver->rhs;
double epssq = eps * eps;
int it = 0;
double res = 1.0;
int pass, jsw, isw;
while ((res >= epssq) && (it < itermax)) {
jsw = 1;
for (pass = 0; pass < 2; pass++) {
isw = jsw;
exchange(solver, p);
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = isw; i < imax + 1; i += 2) {
double r = RHS(i, j) -
((P(i + 1, j) - 2.0 * P(i, j) + P(i - 1, j)) * idx2 +
(P(i, j + 1) - 2.0 * P(i, j) + P(i, j - 1)) * idy2);
P(i, j) -= (factor * r);
res += (r * r);
}
isw = 3 - isw;
}
jsw = 3 - jsw;
}
if (solver->rank == 0) {
for (int i = 1; i < imax + 1; i++) {
P(i, 0) = P(i, 1);
}
}
if (solver->rank == (solver->size - 1)) {
for (int i = 1; i < imax + 1; i++) {
P(i, jmaxLocal + 1) = P(i, jmaxLocal);
}
}
for (int j = 1; j < jmaxLocal + 1; j++) {
P(0, j) = P(1, j);
P(imax + 1, j) = P(imax, j);
}
MPI_Allreduce(MPI_IN_PLACE, &res, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
res = res / (double)(imax * jmax);
#ifdef DEBUG
if (solver->rank == 0) {
printf("%d Residuum: %e\n", it, res);
}
#endif
it++;
}
#ifdef VERBOSE
if (solver->rank == 0) {
printf("Solver took %d iterations to reach %f\n", it, sqrt(res));
}
#endif
}
void solveRBA(Solver* solver)
{
int imax = solver->imax;
int jmax = solver->jmax;
int jmaxLocal = solver->jmaxLocal;
double eps = solver->eps;
int itermax = solver->itermax;
double dx2 = solver->dx * solver->dx;
double dy2 = solver->dy * solver->dy;
double idx2 = 1.0 / dx2;
double idy2 = 1.0 / dy2;
double factor = 0.5 * (dx2 * dy2) / (dx2 + dy2);
double* p = solver->p;
double* rhs = solver->rhs;
double rho = solver->rho;
double epssq = eps * eps;
int it = 0;
double res = 1.0;
double omega = 1.0;
int pass, jsw, isw;
while ((res >= epssq) && (it < itermax)) {
res = 0.0;
jsw = 1;
for (pass = 0; pass < 2; pass++) {
isw = jsw;
exchange(solver, p);
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = isw; i < imax + 1; i += 2) {
double r = RHS(i, j) -
((P(i + 1, j) - 2.0 * P(i, j) + P(i - 1, j)) * idx2 +
(P(i, j + 1) - 2.0 * P(i, j) + P(i, j - 1)) * idy2);
P(i, j) -= (omega * factor * r);
res += (r * r);
}
isw = 3 - isw;
}
jsw = 3 - jsw;
omega = (it == 0 && pass == 0 ? 1.0 / (1.0 - 0.5 * rho * rho)
: 1.0 / (1.0 - 0.25 * rho * rho * omega));
}
if (solver->rank == 0) {
for (int i = 1; i < imax + 1; i++) {
P(i, 0) = P(i, 1);
}
}
if (solver->rank == (solver->size - 1)) {
for (int i = 1; i < imax + 1; i++) {
P(i, jmaxLocal + 1) = P(i, jmaxLocal);
}
}
for (int j = 1; j < jmaxLocal + 1; j++) {
P(0, j) = P(1, j);
P(imax + 1, j) = P(imax, j);
}
MPI_Allreduce(MPI_IN_PLACE, &res, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
res = res / (double)(imax * jmax);
#ifdef DEBUG
if (solver->rank == 0) {
printf("%d Residuum: %e\n", it, res);
}
#endif
it++;
}
#ifdef VERBOSE
if (solver->rank == 0) {
printf("Solver took %d iterations to reach %f\n", it, sqrt(res));
}
#endif
}
static double maxElement(Solver* solver, double* m)
{
int size = (solver->imax + 2) * (solver->jmaxLocal + 2);
double maxval = DBL_MIN;
for (int i = 0; i < size; i++) {
maxval = MAX(maxval, fabs(m[i]));
}
MPI_Allreduce(MPI_IN_PLACE, &maxval, 1, MPI_DOUBLE, MPI_MAX, MPI_COMM_WORLD);
return maxval;
}
void normalizePressure(Solver* solver)
{
int size = (solver->imax + 2) * (solver->jmaxLocal + 2);
double* p = solver->p;
double avgP = 0.0;
for (int i = 0; i < size; i++) {
avgP += p[i];
}
MPI_Allreduce(MPI_IN_PLACE, &avgP, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
avgP /= (solver->imax + 2) * (solver->jmax + 2);
for (int i = 0; i < size; i++) {
p[i] = p[i] - avgP;
}
}
void computeTimestep(Solver* solver)
{
double dt = solver->dtBound;
double dx = solver->dx;
double dy = solver->dy;
double umax = maxElement(solver, solver->u);
double vmax = maxElement(solver, solver->v);
if (umax > 0) {
dt = (dt > dx / umax) ? dx / umax : dt;
}
if (vmax > 0) {
dt = (dt > dy / vmax) ? dy / vmax : dt;
}
solver->dt = dt * solver->tau;
}
void setBoundaryConditions(Solver* solver)
{
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
double* u = solver->u;
double* v = solver->v;
// Left boundary
switch (solver->bcLeft) {
case NOSLIP:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(0, j) = 0.0;
V(0, j) = -V(1, j);
}
break;
case SLIP:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(0, j) = 0.0;
V(0, j) = V(1, j);
}
break;
case OUTFLOW:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(0, j) = U(1, j);
V(0, j) = V(1, j);
}
break;
case PERIODIC:
break;
}
// Right boundary
switch (solver->bcRight) {
case NOSLIP:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(imax, j) = 0.0;
V(imax + 1, j) = -V(imax, j);
}
break;
case SLIP:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(imax, j) = 0.0;
V(imax + 1, j) = V(imax, j);
}
break;
case OUTFLOW:
for (int j = 1; j < jmaxLocal + 1; j++) {
U(imax, j) = U(imax - 1, j);
V(imax + 1, j) = V(imax, j);
}
break;
case PERIODIC:
break;
}
// Bottom boundary
if (solver->rank == 0) {
switch (solver->bcBottom) {
case NOSLIP:
for (int i = 1; i < imax + 1; i++) {
V(i, 0) = 0.0;
U(i, 0) = -U(i, 1);
}
break;
case SLIP:
for (int i = 1; i < imax + 1; i++) {
V(i, 0) = 0.0;
U(i, 0) = U(i, 1);
}
break;
case OUTFLOW:
for (int i = 1; i < imax + 1; i++) {
U(i, 0) = U(i, 1);
V(i, 0) = V(i, 1);
}
break;
case PERIODIC:
break;
}
}
// Top boundary
if (solver->rank == (solver->size - 1)) {
switch (solver->bcTop) {
case NOSLIP:
for (int i = 1; i < imax + 1; i++) {
V(i, jmaxLocal) = 0.0;
U(i, jmaxLocal + 1) = -U(i, jmaxLocal);
}
break;
case SLIP:
for (int i = 1; i < imax + 1; i++) {
V(i, jmaxLocal) = 0.0;
U(i, jmaxLocal + 1) = U(i, jmaxLocal);
}
break;
case OUTFLOW:
for (int i = 1; i < imax + 1; i++) {
U(i, jmaxLocal + 1) = U(i, jmaxLocal);
V(i, jmaxLocal) = V(i, jmaxLocal - 1);
}
break;
case PERIODIC:
break;
}
}
}
void setSpecialBoundaryCondition(Solver* solver)
{
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
double* u = solver->u;
if (strcmp(solver->problem, "dcavity") == 0) {
if (solver->rank == (solver->size - 1)) {
for (int i = 1; i < imax; i++) {
U(i, jmaxLocal + 1) = 2.0 - U(i, jmaxLocal);
}
}
} else if (strcmp(solver->problem, "canal") == 0) {
double ylength = solver->ylength;
double dy = solver->dy;
int rest = solver->jmax % solver->size;
int yc = solver->rank * (solver->jmax / solver->size) + MIN(rest, solver->rank);
double ys = dy * (yc + 0.5);
double y;
/* printf("RANK %d yc: %d ys: %f\n", solver->rank, yc, ys); */
for (int j = 1; j < jmaxLocal + 1; j++) {
y = ys + dy * (j - 0.5);
U(0, j) = y * (ylength - y) * 4.0 / (ylength * ylength);
}
}
/* print(solver, solver->u); */
}
void computeFG(Solver* solver)
{
double* u = solver->u;
double* v = solver->v;
double* f = solver->f;
double* g = solver->g;
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
double gx = solver->gx;
double gy = solver->gy;
double gamma = solver->gamma;
double dt = solver->dt;
double inverseRe = 1.0 / solver->re;
double inverseDx = 1.0 / solver->dx;
double inverseDy = 1.0 / solver->dy;
double du2dx, dv2dy, duvdx, duvdy;
double du2dx2, du2dy2, dv2dx2, dv2dy2;
exchange(solver, u);
exchange(solver, v);
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = 1; i < imax + 1; i++) {
du2dx = inverseDx * 0.25 *
((U(i, j) + U(i + 1, j)) * (U(i, j) + U(i + 1, j)) -
(U(i, j) + U(i - 1, j)) * (U(i, j) + U(i - 1, j))) +
gamma * inverseDx * 0.25 *
(fabs(U(i, j) + U(i + 1, j)) * (U(i, j) - U(i + 1, j)) +
fabs(U(i, j) + U(i - 1, j)) * (U(i, j) - U(i - 1, j)));
duvdy = inverseDy * 0.25 *
((V(i, j) + V(i + 1, j)) * (U(i, j) + U(i, j + 1)) -
(V(i, j - 1) + V(i + 1, j - 1)) * (U(i, j) + U(i, j - 1))) +
gamma * inverseDy * 0.25 *
(fabs(V(i, j) + V(i + 1, j)) * (U(i, j) - U(i, j + 1)) +
fabs(V(i, j - 1) + V(i + 1, j - 1)) *
(U(i, j) - U(i, j - 1)));
du2dx2 = inverseDx * inverseDx * (U(i + 1, j) - 2.0 * U(i, j) + U(i - 1, j));
du2dy2 = inverseDy * inverseDy * (U(i, j + 1) - 2.0 * U(i, j) + U(i, j - 1));
F(i, j) = U(i, j) + dt * (inverseRe * (du2dx2 + du2dy2) - du2dx - duvdy + gx);
duvdx = inverseDx * 0.25 *
((U(i, j) + U(i, j + 1)) * (V(i, j) + V(i + 1, j)) -
(U(i - 1, j) + U(i - 1, j + 1)) * (V(i, j) + V(i - 1, j))) +
gamma * inverseDx * 0.25 *
(fabs(U(i, j) + U(i, j + 1)) * (V(i, j) - V(i + 1, j)) +
fabs(U(i - 1, j) + U(i - 1, j + 1)) *
(V(i, j) - V(i - 1, j)));
dv2dy = inverseDy * 0.25 *
((V(i, j) + V(i, j + 1)) * (V(i, j) + V(i, j + 1)) -
(V(i, j) + V(i, j - 1)) * (V(i, j) + V(i, j - 1))) +
gamma * inverseDy * 0.25 *
(fabs(V(i, j) + V(i, j + 1)) * (V(i, j) - V(i, j + 1)) +
fabs(V(i, j) + V(i, j - 1)) * (V(i, j) - V(i, j - 1)));
dv2dx2 = inverseDx * inverseDx * (V(i + 1, j) - 2.0 * V(i, j) + V(i - 1, j));
dv2dy2 = inverseDy * inverseDy * (V(i, j + 1) - 2.0 * V(i, j) + V(i, j - 1));
G(i, j) = V(i, j) + dt * (inverseRe * (dv2dx2 + dv2dy2) - duvdx - dv2dy + gy);
}
}
/* ----------------------------- boundary of F ---------------------------
*/
for (int j = 1; j < jmaxLocal + 1; j++) {
F(0, j) = U(0, j);
F(imax, j) = U(imax, j);
}
/* ----------------------------- boundary of G ---------------------------
*/
if (solver->rank == 0) {
for (int i = 1; i < imax + 1; i++) {
G(i, 0) = V(i, 0);
}
}
if (solver->rank == (solver->size - 1)) {
for (int i = 1; i < imax + 1; i++) {
G(i, jmaxLocal) = V(i, jmaxLocal);
}
}
}
void adaptUV(Solver* solver)
{
int imax = solver->imax;
int jmaxLocal = solver->jmaxLocal;
double* p = solver->p;
double* u = solver->u;
double* v = solver->v;
double* f = solver->f;
double* g = solver->g;
double factorX = solver->dt / solver->dx;
double factorY = solver->dt / solver->dy;
for (int j = 1; j < jmaxLocal + 1; j++) {
for (int i = 1; i < imax + 1; i++) {
U(i, j) = F(i, j) - (P(i + 1, j) - P(i, j)) * factorX;
V(i, j) = G(i, j) - (P(i, j + 1) - P(i, j)) * factorY;
}
}
}
void writeResult(Solver* solver, double* p, double* u, double* v)
{
int imax = solver->imax;
int jmax = solver->jmax;
double dx = solver->dx;
double dy = solver->dy;
double x = 0.0, y = 0.0;
FILE* fp;
fp = fopen("pressure.dat", "w");
if (fp == NULL) {
printf("Error!\n");
exit(EXIT_FAILURE);
}
for (int j = 1; j < jmax + 1; j++) {
y = (double)(j - 0.5) * dy;
for (int i = 1; i < imax + 1; i++) {
x = (double)(i - 0.5) * dx;
fprintf(fp, "%.2f %.2f %f\n", x, y, P(i, j));
}
fprintf(fp, "\n");
}
fclose(fp);
fp = fopen("velocity.dat", "w");
if (fp == NULL) {
printf("Error!\n");
exit(EXIT_FAILURE);
}
for (int j = 1; j < jmax + 1; j++) {
y = dy * (j - 0.5);
for (int i = 1; i < imax + 1; i++) {
x = dx * (i - 0.5);
double vel_u = (U(i, j) + U(i - 1, j)) / 2.0;
double vel_v = (V(i, j) + V(i, j - 1)) / 2.0;
double len = sqrt((vel_u * vel_u) + (vel_v * vel_v));
fprintf(fp, "%.2f %.2f %f %f %f\n", x, y, vel_u, vel_v, len);
}
}
fclose(fp);
}

View File

@ -1,51 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __SOLVER_H_
#define __SOLVER_H_
#include "parameter.h"
enum BC { NOSLIP = 1, SLIP, OUTFLOW, PERIODIC };
typedef struct {
/* geometry and grid information */
double dx, dy;
int imax, jmax;
int jmaxLocal;
double xlength, ylength;
/* arrays */
double *p, *rhs;
double *f, *g;
double *u, *v;
/* parameters */
double eps, omega, rho;
double re, tau, gamma;
double gx, gy;
/* time stepping */
int itermax;
double dt, te;
double dtBound;
char* problem;
int bcLeft, bcRight, bcBottom, bcTop;
/* mpi */
int rank;
int size;
} Solver;
void initSolver(Solver*, Parameter*);
void computeRHS(Solver*);
void solve(Solver*);
void solveRB(Solver*);
void solveRBA(Solver*);
void normalizePressure(Solver*);
void computeTimestep(Solver*);
void setBoundaryConditions(Solver*);
void setSpecialBoundaryCondition(Solver*);
void computeFG(Solver*);
void adaptUV(Solver*);
void collectResult(Solver*);
void writeResult(Solver*, double*, double*, double*);
#endif

View File

@ -1,24 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <stdlib.h>
#include <time.h>
double getTimeStamp()
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeResolution()
{
struct timespec ts;
clock_getres(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeStamp_() { return getTimeStamp(); }

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __TIMING_H_
#define __TIMING_H_
extern double getTimeStamp();
extern double getTimeResolution();
extern double getTimeStamp_();
#endif // __TIMING_H_

View File

@ -1,23 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __UTIL_H_
#define __UTIL_H_
#define HLINE \
"------------------------------------------------------------------------" \
"----\n"
#ifndef MIN
#define MIN(x, y) ((x) < (y) ? (x) : (y))
#endif
#ifndef MAX
#define MAX(x, y) ((x) > (y) ? (x) : (y))
#endif
#ifndef ABS
#define ABS(a) ((a) >= 0 ? (a) : -(a))
#endif
#endif // __UTIL_H_

View File

@ -1,5 +0,0 @@
set terminal png size 1800,768 enhanced font ,12
set output 'velocity.png'
set datafile separator whitespace
plot 'velocity.dat' using 1:2:3:4:5 with vectors filled head size 0.01,20,60 lc palette

File diff suppressed because it is too large Load Diff

Binary file not shown.

Before

Width:  |  Height:  |  Size: 19 KiB

View File

@ -1,71 +0,0 @@
#=======================================================================================
# Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
# All rights reserved.
# Use of this source code is governed by a MIT-style
# license that can be found in the LICENSE file.
#=======================================================================================
#CONFIGURE BUILD SYSTEM
TARGET = exe-$(TAG)
BUILD_DIR = ./$(TAG)
SRC_DIR = ./src
MAKE_DIR = ./
Q ?= @
#DO NOT EDIT BELOW
include $(MAKE_DIR)/config.mk
include $(MAKE_DIR)/include_$(TAG).mk
INCLUDES += -I$(SRC_DIR) -I$(BUILD_DIR)
VPATH = $(SRC_DIR)
SRC = $(wildcard $(SRC_DIR)/*.c)
ASM = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.s, $(SRC))
OBJ = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.o, $(SRC))
SOURCES = $(SRC) $(wildcard $(SRC_DIR)/*.h)
CPPFLAGS := $(CPPFLAGS) $(DEFINES) $(OPTIONS) $(INCLUDES)
${TARGET}: $(BUILD_DIR) $(OBJ)
$(info ===> LINKING $(TARGET))
$(Q)${LINKER} ${LFLAGS} -o $(TARGET) $(OBJ) $(LIBS)
$(BUILD_DIR)/%.o: %.c $(MAKE_DIR)/include_$(TAG).mk $(MAKE_DIR)/config.mk
$(info ===> COMPILE $@)
$(CC) -c $(CPPFLAGS) $(CFLAGS) $< -o $@
$(Q)$(GCC) $(CPPFLAGS) -MT $(@:.d=.o) -MM $< > $(BUILD_DIR)/$*.d
$(BUILD_DIR)/%.s: %.c
$(info ===> GENERATE ASM $@)
$(CC) -S $(CPPFLAGS) $(CFLAGS) $< -o $@
.PHONY: clean distclean tags info asm format
clean:
$(info ===> CLEAN)
@rm -rf $(BUILD_DIR)
@rm -f tags
distclean: clean
$(info ===> DIST CLEAN)
@rm -f $(TARGET)
info:
$(info $(CFLAGS))
$(Q)$(CC) $(VERSION)
asm: $(BUILD_DIR) $(ASM)
tags:
$(info ===> GENERATE TAGS)
$(Q)ctags -R
format:
@for src in $(SOURCES) ; do \
echo "Formatting $$src" ; \
clang-format -i $$src ; \
done
@echo "Done"
$(BUILD_DIR):
@mkdir $(BUILD_DIR)
-include $(OBJ:.o=.d)

View File

@ -1,48 +0,0 @@
# C source skeleton
## Build
1. Configure the toolchain and additional options in `config.mk`:
```
# Supported: GCC, CLANG, ICC
TAG ?= GCC
ENABLE_OPENMP ?= false
OPTIONS += -DARRAY_ALIGNMENT=64
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER
```
The verbosity options enable detailed output about affinity settings, allocation sizes and timer resolution.
2. Build with:
```
make
```
You can build multiple toolchains in the same directory, but notice that the Makefile is only acting on the one currently set.
Intermediate build results are located in the `<TOOLCHAIN>` directory.
To output the executed commands use:
```
make Q=
```
3. Clean up with:
```
make clean
```
to clean intermediate build results.
```
make distclean
```
to clean intermediate build results and binary.
4. (Optional) Generate assembler:
```
make asm
```
The assembler files will also be located in the `<TOOLCHAIN>` directory.

View File

@ -1,47 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name canal # name of flow setup
bcN 1 # flags for boundary conditions
bcE 3 # 1 = no-slip 3 = outflow
bcS 1 # 2 = free-slip 4 = periodic
bcW 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 100.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 4.0 # domain size in y-direction
imax 200 # number of interior cells in x-direction
jmax 50 # number of interior cells in y-direction
# Time Data:
# ---------
te 100.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.00001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
#===============================================================================

View File

@ -1,10 +0,0 @@
# Supported: GCC, CLANG, ICC
TAG ?= ICC
ENABLE_OPENMP ?= false
#Feature options
OPTIONS += -DARRAY_ALIGNMENT=64
# OPTIONS += -DVERBOSE
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER

View File

@ -1,47 +0,0 @@
#==============================================================================
# Driven Cavity
#==============================================================================
# Problem specific Data:
# ---------------------
name dcavity # name of flow setup
bcN 1 # flags for boundary conditions
bcE 1 # 1 = no-slip 3 = outflow
bcS 1 # 2 = free-slip 4 = periodic
bcW 1 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 1000.0 # Reynolds number
u_init 0.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 1.0 # domain size in x-direction
ylength 1.0 # domain size in y-direction
imax 100 # number of interior cells in x-direction
jmax 100 # number of interior cells in y-direction
# Time Data:
# ---------
te 10.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 1000 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.5
omg 1.7 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
#===============================================================================

View File

@ -1,16 +0,0 @@
CC = mpicc
GCC = cc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
#OPENMP = -Xpreprocessor -fopenmp #required on Macos with homebrew libomp
LIBS = # -lomp
endif
VERSION = --version
CFLAGS = -Ofast -std=c99 $(OPENMP)
#CFLAGS = -Ofast -fnt-store=aggressive -std=c99 $(OPENMP) #AMD CLANG
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE# -DDEBUG
INCLUDES = -I/usr/local/include

View File

@ -1,14 +0,0 @@
CC = gcc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
endif
VERSION = --version
CFLAGS = -Ofast -ffreestanding -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

View File

@ -1,14 +0,0 @@
CC = mpiicc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -qopenmp
endif
VERSION = --version
CFLAGS = -O3 -xHost -qopt-zmm-usage=high -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

File diff suppressed because it is too large Load Diff

View File

@ -1,61 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifdef __linux__
#ifdef _OPENMP
#include <pthread.h>
#include <sched.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <unistd.h>
#define MAX_NUM_THREADS 128
#define gettid() syscall(SYS_gettid)
static int getProcessorID(cpu_set_t* cpu_set)
{
int processorId;
for (processorId = 0; processorId < MAX_NUM_THREADS; processorId++) {
if (CPU_ISSET(processorId, cpu_set)) {
break;
}
}
return processorId;
}
int affinity_getProcessorId()
{
cpu_set_t cpu_set;
CPU_ZERO(&cpu_set);
sched_getaffinity(gettid(), sizeof(cpu_set_t), &cpu_set);
return getProcessorID(&cpu_set);
}
void affinity_pinThread(int processorId)
{
cpu_set_t cpuset;
pthread_t thread;
thread = pthread_self();
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
pthread_setaffinity_np(thread, sizeof(cpu_set_t), &cpuset);
}
void affinity_pinProcess(int processorId)
{
cpu_set_t cpuset;
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
sched_setaffinity(0, sizeof(cpu_set_t), &cpuset);
}
#endif /*_OPENMP*/
#endif /*__linux__*/

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef AFFINITY_H
#define AFFINITY_H
extern int affinity_getProcessorId();
extern void affinity_pinProcess(int);
extern void affinity_pinThread(int);
#endif /*AFFINITY_H*/

View File

@ -1,35 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
void* allocate(int alignment, size_t bytesize)
{
int errorCode;
void* ptr;
errorCode = posix_memalign(&ptr, alignment, bytesize);
if (errorCode) {
if (errorCode == EINVAL) {
fprintf(stderr, "Error: Alignment parameter is not a power of two\n");
exit(EXIT_FAILURE);
}
if (errorCode == ENOMEM) {
fprintf(stderr, "Error: Insufficient memory to fulfill the request\n");
exit(EXIT_FAILURE);
}
}
if (ptr == NULL) {
fprintf(stderr, "Error: posix_memalign failed!\n");
exit(EXIT_FAILURE);
}
return ptr;
}

View File

@ -1,13 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __ALLOCATE_H_
#define __ALLOCATE_H_
#include <stdlib.h>
extern void* allocate(int alignment, size_t bytesize);
#endif

View File

@ -1,54 +0,0 @@
/*
* =======================================================================================
*
* Author: Jan Eitzinger (je), jan.eitzinger@fau.de
* Copyright (c) 2020 RRZE, University Erlangen-Nuremberg
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* =======================================================================================
*/
#ifndef LIKWID_MARKERS_H
#define LIKWID_MARKERS_H
#ifdef LIKWID_PERFMON
#include <likwid.h>
#define LIKWID_MARKER_INIT likwid_markerInit()
#define LIKWID_MARKER_THREADINIT likwid_markerThreadInit()
#define LIKWID_MARKER_SWITCH likwid_markerNextGroup()
#define LIKWID_MARKER_REGISTER(regionTag) likwid_markerRegisterRegion(regionTag)
#define LIKWID_MARKER_START(regionTag) likwid_markerStartRegion(regionTag)
#define LIKWID_MARKER_STOP(regionTag) likwid_markerStopRegion(regionTag)
#define LIKWID_MARKER_CLOSE likwid_markerClose()
#define LIKWID_MARKER_RESET(regionTag) likwid_markerResetRegion(regionTag)
#define LIKWID_MARKER_GET(regionTag, nevents, events, time, count) \
likwid_markerGetRegion(regionTag, nevents, events, time, count)
#else /* LIKWID_PERFMON */
#define LIKWID_MARKER_INIT
#define LIKWID_MARKER_THREADINIT
#define LIKWID_MARKER_SWITCH
#define LIKWID_MARKER_REGISTER(regionTag)
#define LIKWID_MARKER_START(regionTag)
#define LIKWID_MARKER_STOP(regionTag)
#define LIKWID_MARKER_CLOSE
#define LIKWID_MARKER_GET(regionTag, nevents, events, time, count)
#define LIKWID_MARKER_RESET(regionTag)
#endif /* LIKWID_PERFMON */
#endif /*LIKWID_MARKERS_H*/

View File

@ -1,91 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <float.h>
#include <limits.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include "parameter.h"
#include "progress.h"
#include "solver.h"
#include "timing.h"
enum VARIANT { SOR = 1, RB, RBA };
int main (int argc, char** argv)
{
int rank;
int variant = RB;
double S, E;
Parameter params;
Solver solver;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
initParameter(&params);
if (argc < 2) {
printf("Usage: %s <configFile>\n", argv[0]);
exit(EXIT_SUCCESS);
}
readParameter(&params, argv[1]);
if (argc == 3)
{
variant = atoi(argv[2]);
}
if (rank == 0) {
printParameter(&params);
}
initSolver(&solver, &params);
/* debugExchange(&solver); */
/* debugBC(&solver); */
/* exit(EXIT_SUCCESS); */
initProgress(solver.te);
double tau = solver.tau;
double te = solver.te;
double t = 0.0;
int (*solver_generic[])(solver) = {solve, solveRB, solveRBA};
S = getTimeStamp();
while (t <= te) {
if (tau > 0.0) {
computeTimestep(&solver);
}
setBoundaryConditions(&solver);
setSpecialBoundaryCondition(&solver);
computeFG(&solver);
computeRHS(&solver);
(*solver_generic[variant - 1])(&solver);
adaptUV(&solver);
t += solver.dt;
#ifdef VERBOSE
if (rank == 0) {
printf("TIME %f , TIMESTEP %f\n", t, solver.dt);
}
#else
printProgress(t);
#endif
}
E = getTimeStamp();
stopProgress();
if (rank == 0) {
printf("Solution took %.2fs\n", E - S);
}
collectResult(&solver);
MPI_Finalize();
return EXIT_SUCCESS;
}

View File

@ -1,112 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "parameter.h"
#include "util.h"
#define MAXLINE 4096
void initParameter(Parameter* param)
{
param->xlength = 1.0;
param->ylength = 1.0;
param->imax = 100;
param->jmax = 100;
param->itermax = 1000;
param->eps = 0.0001;
param->omg = 1.8;
param->rho = 0.99;
}
void readParameter(Parameter* param, const char* filename)
{
FILE* fp = fopen(filename, "r");
char line[MAXLINE];
int i;
if (!fp) {
fprintf(stderr, "Could not open parameter file: %s\n", filename);
exit(EXIT_FAILURE);
}
while (!feof(fp)) {
line[0] = '\0';
fgets(line, MAXLINE, fp);
for (i = 0; line[i] != '\0' && line[i] != '#'; i++)
;
line[i] = '\0';
char* tok = strtok(line, " ");
char* val = strtok(NULL, " ");
#define PARSE_PARAM(p, f) \
if (strncmp(tok, #p, sizeof(#p) / sizeof(#p[0]) - 1) == 0) { \
param->p = f(val); \
}
#define PARSE_STRING(p) PARSE_PARAM(p, strdup)
#define PARSE_INT(p) PARSE_PARAM(p, atoi)
#define PARSE_REAL(p) PARSE_PARAM(p, atof)
if (tok != NULL && val != NULL) {
PARSE_REAL(xlength);
PARSE_REAL(ylength);
PARSE_INT(imax);
PARSE_INT(jmax);
PARSE_INT(itermax);
PARSE_REAL(eps);
PARSE_REAL(omg);
PARSE_REAL(re);
PARSE_REAL(tau);
PARSE_REAL(gamma);
PARSE_REAL(dt);
PARSE_REAL(te);
PARSE_REAL(gx);
PARSE_REAL(gy);
PARSE_STRING(name);
PARSE_INT(bcN);
PARSE_INT(bcS);
PARSE_INT(bcE);
PARSE_INT(bcW);
PARSE_REAL(u_init);
PARSE_REAL(v_init);
PARSE_REAL(p_init);
PARSE_REAL(rho);
}
}
fclose(fp);
}
void printParameter(Parameter* param)
{
printf("Parameters for %s\n", param->name);
printf("Boundary conditions N:%d E:%d S:%d W:%d\n",
param->bcN,
param->bcE,
param->bcS,
param->bcW);
printf("\tReynolds number: %.2f\n", param->re);
printf("\tInit arrays: U:%.2f V:%.2f P:%.2f\n",
param->u_init,
param->v_init,
param->p_init);
printf("Geometry data:\n");
printf("\tDomain box size (x, y): %.2f, %.2f\n", param->xlength, param->ylength);
printf("\tCells (x, y): %d, %d\n", param->imax, param->jmax);
printf("Timestep parameters:\n");
printf("\tDefault stepsize: %.2f, Final time %.2f\n", param->dt, param->te);
printf("\tTau factor: %.2f\n", param->tau);
printf("Iterative solver parameters:\n");
printf("\tMax iterations: %d\n", param->itermax);
printf("\tepsilon (stopping tolerance) : %f\n", param->eps);
printf("\tgamma (stopping tolerance) : %f\n", param->gamma);
printf("\tomega (SOR relaxation): %f\n", param->omg);
}

View File

@ -1,26 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __PARAMETER_H_
#define __PARAMETER_H_
typedef struct {
double xlength, ylength;
int imax, jmax;
int itermax;
double eps, omg, rho;
double re, tau, gamma;
double te, dt;
double gx, gy;
char* name;
int bcN, bcS, bcE, bcW;
double u_init, v_init, p_init;
} Parameter;
void initParameter(Parameter*);
void readParameter(Parameter*, const char*);
void printParameter(Parameter*);
#endif

View File

@ -1,60 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <math.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "progress.h"
static double _end;
static int _current;
static int _rank = -1;
void initProgress(double end)
{
MPI_Comm_rank(MPI_COMM_WORLD, &_rank);
_end = end;
_current = 0;
if (_rank == 0) {
printf("[ ]");
fflush(stdout);
}
}
void printProgress(double current)
{
if (_rank == 0) {
int new = (int)rint((current / _end) * 10.0);
if (new > _current) {
char progress[11];
_current = new;
progress[0] = 0;
for (int i = 0; i < 10; i++) {
if (i < _current) {
sprintf(progress + strlen(progress), "#");
} else {
sprintf(progress + strlen(progress), " ");
}
}
printf("\r[%s]", progress);
}
fflush(stdout);
}
}
void stopProgress()
{
if (_rank == 0) {
printf("\n");
fflush(stdout);
}
}

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __PROGRESS_H_
#define __PROGRESS_H_
extern void initProgress(double);
extern void printProgress(double);
extern void stopProgress();
#endif

File diff suppressed because it is too large Load Diff

View File

@ -1,58 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __SOLVER_H_
#define __SOLVER_H_
#include "parameter.h"
#include <mpi.h>
enum BC { NOSLIP = 1, SLIP, OUTFLOW, PERIODIC };
typedef struct {
/* geometry and grid information */
double dx, dy;
int imax, jmax;
double xlength, ylength;
/* arrays */
double *p, *rhs;
double *f, *g;
double *u, *v;
/* parameters */
double eps, omega, rho;
double re, tau, gamma;
double gx, gy;
/* time stepping */
int itermax;
double dt, te;
double dtBound;
char* problem;
int bcN, bcS, bcW, bcE;
/* mpi */
int rank;
int size;
MPI_Comm comm;
MPI_Datatype iBufferType, jBufferType;
int iNeighbours[2], jNeighbours[2];
int coords[2], dims[2];
int imaxLocal, jmaxLocal;
} Solver;
void initSolver(Solver*, Parameter*);
void computeRHS(Solver*);
int solve(Solver*);
int solveRB(Solver*);
int solveRBA(Solver*);
void computeTimestep(Solver*);
void setBoundaryConditions(Solver*);
void setSpecialBoundaryCondition(Solver*);
void computeFG(Solver*);
void adaptUV(Solver*);
void collectResult(Solver*);
void writeResult(Solver*, double*, double*, double*);
void debugExchange(Solver*);
void debugBC(Solver*);
void print(Solver*, double*);
#endif

View File

@ -1,24 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <stdlib.h>
#include <time.h>
double getTimeStamp()
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeResolution()
{
struct timespec ts;
clock_getres(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeStamp_() { return getTimeStamp(); }

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __TIMING_H_
#define __TIMING_H_
extern double getTimeStamp();
extern double getTimeResolution();
extern double getTimeStamp_();
#endif // __TIMING_H_

View File

@ -1,22 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __UTIL_H_
#define __UTIL_H_
#define HLINE \
"----------------------------------------------------------------------------\n"
#ifndef MIN
#define MIN(x, y) ((x) < (y) ? (x) : (y))
#endif
#ifndef MAX
#define MAX(x, y) ((x) > (y) ? (x) : (y))
#endif
#ifndef ABS
#define ABS(a) ((a) >= 0 ? (a) : -(a))
#endif
#endif // __UTIL_H_

View File

@ -1,7 +0,0 @@
set terminal png size 1024,768 enhanced font ,12
set output 'p.png'
set datafile separator whitespace
set grid
set hidden3d
splot 'pressure.dat' using 1:2:3 with lines

View File

@ -1,5 +0,0 @@
set terminal png size 1800,768 enhanced font ,12
set output 'velocity.png'
set datafile separator whitespace
plot 'velocity.dat' using 1:2:3:4:5 with vectors filled head size 0.01,20,60 lc palette

File diff suppressed because it is too large Load Diff

Binary file not shown.

Before

Width:  |  Height:  |  Size: 184 KiB

View File

@ -1,76 +0,0 @@
#=======================================================================================
# Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
# All rights reserved.
# Use of this source code is governed by a MIT-style
# license that can be found in the LICENSE file.
#=======================================================================================
#CONFIGURE BUILD SYSTEM
TARGET = exe-$(TAG)
BUILD_DIR = ./$(TAG)
SRC_DIR = ./src
MAKE_DIR = ./
Q ?= @
#DO NOT EDIT BELOW
include $(MAKE_DIR)/config.mk
include $(MAKE_DIR)/include_$(TAG).mk
INCLUDES += -I$(SRC_DIR) -I$(BUILD_DIR)
VPATH = $(SRC_DIR)
SRC = $(wildcard $(SRC_DIR)/*.c)
ASM = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.s, $(SRC))
OBJ = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.o, $(SRC))
SOURCES = $(SRC) $(wildcard $(SRC_DIR)/*.h)
CPPFLAGS := $(CPPFLAGS) $(DEFINES) $(OPTIONS) $(INCLUDES)
${TARGET}: $(BUILD_DIR) $(OBJ)
$(info ===> LINKING $(TARGET))
$(Q)${LINKER} ${LFLAGS} -o $(TARGET) $(OBJ) $(LIBS)
$(BUILD_DIR)/%.o: %.c $(MAKE_DIR)/include_$(TAG).mk $(MAKE_DIR)/config.mk
$(info ===> COMPILE $@)
$(CC) -c $(CPPFLAGS) $(CFLAGS) $< -o $@
$(Q)$(GCC) $(CPPFLAGS) -MT $(@:.d=.o) -MM $< > $(BUILD_DIR)/$*.d
$(BUILD_DIR)/%.s: %.c
$(info ===> GENERATE ASM $@)
$(CC) -S $(CPPFLAGS) $(CFLAGS) $< -o $@
.PHONY: clean distclean tags info asm format
clean: vis
$(info ===> CLEAN)
@rm -rf $(BUILD_DIR)
@rm -f tags
distclean: clean
$(info ===> DIST CLEAN)
@rm -f $(TARGET)
info:
$(info $(CFLAGS))
$(Q)$(CC) $(VERSION)
asm: $(BUILD_DIR) $(ASM)
vis:
$(info ===> REMOVING VIZUALISATION FILES)
@rm -f vtk_files/particle*.vtk
@rm -f vis_files/particle*.dat
tags:
$(info ===> GENERATE TAGS)
$(Q)ctags -R
format:
@for src in $(SOURCES) ; do \
echo "Formatting $$src" ; \
clang-format -i $$src ; \
done
@echo "Done"
$(BUILD_DIR):
@mkdir $(BUILD_DIR)
-include $(OBJ:.o=.d)

View File

@ -1,48 +0,0 @@
# C source skeleton
## Build
1. Configure the toolchain and additional options in `config.mk`:
```
# Supported: GCC, CLANG, ICC
TAG ?= GCC
ENABLE_OPENMP ?= false
OPTIONS += -DARRAY_ALIGNMENT=64
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER
```
The verbosity options enable detailed output about affinity settings, allocation sizes and timer resolution.
2. Build with:
```
make
```
You can build multiple toolchains in the same directory, but notice that the Makefile is only acting on the one currently set.
Intermediate build results are located in the `<TOOLCHAIN>` directory.
To output the executed commands use:
```
make Q=
```
3. Clean up with:
```
make clean
```
to clean intermediate build results.
```
make distclean
```
to clean intermediate build results and binary.
4. (Optional) Generate assembler:
```
make asm
```
The assembler files will also be located in the `<TOOLCHAIN>` directory.

View File

@ -1,73 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name backstep # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 36500.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 1.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 7.0 # domain size in x-direction
ylength 1.5 # domain size in y-direction
imax 210 # number of interior cells in x-direction
jmax 45 # number of interior cells in y-direction
# Time Data:
# ---------
te 60.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.0001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
levels 5 # Multigrid levels
# Particle Tracing Data:
# -----------------------
numberOfParticles 200
startTime 0
injectTimePeriod 1.0
writeTimePeriod 0.5
x1 0.0
y1 0.5
x2 0.0
y2 1.5
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 1
xCenter 0.0
yCenter 0.0
xRectLength 2.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

View File

@ -1,73 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name canal # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 100.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 1.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 4.0 # domain size in y-direction
imax 1200 # number of interior cells in x-direction
jmax 160 # number of interior cells in y-direction
# Time Data:
# ---------
te 60.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.0001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
levels 5 # Multigrid levels
# Particle Tracing Data:
# -----------------------
numberOfParticles 60
startTime 500
injectTimePeriod 2.0
writeTimePeriod 0.5
x1 1.0
y1 0.0
x2 1.0
y2 4.0
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 0
xCenter 10.0
yCenter 2
xRectLength 6.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

View File

@ -1,10 +0,0 @@
# Supported: GCC, CLANG, ICC
TAG ?= ICC
ENABLE_OPENMP ?= false
#Feature options
OPTIONS += -DARRAY_ALIGNMENT=64
OPTIONS += -DVERBOSE
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER

View File

@ -1,61 +0,0 @@
#==============================================================================
# Driven Cavity
#==============================================================================
# Problem specific Data:
# ---------------------
name dcavity # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 1 # 2 = free-slip 4 = periodic
bcRight 1 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 10.0 # Reynolds number
u_init 0.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 1.0 # domain size in x-direction
ylength 1.0 # domain size in y-direction
imax 100 # number of interior cells in x-direction
jmax 100 # number of interior cells in y-direction
# Time Data:
# ---------
te 10.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 1000 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.5 #
omg 1.7 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
levels 5 # Multigrid levels
# Particle Tracing Data:
# -----------------------
numberOfParticles 30
startTime 2
injectTimePeriod 2.0
writeTimePeriod 0.1
x1 0.1
y1 0.5
x2 0.9
y2 0.5
#===============================================================================

View File

@ -1,16 +0,0 @@
CC = mpicc
GCC = cc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
#OPENMP = -Xpreprocessor -fopenmp #required on Macos with homebrew libomp
LIBS = # -lomp
endif
VERSION = --version
CFLAGS = -Ofast -std=c99 $(OPENMP)
#CFLAGS = -Ofast -fnt-store=aggressive -std=c99 $(OPENMP) #AMD CLANG
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE# -DDEBUG
INCLUDES = -I/usr/local/include

View File

@ -1,14 +0,0 @@
CC = gcc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
endif
VERSION = --version
CFLAGS = -Ofast -ffreestanding -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

View File

@ -1,14 +0,0 @@
CC = mpiicc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -qopenmp
endif
VERSION = --version
CFLAGS = -O3 -xHost -qopt-zmm-usage=high -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

View File

@ -1,73 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name karman # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 5050.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 8.0 # domain size in y-direction
imax 1400 # number of interior cells in x-direction
jmax 320 # number of interior cells in y-direction
# Time Data:
# ---------
te 150.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 200 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.75 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
levels 5 # Multigrid levels
# Particle Tracing Data:
# -----------------------
numberOfParticles 200
startTime 150
injectTimePeriod 1.0
writeTimePeriod 0.5
x1 0.0
y1 3.8
x2 0.0
y2 4.1
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 2
xCenter 5.0
yCenter 4.0
xRectLength 2.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

File diff suppressed because it is too large Load Diff

View File

@ -1,17 +0,0 @@
#!/bin/bash
rm data.csv
echo "Rank, SOR, RB" >> data.csv
for i in {1..72}
do
res=$(mpirun -n $i ./exe-ICC dcavity.par 1 | grep "Solution took" | cut -c 14- )
res2=$(mpirun -n $i ./exe-ICC dcavity.par 2 | grep "Solution took" | cut -c 14- )
s="$i, $res, $res2"
s=${s//"s"/}
echo $s >> data.csv
done

View File

@ -1,61 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifdef __linux__
#ifdef _OPENMP
#include <pthread.h>
#include <sched.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <unistd.h>
#define MAX_NUM_THREADS 128
#define gettid() syscall(SYS_gettid)
static int getProcessorID(cpu_set_t* cpu_set)
{
int processorId;
for (processorId = 0; processorId < MAX_NUM_THREADS; processorId++) {
if (CPU_ISSET(processorId, cpu_set)) {
break;
}
}
return processorId;
}
int affinity_getProcessorId()
{
cpu_set_t cpu_set;
CPU_ZERO(&cpu_set);
sched_getaffinity(gettid(), sizeof(cpu_set_t), &cpu_set);
return getProcessorID(&cpu_set);
}
void affinity_pinThread(int processorId)
{
cpu_set_t cpuset;
pthread_t thread;
thread = pthread_self();
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
pthread_setaffinity_np(thread, sizeof(cpu_set_t), &cpuset);
}
void affinity_pinProcess(int processorId)
{
cpu_set_t cpuset;
CPU_ZERO(&cpuset);
CPU_SET(processorId, &cpuset);
sched_setaffinity(0, sizeof(cpu_set_t), &cpuset);
}
#endif /*_OPENMP*/
#endif /*__linux__*/

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef AFFINITY_H
#define AFFINITY_H
extern int affinity_getProcessorId();
extern void affinity_pinProcess(int);
extern void affinity_pinThread(int);
#endif /*AFFINITY_H*/

View File

@ -1,35 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
void* allocate(int alignment, size_t bytesize)
{
int errorCode;
void* ptr;
errorCode = posix_memalign(&ptr, alignment, bytesize);
if (errorCode) {
if (errorCode == EINVAL) {
fprintf(stderr, "Error: Alignment parameter is not a power of two\n");
exit(EXIT_FAILURE);
}
if (errorCode == ENOMEM) {
fprintf(stderr, "Error: Insufficient memory to fulfill the request\n");
exit(EXIT_FAILURE);
}
}
if (ptr == NULL) {
fprintf(stderr, "Error: posix_memalign failed!\n");
exit(EXIT_FAILURE);
}
return ptr;
}

View File

@ -1,13 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __ALLOCATE_H_
#define __ALLOCATE_H_
#include <stdlib.h>
extern void* allocate(int alignment, size_t bytesize);
#endif

View File

@ -1,54 +0,0 @@
/*
* =======================================================================================
*
* Author: Jan Eitzinger (je), jan.eitzinger@fau.de
* Copyright (c) 2020 RRZE, University Erlangen-Nuremberg
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* =======================================================================================
*/
#ifndef LIKWID_MARKERS_H
#define LIKWID_MARKERS_H
#ifdef LIKWID_PERFMON
#include <likwid.h>
#define LIKWID_MARKER_INIT likwid_markerInit()
#define LIKWID_MARKER_THREADINIT likwid_markerThreadInit()
#define LIKWID_MARKER_SWITCH likwid_markerNextGroup()
#define LIKWID_MARKER_REGISTER(regionTag) likwid_markerRegisterRegion(regionTag)
#define LIKWID_MARKER_START(regionTag) likwid_markerStartRegion(regionTag)
#define LIKWID_MARKER_STOP(regionTag) likwid_markerStopRegion(regionTag)
#define LIKWID_MARKER_CLOSE likwid_markerClose()
#define LIKWID_MARKER_RESET(regionTag) likwid_markerResetRegion(regionTag)
#define LIKWID_MARKER_GET(regionTag, nevents, events, time, count) \
likwid_markerGetRegion(regionTag, nevents, events, time, count)
#else /* LIKWID_PERFMON */
#define LIKWID_MARKER_INIT
#define LIKWID_MARKER_THREADINIT
#define LIKWID_MARKER_SWITCH
#define LIKWID_MARKER_REGISTER(regionTag)
#define LIKWID_MARKER_START(regionTag)
#define LIKWID_MARKER_STOP(regionTag)
#define LIKWID_MARKER_CLOSE
#define LIKWID_MARKER_GET(regionTag, nevents, events, time, count)
#define LIKWID_MARKER_RESET(regionTag)
#endif /* LIKWID_PERFMON */
#endif /*LIKWID_MARKERS_H*/

View File

@ -1,92 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <float.h>
#include <limits.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include "parameter.h"
#include "particletracing.h"
#include "progress.h"
#include "timing.h"
#include <mpi.h>
int main(int argc, char** argv)
{
int rank;
double S, E;
Parameter params;
Solver solver;
ParticleTracer particletracer;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
initParameter(&params);
if (argc < 2) {
printf("Usage: %s <configFile>\n", argv[0]);
exit(EXIT_SUCCESS);
}
readParameter(&params, argv[1]);
initSolver(&solver, &params);
initParticleTracer(&particletracer, &params, &solver);
if (rank == 0) {
printParameter(&params);
}
printParticleTracerParameters(&particletracer);
initProgress(solver.te);
double tau = solver.tau;
double te = solver.te;
double t = 0.0;
S = getTimeStamp();
while (t <= te) {
if (tau > 0.0) {
computeTimestep(&solver);
}
setBoundaryConditions(&solver);
setSpecialBoundaryCondition(&solver);
setObjectBoundaryCondition(&solver);
computeFG(&solver);
computeRHS(&solver);
multiGrid(&solver);
adaptUV(&solver);
trace(&particletracer, solver.u, solver.v, solver.s, t);
t += solver.dt;
#ifdef VERBOSE
if (rank == 0) {
printf("TIME %f , TIMESTEP %f\n", t, solver.dt);
}
#else
printProgress(t);
#endif
}
E = getTimeStamp();
stopProgress();
if (rank == 0) {
printf("Solution took %.2fs\n", E - S);
}
collectResult(&solver);
freeParticles(&particletracer);
MPI_Finalize();
return EXIT_SUCCESS;
}

View File

@ -1,135 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "parameter.h"
#include "util.h"
#define MAXLINE 4096
void initParameter(Parameter* param)
{
param->xlength = 1.0;
param->ylength = 1.0;
param->imax = 100;
param->jmax = 100;
param->itermax = 1000;
param->eps = 0.0001;
param->omg = 1.7;
param->re = 100.0;
param->gamma = 0.9;
param->tau = 0.5;
param->rho = 0.99;
param->levels = 5;
}
void readParameter(Parameter* param, const char* filename)
{
FILE* fp = fopen(filename, "r");
char line[MAXLINE];
int i;
if (!fp) {
fprintf(stderr, "Could not open parameter file: %s\n", filename);
exit(EXIT_FAILURE);
}
while (!feof(fp)) {
line[0] = '\0';
fgets(line, MAXLINE, fp);
for (i = 0; line[i] != '\0' && line[i] != '#'; i++)
;
line[i] = '\0';
char* tok = strtok(line, " ");
char* val = strtok(NULL, " ");
#define PARSE_PARAM(p, f) \
if (strncmp(tok, #p, sizeof(#p) / sizeof(#p[0]) - 1) == 0) { \
param->p = f(val); \
}
#define PARSE_STRING(p) PARSE_PARAM(p, strdup)
#define PARSE_INT(p) PARSE_PARAM(p, atoi)
#define PARSE_REAL(p) PARSE_PARAM(p, atof)
if (tok != NULL && val != NULL) {
PARSE_REAL(xlength);
PARSE_REAL(ylength);
PARSE_INT(imax);
PARSE_INT(jmax);
PARSE_INT(itermax);
PARSE_INT(levels);
PARSE_REAL(eps);
PARSE_REAL(omg);
PARSE_REAL(re);
PARSE_REAL(tau);
PARSE_REAL(gamma);
PARSE_REAL(dt);
PARSE_REAL(te);
PARSE_REAL(gx);
PARSE_REAL(gy);
PARSE_STRING(name);
PARSE_INT(bcLeft);
PARSE_INT(bcRight);
PARSE_INT(bcBottom);
PARSE_INT(bcTop);
PARSE_REAL(u_init);
PARSE_REAL(v_init);
PARSE_REAL(p_init);
PARSE_REAL(rho);
/* Added new particle tracing parameters */
PARSE_INT(numberOfParticles);
PARSE_REAL(startTime);
PARSE_REAL(injectTimePeriod);
PARSE_REAL(writeTimePeriod);
PARSE_REAL(x1);
PARSE_REAL(y1);
PARSE_REAL(x2);
PARSE_REAL(y2);
/* Added obstacle geometry parameters */
PARSE_INT(shape);
PARSE_REAL(xCenter);
PARSE_REAL(yCenter);
PARSE_REAL(xRectLength);
PARSE_REAL(yRectLength);
PARSE_REAL(circleRadius);
}
}
fclose(fp);
}
void printParameter(Parameter* param)
{
printf("Parameters for %s\n", param->name);
printf("Boundary conditions Left:%d Right:%d Bottom:%d Top:%d\n",
param->bcLeft,
param->bcRight,
param->bcBottom,
param->bcTop);
printf("\tReynolds number: %.2f\n", param->re);
printf("\tInit arrays: U:%.2f V:%.2f P:%.2f\n",
param->u_init,
param->v_init,
param->p_init);
printf("Geometry data:\n");
printf("\tDomain box size (x, y): %.2f, %.2f\n", param->xlength, param->ylength);
printf("\tCells (x, y): %d, %d\n", param->imax, param->jmax);
printf("Timestep parameters:\n");
printf("\tDefault stepsize: %.2f, Final time %.2f\n", param->dt, param->te);
printf("\tTau factor: %.2f\n", param->tau);
printf("Iterative solver parameters:\n");
printf("\tMax iterations: %d\n", param->itermax);
printf("\tepsilon (stopping tolerance) : %f\n", param->eps);
printf("\tgamma (stopping tolerance) : %f\n", param->gamma);
printf("\tomega (SOR relaxation): %f\n", param->omg);
}

View File

@ -1,34 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __PARAMETER_H_
#define __PARAMETER_H_
typedef struct {
double xlength, ylength;
int imax, jmax;
int itermax, levels;
double eps, omg, rho;
double re, tau, gamma;
double te, dt;
double gx, gy;
char* name;
int bcLeft, bcRight, bcBottom, bcTop;
double u_init, v_init, p_init;
int numberOfParticles;
double startTime, injectTimePeriod, writeTimePeriod;
double x1, y1, x2, y2;
int shape;
double xCenter, yCenter, xRectLength, yRectLength, circleRadius;
} Parameter;
void initParameter(Parameter*);
void readParameter(Parameter*, const char*);
void printParameter(Parameter*);
#endif

View File

@ -1,629 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <float.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "particletracing.h"
#define U(i, j) u[(j) * (imaxLocal + 2) + (i)]
#define V(i, j) v[(j) * (imaxLocal + 2) + (i)]
#define S(i, j) s[(j) * (imaxLocal + 2) + (i)]
static int ts = 0;
#define IDIM 0
#define JDIM 1
#define XOFFSET 0
#define YOFFSET 1
#define XOFFSETEND 2
#define YOFFSETEND 3
static double sum(int* sizes, int size)
{
double sum = 0;
for (int i = 0; i < size; ++i) {
sum += sizes[i];
}
return sum;
}
void printUV(ParticleTracer* particletracer, double* u, double* v)
{
int imaxLocal = particletracer->imaxLocal;
for (int i = 0; i < particletracer->size; i++) {
if (i == particletracer->rank) {
printf(
"\n### RANK %d #######################################################\n",
particletracer->rank);
printf("\nGrid U : \n");
for (int j = 0; j < particletracer->jmaxLocal + 2; j++) {
printf("%02d: ", j);
for (int i = 0; i < particletracer->imaxLocal + 2; i++) {
printf("%4.2f ", u[j * (imaxLocal + 2) + i]);
}
printf("\n");
}
fflush(stdout);
printf("\nGrid V : \n");
for (int j = 0; j < particletracer->jmaxLocal + 2; j++) {
printf("%02d: ", j);
for (int i = 0; i < particletracer->imaxLocal + 2; i++) {
printf("%4.2f ", v[j * (imaxLocal + 2) + i]);
}
printf("\n");
}
fflush(stdout);
}
MPI_Barrier(MPI_COMM_WORLD);
}
}
void printParticles(ParticleTracer* particletracer)
{
for (int i = 0; i < particletracer->totalParticles; ++i) {
printf("Rank : %d Particle position X : %.2f, Y : %.2f, flag : %d, total pt : "
"%d, pointer : %d, xOffset : %.2f, yOffset : %.2f, xOffsetEnd : %.2f, "
"yOffsetEnd : %.2f\n",
particletracer->rank,
particletracer->particlePool[i].x,
particletracer->particlePool[i].y,
particletracer->particlePool[i].flag,
particletracer->totalParticles,
particletracer->pointer,
particletracer->xOffset,
particletracer->yOffset,
particletracer->xOffsetEnd,
particletracer->yOffsetEnd);
}
}
void injectParticles(ParticleTracer* particletracer)
{
double x, y;
for (int i = 0; i < particletracer->numberOfParticles; ++i) {
x = particletracer->linSpaceLine[i].x;
y = particletracer->linSpaceLine[i].y;
if (x >= particletracer->xOffset && y >= particletracer->yOffset &&
x <= particletracer->xOffsetEnd && y <= particletracer->yOffsetEnd) {
// printf("\nRank : %d\n", particletracer->rank);
// printf("\t%.2f >= %.2f && %.2f >= %.2f && %.2f <= %.2f && %.2f <= %.2f\n",x
// , particletracer->xOffset ,y , particletracer->yOffset, x ,
// particletracer->xOffsetEnd ,y , particletracer->yOffsetEnd);
particletracer->particlePool[particletracer->pointer].x = x;
particletracer->particlePool[particletracer->pointer].y = y;
particletracer->particlePool[particletracer->pointer].flag = true;
++(particletracer->pointer);
++(particletracer->totalParticles);
}
}
}
void advanceParticles(ParticleTracer* particletracer,
double* restrict u,
double* restrict v,
int* restrict s,
double time)
{
int imax = particletracer->imax;
int jmax = particletracer->jmax;
int imaxLocal = particletracer->imaxLocal;
int jmaxLocal = particletracer->jmaxLocal;
double dx = particletracer->dx;
double dy = particletracer->dy;
double xlength = particletracer->xlength;
double ylength = particletracer->ylength;
Particle buff[particletracer->size][100];
for (int i = 0; i < particletracer->size; ++i) {
for (int j = 0; j < 100; ++j) {
buff[i][j].x = 0.0;
buff[i][j].y = 0.0;
buff[i][j].flag = false;
}
}
int particleBufIndex[particletracer->size];
memset(particleBufIndex, 0, sizeof(particleBufIndex));
for (int i = 0; i < particletracer->totalParticles; ++i) {
if (particletracer->particlePool[i].flag == true) {
double xTemp = particletracer->particlePool[i].x;
double yTemp = particletracer->particlePool[i].y;
double x = xTemp - particletracer->xOffset;
double y = yTemp - particletracer->yOffset;
int iCoord = (int)(x / dx) + 1;
int jCoord = (int)((y + 0.5 * dy) / dy) + 1;
double x1 = (double)(iCoord - 1) * dx;
double y1 = ((double)(jCoord - 1) - 0.5) * dy;
double x2 = (double)iCoord * dx;
double y2 = ((double)jCoord - 0.5) * dy;
double u_n = (1.0 / (dx * dy)) *
((x2 - x) * (y2 - y) * U(iCoord - 1, jCoord - 1) +
(x - x1) * (y2 - y) * U(iCoord, jCoord - 1) +
(x2 - x) * (y - y1) * U(iCoord - 1, jCoord) +
(x - x1) * (y - y1) * U(iCoord, jCoord));
double new_x = (x + particletracer->xOffset) + particletracer->dt * u_n;
particletracer->particlePool[i].x = new_x;
iCoord = (int)((x + 0.5 * dx) / dx) + 1;
jCoord = (int)(y / dy) + 1;
x1 = ((double)(iCoord - 1) - 0.5) * dx;
y1 = (double)(jCoord - 1) * dy;
x2 = ((double)iCoord - 0.5) * dx;
y2 = (double)jCoord * dy;
double v_n = (1.0 / (dx * dy)) *
((x2 - x) * (y2 - y) * V(iCoord - 1, jCoord - 1) +
(x - x1) * (y2 - y) * V(iCoord, jCoord - 1) +
(x2 - x) * (y - y1) * V(iCoord - 1, jCoord) +
(x - x1) * (y - y1) * V(iCoord, jCoord));
double new_y = (y + particletracer->yOffset) + particletracer->dt * v_n;
particletracer->particlePool[i].y = new_y;
// printf("Rank : %d\n", particletracer->rank);
// printf("\tOld X : %.2f, translated X : %.2f, xOffset : %.2f, New X : %.2f,
// iCoord : %d\n\tOld Y : %.2f, translated X : %.2f, yOffset : %.2f, New Y :
// %.2f, jCoord : %d\n\n",xTemp, x, particletracer->xOffset, new_x, iCoord,
// yTemp, y, particletracer->yOffset , new_y, jCoord); printf("\tU(iCoord - 1,
// jCoord - 1) : %.2f, U(iCoord, jCoord - 1) : %.2f, U(iCoord - 1, jCoord) :
// %.2f, U(iCoord, jCoord) : %.2f\n", U(iCoord - 1, jCoord - 1), U(iCoord,
// jCoord - 1), U(iCoord - 1, jCoord), U(iCoord, jCoord)); printf("\tV(iCoord
// - 1, jCoord - 1) : %.2f, V(iCoord, jCoord - 1) : %.2f, V(iCoord - 1,
// jCoord) : %.2f, V(iCoord, jCoord) : %.2f\n\n", V(iCoord - 1, jCoord - 1),
// V(iCoord, jCoord - 1), V(iCoord - 1, jCoord), V(iCoord, jCoord));
// printf("\t U N : %.2f, V N : %.2f\n\n", u_n, v_n);
if (((new_x < particletracer->xOffset) ||
(new_x >= particletracer->xOffsetEnd) ||
(new_y < particletracer->yOffset) ||
(new_y >= particletracer->yOffsetEnd))) {
// New logic to transfer particles to neighbouring ranks or discard the
// particle.
for (int i = 0; i < particletracer->size; ++i) {
if ((new_x >=
particletracer->offset[i + particletracer->size * XOFFSET]) &&
(new_x <= particletracer
->offset[i + particletracer->size * XOFFSETEND]) &&
(new_y >=
particletracer->offset[i + particletracer->size * YOFFSET]) &&
(new_y <= particletracer
->offset[i + particletracer->size * YOFFSETEND]) &&
i != particletracer->rank) {
buff[i][particleBufIndex[i]].x = new_x;
buff[i][particleBufIndex[i]].y = new_y;
buff[i][particleBufIndex[i]].flag = true;
++particleBufIndex[i];
}
}
particletracer->particlePool[i].flag = false;
}
int i_new = new_x / dx, j_new = new_y / dy;
int iOffset = particletracer->xOffset / dx,
jOffset = particletracer->yOffset / dy;
if (S(i_new - iOffset, j_new - jOffset) != NONE) {
particletracer->particlePool[i].flag = false;
}
}
}
for (int i = 0; i < particletracer->size; ++i) {
if (i != particletracer->rank) {
MPI_Send(buff[i],
100,
particletracer->mpi_particle,
i,
0,
particletracer->comm);
}
}
for (int i = 0; i < particletracer->size; ++i) {
if (i != particletracer->rank) {
MPI_Recv(buff[i],
100,
particletracer->mpi_particle,
i,
0,
particletracer->comm,
MPI_STATUS_IGNORE);
}
}
for (int i = 0; i < particletracer->size; ++i) {
if (i != particletracer->rank) {
for (int j = 0; j < 100; ++j) {
if (buff[i][j].flag == true) {
particletracer->particlePool[particletracer->pointer].x = buff[i][j]
.x;
particletracer->particlePool[particletracer->pointer].y = buff[i][j]
.y;
particletracer->particlePool[particletracer->pointer].flag = true;
++(particletracer->pointer);
++(particletracer->totalParticles);
}
}
}
}
}
void freeParticles(ParticleTracer* particletracer)
{
free(particletracer->particlePool);
free(particletracer->linSpaceLine);
free(particletracer->offset);
}
void writeParticles(ParticleTracer* particletracer)
{
int collectedBuffIndex[particletracer->size];
MPI_Gather(&particletracer->totalParticles,
1,
MPI_INT,
collectedBuffIndex,
1,
MPI_INT,
0,
particletracer->comm);
if (particletracer->rank != 0) {
Particle buff[particletracer->totalParticles];
for (int i = 0; i < particletracer->totalParticles; ++i) {
buff[i].x = particletracer->particlePool[i].x;
buff[i].y = particletracer->particlePool[i].y;
buff[i].flag = particletracer->particlePool[i].flag;
// printf("Rank : %d sending to rank 0 X : %.2f, Y : %.2f with totalpt :
// %d\n", particletracer->rank, buff[i].x, buff[i].y,
// particletracer->totalParticles);
}
MPI_Send(buff,
particletracer->totalParticles,
particletracer->mpi_particle,
0,
1,
particletracer->comm);
}
if (particletracer->rank == 0) {
char filename[50];
FILE* fp;
snprintf(filename, 50, "vis_files/particles_%d.dat", ts);
fp = fopen(filename, "w");
if (fp == NULL) {
printf("Error!\n");
exit(EXIT_FAILURE);
}
// fprintf(fp, "# vtk DataFile Version 3.0\n");
// fprintf(fp, "PAMPI cfd solver particle tracing file\n");
// fprintf(fp, "ASCII\n");
// fprintf(fp, "DATASET UNSTRUCTURED_GRID\n");
// fprintf(fp, "FIELD FieldData 2\n");
// fprintf(fp, "TIME 1 1 double\n");
// fprintf(fp, "%d\n", ts);
// fprintf(fp, "CYCLE 1 1 int\n");
// fprintf(fp, "1\n");
int overallTotalParticles = sum(collectedBuffIndex, particletracer->size);
// fprintf(fp, "POINTS %d float\n", overallTotalParticles);
// printf("Total particles : %d\n", overallTotalParticles);
for (int i = 1; i < particletracer->size; ++i) {
Particle recvBuff[collectedBuffIndex[i]];
MPI_Recv(&recvBuff,
collectedBuffIndex[i],
particletracer->mpi_particle,
i,
1,
particletracer->comm,
MPI_STATUS_IGNORE);
for (int j = 0; j < collectedBuffIndex[i]; ++j) {
double x = recvBuff[j].x;
double y = recvBuff[j].y;
fprintf(fp, "%f %f\n", x, y);
// printf("Rank : 0 receiving from rank %d X : %.2f, Y : %.2f with totalpt
// : %d\n", i, x, y, particletracer->totalParticles);
}
}
for (int i = 0; i < particletracer->totalParticles; ++i) {
double x = particletracer->particlePool[i].x;
double y = particletracer->particlePool[i].y;
fprintf(fp, "%f %f\n", x, y);
}
// fprintf(fp, "CELLS %d %d\n", overallTotalParticles, 2 * overallTotalParticles);
// for (int i = 0; i < overallTotalParticles; ++i)
// {
// fprintf(fp, "1 %d\n", i);
// }
// fprintf(fp, "CELL_TYPES %d\n", overallTotalParticles);
// for (int i = 0; i < overallTotalParticles; ++i)
// {
// fprintf(fp, "1\n");
// }
fclose(fp);
}
++ts;
}
void initParticleTracer(ParticleTracer* particletracer, Parameter* params, Solver* solver)
{
int dims[NDIMS] = { 0, 0 };
int periods[NDIMS] = { 0, 0 };
/* initializing local properties from params */
particletracer->numberOfParticles = params->numberOfParticles;
particletracer->startTime = params->startTime;
particletracer->injectTimePeriod = params->injectTimePeriod;
particletracer->writeTimePeriod = params->writeTimePeriod;
particletracer->dt = params->dt;
particletracer->dx = params->xlength / params->imax;
particletracer->dy = params->ylength / params->jmax;
particletracer->xlength = params->xlength;
particletracer->ylength = params->ylength;
particletracer->x1 = params->x1;
particletracer->y1 = params->y1;
particletracer->x2 = params->x2;
particletracer->y2 = params->y2;
particletracer->lastInjectTime = params->startTime;
particletracer->lastUpdateTime = params->startTime;
particletracer->lastWriteTime = params->startTime;
particletracer->pointer = 0;
particletracer->totalParticles = 0;
particletracer->imax = params->imax;
particletracer->jmax = params->jmax;
particletracer->imaxLocal = solver->imaxLocal;
particletracer->jmaxLocal = solver->jmaxLocal;
particletracer->estimatedNumParticles = (particletracer->imaxLocal *
particletracer->jmaxLocal);
particletracer->particlePool = malloc(
sizeof(Particle) * particletracer->estimatedNumParticles);
for (int i = 0; i < particletracer->estimatedNumParticles; ++i) {
particletracer->particlePool[i].x = 0.0;
particletracer->particlePool[i].y = 0.0;
particletracer->particlePool[i].flag = false;
}
particletracer->linSpaceLine = malloc(
sizeof(Particle) * particletracer->numberOfParticles);
/* duplicating communication from solver */
MPI_Comm_dup(solver->comm, &particletracer->comm);
particletracer->rank = solver->rank;
particletracer->size = solver->size;
particletracer->offset = (double*)malloc(sizeof(double) * 4 * particletracer->size);
memcpy(particletracer->dims, solver->dims, sizeof(solver->dims));
memcpy(particletracer->coords, solver->coords, sizeof(solver->coords));
memcpy(particletracer->iNeighbours, solver->iNeighbours, sizeof(solver->iNeighbours));
memcpy(particletracer->jNeighbours, solver->jNeighbours, sizeof(solver->jNeighbours));
particletracer->xLocal = particletracer->imaxLocal * particletracer->dx;
particletracer->yLocal = particletracer->jmaxLocal * particletracer->dy;
double offset[4][particletracer->size];
particletracer->xOffset = solver->xOffset;
particletracer->yOffset = solver->yOffset;
particletracer->xOffsetEnd = particletracer->xOffset + particletracer->xLocal;
particletracer->yOffsetEnd = particletracer->yOffset + particletracer->yLocal;
printf("Rank : %d, xOffset : %.2f, yOffset : %.2f, xOffsetEnd : %.2f, yOffsetEnd : "
"%.2f\n",
particletracer->rank,
particletracer->xOffset,
particletracer->yOffset,
particletracer->xOffsetEnd,
particletracer->yOffsetEnd);
MPI_Allgather(&particletracer->xOffset,
1,
MPI_DOUBLE,
offset[0],
1,
MPI_DOUBLE,
particletracer->comm);
MPI_Allgather(&particletracer->yOffset,
1,
MPI_DOUBLE,
offset[1],
1,
MPI_DOUBLE,
particletracer->comm);
MPI_Allgather(&particletracer->xOffsetEnd,
1,
MPI_DOUBLE,
offset[2],
1,
MPI_DOUBLE,
particletracer->comm);
MPI_Allgather(&particletracer->yOffsetEnd,
1,
MPI_DOUBLE,
offset[3],
1,
MPI_DOUBLE,
particletracer->comm);
memcpy(particletracer->offset, offset, sizeof(offset));
// if(particletracer->rank == 0)
// {
// for(int i = 0;i < particletracer->size; ++i)
// {
// printf("Rank : %d and its xOffset : %.2f, yOffset : %.2f, xOffsetEnd :
// %.2f, yOffsetEnd : %.2f\n", i, particletracer->offset[i +
// particletracer->size * XOFFSET], particletracer->offset[i +
// particletracer->size * YOFFSET], particletracer->offset[i +
// particletracer->size * XOFFSETEND], particletracer->offset[i +
// particletracer->size * YOFFSETEND]);
// }
// }
for (int i = 0; i < particletracer->numberOfParticles; ++i) {
double spacing = (double)i / (double)(particletracer->numberOfParticles - 1);
particletracer->linSpaceLine[i].x = spacing * particletracer->x1 +
(1.0 - spacing) * particletracer->x2;
particletracer->linSpaceLine[i].y = spacing * particletracer->y1 +
(1.0 - spacing) * particletracer->y2;
particletracer->linSpaceLine[i].flag = true;
// if(particletracer->rank == 1) printf("\nRank : %d, with linspace X : %.2f and
// linspace Y : %.2f\n", particletracer->rank, particletracer->linSpaceLine[i].x ,
// particletracer->linSpaceLine[i].y);
}
// Create the mpi_particle datatype
MPI_Datatype mpi_particle;
int lengths[3] = { 1, 1, 1 };
MPI_Aint displacements[3];
Particle dummy_particle;
MPI_Aint base_address;
MPI_Get_address(&dummy_particle, &base_address);
MPI_Get_address(&dummy_particle.x, &displacements[0]);
MPI_Get_address(&dummy_particle.y, &displacements[1]);
MPI_Get_address(&dummy_particle.flag, &displacements[2]);
displacements[0] = MPI_Aint_diff(displacements[0], base_address);
displacements[1] = MPI_Aint_diff(displacements[1], base_address);
displacements[2] = MPI_Aint_diff(displacements[2], base_address);
MPI_Datatype types[3] = { MPI_DOUBLE, MPI_DOUBLE, MPI_C_BOOL };
MPI_Type_create_struct(3,
lengths,
displacements,
types,
&particletracer->mpi_particle);
MPI_Type_commit(&particletracer->mpi_particle);
}
void printParticleTracerParameters(ParticleTracer* particletracer)
{
printf("Particle Tracing data:\n");
printf("Rank : %d\n", particletracer->rank);
printf("\tNumber of particles : %d being injected for every period of %.2f\n",
particletracer->numberOfParticles,
particletracer->injectTimePeriod);
printf("\tstartTime : %.2f\n", particletracer->startTime);
printf("\t(Line along which the particles are to be injected) \n\tx1 : %.2f, y1 : "
"%.2f, x2 : %.2f, y2 : %.2f\n",
particletracer->x1,
particletracer->y1,
particletracer->x2,
particletracer->y2);
printf("\tPointer : %d, TotalParticles : %d\n",
particletracer->pointer,
particletracer->totalParticles);
printf("\tdt : %.2f, dx : %.2f, dy : %.2f\n",
particletracer->dt,
particletracer->dx,
particletracer->dy);
printf("\tcoord[0] : %d, coord[1] : %d\n",
particletracer->coords[IDIM],
particletracer->coords[JDIM]);
printf("\txOffset : %.2f, yOffset : %.2f\n",
particletracer->xOffset,
particletracer->yOffset);
printf("\txOffsetEnd : %.2f, yOffsetEnd : %.2f\n",
particletracer->xOffsetEnd,
particletracer->yOffsetEnd);
printf("\txLocal : %.2f, yLocal : %.2f\n",
particletracer->xLocal,
particletracer->yLocal);
}
void trace(ParticleTracer* particletracer,
double* restrict u,
double* restrict v,
int* restrict s,
double time)
{
if (time >= particletracer->startTime) {
if ((time - particletracer->lastInjectTime) >= particletracer->injectTimePeriod) {
injectParticles(particletracer);
particletracer->lastInjectTime = time;
}
if ((time - particletracer->lastWriteTime) >= particletracer->writeTimePeriod) {
writeParticles(particletracer);
particletracer->lastWriteTime = time;
}
advanceParticles(particletracer, u, v, s, time);
compress(particletracer);
particletracer->lastUpdateTime = time;
}
}
void compress(ParticleTracer* particletracer)
{
Particle* memPool = particletracer->particlePool;
Particle tempPool[particletracer->totalParticles];
for (int i = 0; i < particletracer->totalParticles; ++i) {
tempPool[i].x = 0.0;
tempPool[i].y = 0.0;
tempPool[i].flag = false;
}
int totalParticles = 0;
for (int i = 0; i < particletracer->totalParticles; ++i) {
if (memPool[i].flag == true) {
tempPool[totalParticles].x = memPool[i].x;
tempPool[totalParticles].y = memPool[i].y;
tempPool[totalParticles].flag = memPool[i].flag;
++totalParticles;
}
}
particletracer->totalParticles = totalParticles;
particletracer->pointer = totalParticles;
memcpy(particletracer->particlePool, tempPool, totalParticles * sizeof(Particle));
}

View File

@ -1,65 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __PARTICLETRACING_H_
#define __PARTICLETRACING_H_
#include "allocate.h"
#include "parameter.h"
#include "solver.h"
#include <mpi.h>
#include <stdbool.h>
#define NDIMS 2
typedef enum COORD { X = 0, Y, NCOORD } COORD;
typedef struct {
double x, y;
bool flag;
} Particle;
typedef struct {
int numberOfParticles, totalParticles;
double startTime, injectTimePeriod, writeTimePeriod, lastInjectTime, lastUpdateTime,
lastWriteTime;
int estimatedNumParticles;
double dx, dy, dt;
Particle* linSpaceLine;
Particle* particlePool;
int pointer;
double imax, jmax, xlength, ylength, imaxLocal, jmaxLocal;
double x1, y1, x2, y2;
MPI_Comm comm;
MPI_Datatype mpi_particle;
int rank, size;
int iNeighbours[NDIMS], jNeighbours[NDIMS];
int coords[NDIMS], dims[NDIMS];
double xLocal, yLocal, xOffset, yOffset, xOffsetEnd, yOffsetEnd;
double* offset;
} ParticleTracer;
extern void initParticleTracer(ParticleTracer*, Parameter*, Solver*);
extern void injectParticles(ParticleTracer*);
extern void advanceParticles(ParticleTracer*, double*, double*, int*, double);
extern void freeParticles(ParticleTracer*);
extern void writeParticles(ParticleTracer*);
extern void printParticleTracerParameters(ParticleTracer*);
extern void printParticles(ParticleTracer*);
extern void trace(ParticleTracer*, double*, double*, int*, double);
extern void compress(ParticleTracer*);
#endif

View File

@ -1,60 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#include <math.h>
#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "progress.h"
static double _end;
static int _current;
static int _rank = -1;
void initProgress(double end)
{
MPI_Comm_rank(MPI_COMM_WORLD, &_rank);
_end = end;
_current = 0;
if (_rank == 0) {
printf("[ ]");
fflush(stdout);
}
}
void printProgress(double current)
{
if (_rank == 0) {
int new = (int)rint((current / _end) * 10.0);
if (new > _current) {
char progress[11];
_current = new;
progress[0] = 0;
for (int i = 0; i < 10; i++) {
if (i < _current) {
sprintf(progress + strlen(progress), "#");
} else {
sprintf(progress + strlen(progress), " ");
}
}
printf("\r[%s]", progress);
}
fflush(stdout);
}
}
void stopProgress()
{
if (_rank == 0) {
printf("\n");
fflush(stdout);
}
}

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __PROGRESS_H_
#define __PROGRESS_H_
extern void initProgress(double);
extern void printProgress(double);
extern void stopProgress();
#endif

File diff suppressed because it is too large Load Diff

View File

@ -1,91 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved. This file is part of nusif-solver.
* Use of this source code is governed by a MIT style
* license that can be found in the LICENSE file.
*/
#ifndef __SOLVER_H_
#define __SOLVER_H_
#include "parameter.h"
#include <mpi.h>
#define NDIMS 2
enum OBJECTBOUNDARY {
NONE = 0,
TOP,
BOTTOM,
LEFT,
RIGHT,
TOPLEFT,
BOTTOMLEFT,
TOPRIGHT,
BOTTOMRIGHT,
LOCAL
};
enum BC { NOSLIP = 1, SLIP, OUTFLOW, PERIODIC };
/// @brief
enum SHAPE { NOSHAPE = 0, RECT, CIRCLE };
typedef struct {
/* geometry and grid information */
double dx, dy;
int imax, jmax;
double xlength, ylength;
/* arrays */
double *p, *rhs, **r, **e;
;
double *f, *g;
double *u, *v;
int* s;
/* parameters */
double eps, omega, rho;
double re, tau, gamma;
double gx, gy;
/* time stepping */
int itermax, levels, currentlevel;
double dt, te;
double dtBound;
char* problem;
int bcLeft, bcRight, bcBottom, bcTop;
/* mpi */
int rank;
int size;
MPI_Comm comm;
MPI_Datatype bufferTypes[NDIMS * 2];
MPI_Aint sdispls[NDIMS * 2], rdispls[NDIMS * 2];
MPI_Datatype bufferIntTypes[NDIMS * 2];
MPI_Aint sdisplsInt[NDIMS * 2], rdisplsInt[NDIMS * 2];
int iNeighbours[NDIMS], jNeighbours[NDIMS];
int coords[NDIMS], dims[NDIMS];
int imaxLocal, jmaxLocal;
double xLocal, yLocal, xOffset, yOffset, xOffsetEnd, yOffsetEnd;
} Solver;
extern void initSolver(Solver*, Parameter*);
extern void computeRHS(Solver*);
extern double smoothRB(Solver*);
extern void restrictMG(Solver*);
extern void prolongate(Solver*);
extern void correct(Solver*);
extern Solver copySolver(Solver*);
extern void multiGrid(Solver*);
extern void computeTimestep(Solver*);
extern void setBoundaryConditions(Solver*);
extern void setSpecialBoundaryCondition(Solver*);
extern void setObjectBoundaryCondition(Solver*);
extern void setPressureBoundaryCondition(Solver*);
extern void computeFG(Solver*);
extern void adaptUV(Solver*);
extern void collectResult(Solver*);
extern void writeResult(Solver*, double*, double*, double*);
extern void debugExchange(Solver*, double*);
extern void print(Solver*, double*);
extern void printInt(Solver*, int*);
#endif

View File

@ -1,24 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#include <stdlib.h>
#include <time.h>
double getTimeStamp()
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeResolution()
{
struct timespec ts;
clock_getres(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
}
double getTimeStamp_() { return getTimeStamp(); }

View File

@ -1,14 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __TIMING_H_
#define __TIMING_H_
extern double getTimeStamp();
extern double getTimeResolution();
extern double getTimeStamp_();
#endif // __TIMING_H_

View File

@ -1,22 +0,0 @@
/*
* Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
* All rights reserved.
* Use of this source code is governed by a MIT-style
* license that can be found in the LICENSE file.
*/
#ifndef __UTIL_H_
#define __UTIL_H_
#define HLINE \
"----------------------------------------------------------------------------\n"
#ifndef MIN
#define MIN(x, y) ((x) < (y) ? (x) : (y))
#endif
#ifndef MAX
#define MAX(x, y) ((x) > (y) ? (x) : (y))
#endif
#ifndef ABS
#define ABS(a) ((a) >= 0 ? (a) : -(a))
#endif
#endif // __UTIL_H_

View File

@ -1,7 +0,0 @@
set terminal png size 1024,768 enhanced font ,12
set output 'p.png'
set datafile separator whitespace
set grid
set hidden3d
splot 'pressure.dat' using 1:2:3 with lines

View File

@ -1,6 +0,0 @@
set terminal png size 3600,768 enhanced font ,28
set output 'velocity.png'
set size ratio -1
set datafile separator whitespace
plot 'velocity.dat' using 1:2:3:4:5 with vectors filled head size 0.01,20,60 lc palette

File diff suppressed because it is too large Load Diff

Binary file not shown.

Before

Width:  |  Height:  |  Size: 338 KiB

View File

@ -1,12 +0,0 @@
unset border; unset tics; unset key;
set term gif animate delay 10
set output "trace.gif"
set xrange [0:7]
set yrange [0:1.5]
set object 1 rect from 0.0,0.0 to 1.0,0.5 lw 2
do for [ts=0:300] {
plot "particles_".ts.".dat" with points pointtype 7
}
unset output

Binary file not shown.

Before

Width:  |  Height:  |  Size: 751 KiB

View File

@ -1,76 +0,0 @@
#=======================================================================================
# Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
# All rights reserved.
# Use of this source code is governed by a MIT-style
# license that can be found in the LICENSE file.
#=======================================================================================
#CONFIGURE BUILD SYSTEM
TARGET = exe-$(TAG)
BUILD_DIR = ./$(TAG)
SRC_DIR = ./src
MAKE_DIR = ./
Q ?= @
#DO NOT EDIT BELOW
include $(MAKE_DIR)/config.mk
include $(MAKE_DIR)/include_$(TAG).mk
INCLUDES += -I$(SRC_DIR) -I$(BUILD_DIR)
VPATH = $(SRC_DIR)
SRC = $(wildcard $(SRC_DIR)/*.c)
ASM = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.s, $(SRC))
OBJ = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.o, $(SRC))
SOURCES = $(SRC) $(wildcard $(SRC_DIR)/*.h)
CPPFLAGS := $(CPPFLAGS) $(DEFINES) $(OPTIONS) $(INCLUDES)
${TARGET}: $(BUILD_DIR) $(OBJ)
$(info ===> LINKING $(TARGET))
$(Q)${LINKER} ${LFLAGS} -o $(TARGET) $(OBJ) $(LIBS)
$(BUILD_DIR)/%.o: %.c $(MAKE_DIR)/include_$(TAG).mk $(MAKE_DIR)/config.mk
$(info ===> COMPILE $@)
$(CC) -c $(CPPFLAGS) $(CFLAGS) $< -o $@
$(Q)$(GCC) $(CPPFLAGS) -MT $(@:.d=.o) -MM $< > $(BUILD_DIR)/$*.d
$(BUILD_DIR)/%.s: %.c
$(info ===> GENERATE ASM $@)
$(CC) -S $(CPPFLAGS) $(CFLAGS) $< -o $@
.PHONY: clean distclean tags info asm format
clean: vis
$(info ===> CLEAN)
@rm -rf $(BUILD_DIR)
@rm -f tags
distclean: clean
$(info ===> DIST CLEAN)
@rm -f $(TARGET)
info:
$(info $(CFLAGS))
$(Q)$(CC) $(VERSION)
asm: $(BUILD_DIR) $(ASM)
vis:
$(info ===> REMOVING VIZUALISATION FILES)
@rm -f vtk_files/particle*.vtk
@rm -f vis_files/particle*.dat
tags:
$(info ===> GENERATE TAGS)
$(Q)ctags -R
format:
@for src in $(SOURCES) ; do \
echo "Formatting $$src" ; \
clang-format -i $$src ; \
done
@echo "Done"
$(BUILD_DIR):
@mkdir $(BUILD_DIR)
-include $(OBJ:.o=.d)

View File

@ -1,48 +0,0 @@
# C source skeleton
## Build
1. Configure the toolchain and additional options in `config.mk`:
```
# Supported: GCC, CLANG, ICC
TAG ?= GCC
ENABLE_OPENMP ?= false
OPTIONS += -DARRAY_ALIGNMENT=64
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER
```
The verbosity options enable detailed output about affinity settings, allocation sizes and timer resolution.
2. Build with:
```
make
```
You can build multiple toolchains in the same directory, but notice that the Makefile is only acting on the one currently set.
Intermediate build results are located in the `<TOOLCHAIN>` directory.
To output the executed commands use:
```
make Q=
```
3. Clean up with:
```
make clean
```
to clean intermediate build results.
```
make distclean
```
to clean intermediate build results and binary.
4. (Optional) Generate assembler:
```
make asm
```
The assembler files will also be located in the `<TOOLCHAIN>` directory.

View File

@ -1,72 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name backstep # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 36500.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 1.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 7.0 # domain size in x-direction
ylength 1.5 # domain size in y-direction
imax 210 # number of interior cells in x-direction
jmax 45 # number of interior cells in y-direction
# Time Data:
# ---------
te 60.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.0001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
# Particle Tracing Data:
# -----------------------
numberOfParticles 200
startTime 0
injectTimePeriod 1.0
writeTimePeriod 0.5
x1 0.0
y1 0.5
x2 0.0
y2 1.5
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 1
xCenter 0.0
yCenter 0.0
xRectLength 2.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

View File

@ -1,72 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name canal # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 100.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 1.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 4.0 # domain size in y-direction
imax 200 # number of interior cells in x-direction
jmax 50 # number of interior cells in y-direction
# Time Data:
# ---------
te 60.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 500 # maximal number of pressure iteration in one time step
eps 0.0001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.8 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
# Particle Tracing Data:
# -----------------------
numberOfParticles 60
startTime 500
injectTimePeriod 2.0
writeTimePeriod 0.5
x1 1.0
y1 0.0
x2 1.0
y2 4.0
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 1
xCenter 10.0
yCenter 2
xRectLength 6.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

View File

@ -1,10 +0,0 @@
# Supported: GCC, CLANG, ICC
TAG ?= ICC
ENABLE_OPENMP ?= false
#Feature options
OPTIONS += -DARRAY_ALIGNMENT=64
OPTIONS += -DVERBOSE
#OPTIONS += -DVERBOSE_AFFINITY
#OPTIONS += -DVERBOSE_DATASIZE
#OPTIONS += -DVERBOSE_TIMER

View File

@ -1,73 +0,0 @@
Rank, SOR, RB
1, 86.28, 46.72
2, 41.61, 32.01
3, 30.54, 29.39
4, 25.42, 21.58
5, 22.41, 24.34
6, 19.24, 18.86
7, 17.98, 22.45
8, 17.08, 18.98
9, 16.59, 19.84
10, 15.30, 18.60
11, 15.49, 21.93
12, 15.05, 19.61
13, 15.78, 21.57
14, 14.44, 16.53
15, 13.94, 17.54
16, 13.14, 18.97
17, 17.31, 19.64
18, 13.59, 18.33
19, 22.20, 34.39
20, 12.55, 16.40
21, 12.75, 18.14
22, 12.61, 17.22
23, 3.30, 30.04
24, 12.25, 18.27
25, 12.42, 16.13
26, 12.74, 17.17
27, 12.08, 19.87
28, 12.31, 17.62
29, 2.95, 3.37
30, 12.14, 16.16
31, 2.78, 3.53
32, 11.80, 19.34
33, 12.52, 21.27
34, 13.06, 15.24
35, 12.13, 15.23
36, 12.13, 18.15
37, 3.11, 3.67
38, 26.77, 26.82
39, 12.39, 21.69
40, 11.34, 15.61
41, 2.89, 3.34
42, 11.92, 18.12
43, 2.79, 3.40
44, 14.22, 37.07
45, 12.12, 15.90
46, 2.29, 23.39
47, 2.91, 3.55
48, 12.27, 21.39
49, 12.71, 19.82
50, 13.01, 16.37
51, 2.91, 17.26
52, 2.39, 5.00
53, 2.95, 3.47
54, 15.58, 65.25
55, 2.59, 15.84
56, 11.78, 21.47
57, 2.36, 3.88
58, 2.18, 2.34
59, 2.95, 3.33
60, 16.75, 14.88
61, 2.73, 3.20
62, 2.32, 2.73
63, 2.60, 25.96
64, 19.61, 36.29
65, 2.08, 17.16
66, 1.93, 2.32
67, 2.72, 3.44
68, 2.06, 17.25
69, 2.13, 129.62
70, 2.05, 15.92
71, 2.66, 3.21
72, 2.16, 2.54
1 Rank SOR RB
2 1 86.28 46.72
3 2 41.61 32.01
4 3 30.54 29.39
5 4 25.42 21.58
6 5 22.41 24.34
7 6 19.24 18.86
8 7 17.98 22.45
9 8 17.08 18.98
10 9 16.59 19.84
11 10 15.30 18.60
12 11 15.49 21.93
13 12 15.05 19.61
14 13 15.78 21.57
15 14 14.44 16.53
16 15 13.94 17.54
17 16 13.14 18.97
18 17 17.31 19.64
19 18 13.59 18.33
20 19 22.20 34.39
21 20 12.55 16.40
22 21 12.75 18.14
23 22 12.61 17.22
24 23 3.30 30.04
25 24 12.25 18.27
26 25 12.42 16.13
27 26 12.74 17.17
28 27 12.08 19.87
29 28 12.31 17.62
30 29 2.95 3.37
31 30 12.14 16.16
32 31 2.78 3.53
33 32 11.80 19.34
34 33 12.52 21.27
35 34 13.06 15.24
36 35 12.13 15.23
37 36 12.13 18.15
38 37 3.11 3.67
39 38 26.77 26.82
40 39 12.39 21.69
41 40 11.34 15.61
42 41 2.89 3.34
43 42 11.92 18.12
44 43 2.79 3.40
45 44 14.22 37.07
46 45 12.12 15.90
47 46 2.29 23.39
48 47 2.91 3.55
49 48 12.27 21.39
50 49 12.71 19.82
51 50 13.01 16.37
52 51 2.91 17.26
53 52 2.39 5.00
54 53 2.95 3.47
55 54 15.58 65.25
56 55 2.59 15.84
57 56 11.78 21.47
58 57 2.36 3.88
59 58 2.18 2.34
60 59 2.95 3.33
61 60 16.75 14.88
62 61 2.73 3.20
63 62 2.32 2.73
64 63 2.60 25.96
65 64 19.61 36.29
66 65 2.08 17.16
67 66 1.93 2.32
68 67 2.72 3.44
69 68 2.06 17.25
70 69 2.13 129.62
71 70 2.05 15.92
72 71 2.66 3.21
73 72 2.16 2.54

View File

@ -1,60 +0,0 @@
#==============================================================================
# Driven Cavity
#==============================================================================
# Problem specific Data:
# ---------------------
name dcavity # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 1 # 2 = free-slip 4 = periodic
bcRight 1 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 10.0 # Reynolds number
u_init 0.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 1.0 # domain size in x-direction
ylength 1.0 # domain size in y-direction
imax 100 # number of interior cells in x-direction
jmax 100 # number of interior cells in y-direction
# Time Data:
# ---------
te 10.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 1000 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.5 #
omg 1.7 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
# Particle Tracing Data:
# -----------------------
numberOfParticles 30
startTime 2
injectTimePeriod 2.0
writeTimePeriod 0.1
x1 0.1
y1 0.5
x2 0.9
y2 0.5
#===============================================================================

View File

@ -1,16 +0,0 @@
CC = mpicc
GCC = cc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
#OPENMP = -Xpreprocessor -fopenmp #required on Macos with homebrew libomp
LIBS = # -lomp
endif
VERSION = --version
CFLAGS = -Ofast -std=c99 $(OPENMP)
#CFLAGS = -Ofast -fnt-store=aggressive -std=c99 $(OPENMP) #AMD CLANG
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE# -DDEBUG
INCLUDES = -I/usr/local/include

View File

@ -1,14 +0,0 @@
CC = gcc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -fopenmp
endif
VERSION = --version
CFLAGS = -Ofast -ffreestanding -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

View File

@ -1,14 +0,0 @@
CC = mpiicc
GCC = gcc
LINKER = $(CC)
ifeq ($(ENABLE_OPENMP),true)
OPENMP = -qopenmp
endif
VERSION = --version
CFLAGS = -O3 -xHost -qopt-zmm-usage=high -std=c99 $(OPENMP)
LFLAGS = $(OPENMP)
DEFINES = -D_GNU_SOURCE
INCLUDES =
LIBS =

View File

@ -1,72 +0,0 @@
#==============================================================================
# Laminar Canal Flow
#==============================================================================
# Problem specific Data:
# ---------------------
name karman # name of flow setup
bcTop 1 # flags for boundary conditions
bcBottom 1 # 1 = no-slip 3 = outflow
bcLeft 3 # 2 = free-slip 4 = periodic
bcRight 3 #
gx 0.0 # Body forces (e.g. gravity)
gy 0.0 #
re 5050.0 # Reynolds number
u_init 1.0 # initial value for velocity in x-direction
v_init 0.0 # initial value for velocity in y-direction
p_init 0.0 # initial value for pressure
# Geometry Data:
# -------------
xlength 30.0 # domain size in x-direction
ylength 8.0 # domain size in y-direction
imax 400 # number of interior cells in x-direction
jmax 200 # number of interior cells in y-direction
# Time Data:
# ---------
te 150.0 # final time
dt 0.02 # time stepsize
tau 0.5 # safety factor for time stepsize control (<0 constant delt)
# Pressure Iteration Data:
# -----------------------
itermax 200 # maximal number of pressure iteration in one time step
eps 0.001 # stopping tolerance for pressure iteration
rho 0.52
omg 1.75 # relaxation parameter for SOR iteration
gamma 0.9 # upwind differencing factor gamma
# Particle Tracing Data:
# -----------------------
numberOfParticles 200
startTime 50
injectTimePeriod 1.0
writeTimePeriod 0.5
x1 0.0
y1 3.8
x2 0.0
y2 4.1
# Obstacle Geometry Data:
# -----------------------
# Shape 0 disable, 1 Rectangle/Square, 2 Circle
shape 2
xCenter 5.0
yCenter 4.0
xRectLength 2.0
yRectLength 1.0
circleRadius 1.0
#===============================================================================

File diff suppressed because it is too large Load Diff

View File

@ -1,17 +0,0 @@
#!/bin/bash
rm data.csv
echo "Rank, SOR, RB" >> data.csv
for i in {1..72}
do
res=$(mpirun -n $i ./exe-ICC dcavity.par 1 | grep "Solution took" | cut -c 14- )
res2=$(mpirun -n $i ./exe-ICC dcavity.par 2 | grep "Solution took" | cut -c 14- )
s="$i, $res, $res2"
s=${s//"s"/}
echo $s >> data.csv
done

Some files were not shown because too many files have changed in this diff Show More