idefix/src/timeIntegrator.cpp at master · idefix-code/idefix · GitHub

Repository navigation

History

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

// ***********************************************************************************

// Idefix MHD astrophysical code

// Source file src/timeIntegrator.cpp

// Last modified : 09/2026

// Copyright(C) by :

// - Geoffroy Lesur <geoffroy.lesur@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2020 - 2026)

// - Soufiane Baghdadi <soufiane.baghdadi@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2020 - 2021)

// - Clément Robert <clement.robert@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2021 - 2022)

// - Gaylor Wafflard <gaylor.wafflard@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2024)

// - Sébastien Valat <sebastien.valat@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2026)

// - Nicolas Scepi <nicolas.scepi@univ-grenoble-alpes.fr> (IPAG/UGA/CNRS - 2026)

// and other code contributors

// Licensed under CeCILL 2.1 License, see COPYING for more information

// ***********************************************************************************

//#define WITH_TEMPERATURE_SENSOR

#include <cstdio>

#include <iomanip>

#include <string>

#include <vector>

#include "idefix.hpp"

#include "timeIntegrator.hpp"

#include "input.hpp"

#include "dataBlock.hpp"

#include "stateContainer.hpp"

#include "fluid.hpp"

#include "planetarySystem.hpp"

TimeIntegrator::TimeIntegrator(Input & input, DataBlock & data) {

idfx::pushRegion("TimeIntegrator::TimeIntegrator(Input...)");

this->timer.reset();

this->lastLog=timer.seconds();

this->lastMpiLog=idfx::mpiCallsTimer + idfx::mpiCallsTimer;

nstages=input.Get<int>("TimeIntegrator","nstages",0);

if(input.CheckEntry("TimeIntegrator","fixed_dt")>0) {

this->haveFixedDt = true;

this->fixedDt = input.Get<real>("TimeIntegrator","fixed_dt",0);

data.dt=fixedDt;

}

if(!haveFixedDt) {

cfl=input.Get<real>("TimeIntegrator","CFL",0);

cflMaxVar = input.GetOrSet<real>("TimeIntegrator","CFL_max_var",0, 1.1);

data.dt = input.GetOrSet<real>("TimeIntegrator","first_dt",0, 1.0e-10);

}

this->cyclePeriod = input.GetOrSet<int>("Output","log",0, 100);

this->maxRuntime = 3600*input.GetOrSet<double>("TimeIntegrator","max_runtime",0.0,-1.0);

// check nan periodicity every 100 loops

this->checkNanPeriodicity = input.GetOrSet<int>("TimeIntegrator","check_nan", 0, 100);

#ifndef SINGLE_PRECISION

const real maxdivBDefault = 1e-6;

#else

const real maxdivBDefault = 1e-2;

#endif

this->maxdivB = input.GetOrSet<real>("TimeIntegrator","maxdivB", 0,maxdivBDefault);

data.t=0.0;

ncycles=0;

if(nstages==2) {

wc[0] = 0.5;

w0[0] = 0.5;

}

if(nstages==3) {

wc[0] = 0.25;

w0[0] = 0.75;

wc[1] = 2.0/3.0;

w0[1] = 1.0/3.0;

}

// Init the RKL scheme if it's needed

if(data.hydro->haveRKLParabolicTerms) {

haveRKL = true;

}

// If multi-stage, create a new state in the datablock called "begin"

if(nstages>1) {

data.states["begin"] = StateContainer();

data.states["begin"].AllocateAs(data.states["current"]);

}

idfx::popRegion();

}

void TimeIntegrator::ShowLog(DataBlock &data) {

if(isSilent) return;

double rawperf = (timer.seconds()-lastLog)/data.mygrid->np_int[IDIR]/data.mygrid->np_int[JDIR]

/data.mygrid->np_int[KDIR]/cyclePeriod * idfx::psize;

#ifdef WITH_MPI

// measure time spent in expensive MPI calls

double mpiCycleTime = idfx::mpiCallsTimer - lastMpiLog;

// reduce to an normalized overhead in %

double mpiOverhead = 100.0 * mpiCycleTime / (timer.seconds() - lastLog);

lastMpiLog = idfx::mpiCallsTimer;

#endif

double sgOverhead;

if(data.haveGravity && data.gravity->haveSelfGravityPotential) {

double sgCycleTime = data.gravity->selfGravity.elapsedTime - lastSGLog;

sgOverhead = 100.0 * sgCycleTime / (timer.seconds() - lastLog);

lastSGLog = data.gravity->selfGravity.elapsedTime;

}

lastLog = timer.seconds();

int col_width{16};

if (ncycles == 0) {

idfx::cout << "TimeIntegrator: ";

idfx::cout << std::setw(col_width) << "time";

idfx::cout << " | " << std::setw(col_width) << "cycle";

idfx::cout << " | " << std::setw(col_width) << "time step";

idfx::cout << " | " << std::setw(col_width) << "cell (updates/s)";

#ifdef WITH_MPI

idfx::cout << " | " << std::setw(col_width) << "MPI overhead (%)";

if(idfx::prank==0) {

idfx::cout << " | " << std::setw(col_width) << "MPI imbalance(%)";

}

#endif

#if MHD == YES

idfx::cout << " | " << std::setw(col_width) << "div B";

#endif

if(haveRKL) {

idfx::cout << " | " << std::setw(col_width) << "RKL stages";

}

if(data.haveGravity && data.gravity->haveSelfGravityPotential) {

idfx::cout << " | " << std::setw(col_width) << "SG iterations";

idfx::cout << " | " << std::setw(col_width) << "SG error";

idfx::cout << " | " << std::setw(col_width) << "SG overhead (%)";

}

idfx::cout << std::endl;

}

#ifdef WITH_MPI

double imbalance = 0;

if(ncycles>=cyclePeriod) imbalance = ComputeBalance();

#endif

idfx::cout << "TimeIntegrator: ";

idfx::cout << std::scientific;

idfx::cout << std::setw(col_width) << data.t;

idfx::cout << " | " << std::setw(col_width) << ncycles;

idfx::cout << " | " << std::setw(col_width) << data.dt;

if(ncycles>=cyclePeriod) {

idfx::cout << " | " << std::setw(col_width) << 1 / rawperf;

#ifdef WITH_MPI

idfx::cout << std::fixed;

idfx::cout << " | " << std::setw(col_width) << mpiOverhead;

if(idfx::prank==0) {

idfx::cout << " | " << std::setw(col_width) << imbalance;

}

#endif

} else {

idfx::cout << " | " << std::setw(col_width) << "N/A";

#if WITH_MPI

idfx::cout << " | " << std::setw(col_width) << "N/A";

if(idfx::prank==0) {

idfx::cout << " | " << std::setw(col_width) << "N/A";

}

#endif

}

#if MHD == YES

// Check divB

real divB = data.hydro->CheckDivB();

idfx::cout << std::scientific;

idfx::cout << " | " << std::setw(col_width) << divB;

if(divB>maxdivB) {

std::stringstream msg;

msg << std::endl << "divB too large, check your calculation";

throw std::runtime_error(msg.str());

}

#endif

if(haveRKL) {

idfx::cout << " | " << std::setw(col_width) << data.hydro->rkl->stage;

}

if(data.haveGravity && data.gravity->haveSelfGravityPotential) {

if(ncycles>=cyclePeriod) {

idfx::cout << " | " << std::setw(col_width) << data.gravity->selfGravity.nsteps;

idfx::cout << std::scientific;

idfx::cout << " | " << std::setw(col_width) << data.gravity->selfGravity.currentError;

idfx::cout << std::fixed;

idfx::cout << " | " << std::setw(col_width) << sgOverhead;

} else {

idfx::cout << " | " << std::setw(col_width) << "N/A";

}

idfx::cout << std::endl;

}

double TimeIntegrator::ComputeBalance() {

// Check MPI imbalance

double imbalance = 0;

#ifdef WITH_MPI

const double allowedImbalance = 20.0;

std::vector<double> computeLogPerCore(idfx::psize);

idfx::MPI_Gather(&computeLastLog, 1, MPI_DOUBLE, computeLogPerCore, 1, MPI_DOUBLE, 0,

MPI_COMM_WORLD);

computeLastLog = 0; // reset timer for all cores

if(idfx::prank==0) {

// Compute the average, the min and the max

double computeMin = computeLogPerCore[0];

double computeMax = computeLogPerCore[0];

double computeMean = 0;

for(int i = 0 ; i < idfx::psize ; i++) {

computeMean += computeLogPerCore[i];

if(computeLogPerCore[i]>computeMax) {

computeMax = computeLogPerCore[i];

}

if(computeLogPerCore[i]<computeMin) {

computeMin = computeLogPerCore[i];

}

computeMean /= idfx::psize;

imbalance = (computeMax-computeMin)/computeMean*100;

// only show warnings if the imbalance is higher than 20%on GPUs

#if defined(KOKKOS_ENABLE_CUDA) || defined(KOKKOS_ENABLE_HIP) || defined(KOKKOS_ENABLE_SYCL)

if(imbalance>allowedImbalance ) {

idfx::cout << "-------------------------------------------------------------"

<< std::endl;

idfx::cout << "Warning: MPI imbalance found in this run " << std::endl;

idfx::cout << std::fixed;

for(int i = 0 ; i < idfx::psize ; i++) {

if(computeLogPerCore[i]/computeMean - 1> allowedImbalance/2/100) {

idfx::cout << "+" << 100*(computeLogPerCore[i]/computeMean-1)

<< "% (proc " << i << ")" << std::endl;

}

if(1-computeLogPerCore[i]/computeMean > allowedImbalance/2/100) {

idfx::cout << "-" << 100*(1-computeLogPerCore[i]/computeMean)

<< "% (proc " << i << ")" << std::endl;

}

idfx::cout << "You should probably check these nodes are running properly."

<< std::endl;

idfx::cout << "-------------------------------------------------------------"

<< std::endl;

}

#endif

}

#endif

return(imbalance);

}

// Compute one full cycle of the time Integrator

void TimeIntegrator::Cycle(DataBlock &data) {

// Do one cycle

IdefixArray3D<real> InvDt = data.hydro->InvDt;

real newdt;

idfx::pushRegion("TimeIntegrator::Cycle");

if(ncycles%cyclePeriod==0) ShowLog(data);

// Launch user step before everything

data.LaunchUserStepFirst();

if(haveRKL && (ncycles%2)==1) { // Runge-Kutta-Legendre cycle

data.EvolveRKLStage();

}

// save t at the begining of the cycle

const real t0 = data.t;

// Reinit datablock for a new stage

data.ResetStage();

#ifdef WITH_MPI

MPI_Request dtReduce;

#endif

/////////////////////////////////////////////////

// BEGIN STAGES LOOP //

/////////////////////////////////////////////////

for(int stage=0; stage < nstages ; stage++) {

// Apply Boundary conditions

data.SetBoundaries();

// Remove Fargo velocity so that the integrator works on the residual

if(data.haveFargo) data.fargo->SubstractVelocity(data.t);

// Convert current state into conservative variable and save it

data.PrimToCons();

// Store (deep copy) initial stage for multi-stage time integrators

if(nstages>1 && stage==0) {

data.states["begin"].CopyFrom(data.states["current"]);

}

// If gravity is needed, update it

if(data.haveGravity) {

if(ncycles % data.gravity->skipGravity == 0) data.gravity->ComputeGravity(ncycles);

}

Kokkos::fence();

computeLastLog -= timer.seconds();

// Update Uc & Vs

data.EvolveStage();

Kokkos::fence();

computeLastLog += timer.seconds();

// evolve dt accordingly

data.t += data.dt;

// Compute next time_step during first stage

if(stage==0) {

if(!haveFixedDt) {

newdt = cfl*data.ComputeTimestep();

#ifdef WITH_MPI

if(idfx::psize>1) {

MPI_SAFE_CALL(MPI_Iallreduce(MPI_IN_PLACE, &newdt, 1, realMPI, MPI_MIN, MPI_COMM_WORLD,

&dtReduce));

}

#endif

}

// Is this not the first stage?

if(stage>0) {

// do the partial evolution required by the multi-step

real wcs=wc[stage-1];

real w0s=w0[stage-1];

data.states["current"].AddAndStore(wcs, w0s, data.states["begin"]);

// update t

data.t = wcs*data.t + w0s*t0;

}

// Shift solution according to fargo if this is our last stage

if(data.haveFargo && stage==nstages-1) {

data.fargo->ShiftSolution(t0,data.dt);

}

// Coarsen conservative variables once they have been evolved

if(data.haveGridCoarsening) {

data.Coarsen();

}

// Back to using Vc

data.ConsToPrim();

// Add back fargo velocity so that boundary conditions are applied on the total V

if(data.haveFargo) data.fargo->AddVelocity(data.t);

}

/////////////////////////////////////////////////

// END STAGES LOOP //

/////////////////////////////////////////////////

// Wait for dt MPI reduction

#ifdef WITH_MPI

if(!haveFixedDt && idfx::psize>1) {

MPI_SAFE_CALL(MPI_Wait(&dtReduce, MPI_STATUS_IGNORE));

}

#endif

if(haveRKL && (ncycles%2)==0) { // Runge-Kutta-Legendre cycle

data.EvolveRKLStage();

}

// Update planet position

if(data.haveplanetarySystem) {

data.planetarySystem->EvolveSystem(data, data.dt);

}

// Coarsen the grid

if(data.haveGridCoarsening) {

data.Coarsen();

}

// Launch user step last

data.LaunchUserStepLast();

// Look for Nans every now and then (this actually cost a lot of time on GPUs

// because streams are divergent)

if(ncycles%checkNanPeriodicity==0) {

if(data.CheckNan()>0) {

throw std::runtime_error(std::string("Nan found after integration cycle"));

}

// Update current time (should have already been done, but this gets rid of roundoff errors)

data.t=t0+data.dt;

if(haveRKL) {

// update next time step

real tt = newdt/data.hydro->rkl->dt;

newdt *= std::fmin(ONE_F, data.hydro->rkl->rmax_par/(tt));

}

// Next time step

if(!haveFixedDt) {

if(newdt>cflMaxVar*data.dt) {

data.dt=cflMaxVar*data.dt;

} else {

if(ncycles==0 && newdt < 0.5*data.dt) {

std::stringstream msg;

msg << "Your guessed first_dt is too large. My next dt=" << newdt << std::endl;

msg << "Try to reduce first_dt in the ini file.";

IDEFIX_ERROR(msg);

}

data.dt=newdt;

}

if(data.dt < 1e-15) {

std::stringstream msg;

msg << "dt = " << data.dt << " is too small.";

throw std::runtime_error(msg.str());

}

} else {

data.dt = fixedDt;

}

ncycles++;

idfx::popRegion();

}

int64_t TimeIntegrator::GetNCycles() {

return(ncycles);

}

// Check whether our maximumruntime has been reached. Reduce the results on all of the cores

// to make sure they stop simultaneously even if running time are not perfectly in sync

bool TimeIntegrator::CheckForMaxRuntime() {

idfx::pushRegion("TimeIntegrator::CheckForMaxRuntime");

// if maxRuntime is negative, this function is disabled (default)

if(this->maxRuntime < 0) {

idfx::popRegion();

return(false);

}

double runtime = timer.seconds();

bool runtimeReached{false};

#ifdef WITH_MPI

int runtimeValue = 0;

if(runtime >= this->maxRuntime) runtimeValue = 1;

MPI_Bcast(&runtimeValue, 1, MPI_INT, 0, MPI_COMM_WORLD);

runtimeReached = runtimeValue > 0;

#else

runtimeReached = runtime >= this->maxRuntime;

#endif

if(runtimeReached) {

idfx::cout << "TimeIntegrator:CheckForMaxRuntime: Maximum runtime reached."

<< std::endl;

}

idfx::popRegion();

return(runtimeReached);

}

void TimeIntegrator::ShowConfig() {

if(nstages==1) {

idfx::cout << "TimeIntegrator: using 1st Order (EULER) integrator." << std::endl;

} else if(nstages==2) {

idfx::cout << "TimeIntegrator: using 2nd Order (RK2) integrator." << std::endl;

} else if(nstages==3) {

idfx::cout << "TimeIntegrator: using 3rd Order (RK3) integrator." << std::endl;

} else {

IDEFIX_ERROR("Unknown time integrator");

}

if(haveFixedDt) {

idfx::cout << "TimeIntegrator: Using fixed dt=" << fixedDt << ". Ignoring CFL and first_dt."

<< std::endl;

} else {

idfx::cout << "TimeIntegrator: Using adaptive dt with CFL=" << cfl << " ." << std::endl;

}

if(maxRuntime>0) {

idfx::cout << "TimeIntegrator: will stop after " << maxRuntime/3600 << " hours." << std::endl;

}

idefix/src/timeIntegrator.cpp at master · idefix-code/idefix · GitHub

Repository navigation

Expand file tree

Breadcrumbs

timeIntegrator.cpp

Latest commit

History

Breadcrumbs

timeIntegrator.cpp

File metadata and controls