Differential D7475 Diff 23907 source/blender/blenkernel/intern/ocean.c

Changeset View

Standalone View

source/blender/blenkernel/intern/ocean.c

Show First 20 Lines • Show All 447 Lines • ▼ Show 20 Lines	for (j = 0; j <= o->_N / 2; j++) {
mul_complex_c(exp_param1, o->_h0[i * o->_N + j], exp_param1);		mul_complex_c(exp_param1, o->_h0[i * o->_N + j], exp_param1);
mul_complex_c(exp_param2, conj_param, exp_param2);		mul_complex_c(exp_param2, conj_param, exp_param2);

add_comlex_c(o->_htilda[i * (1 + o->_N / 2) + j], exp_param1, exp_param2);		add_comlex_c(o->_htilda[i * (1 + o->_N / 2) + j], exp_param1, exp_param2);
mul_complex_f(o->_fft_in[i * (1 + o->_N / 2) + j], o->_htilda[i * (1 + o->_N / 2) + j], scale);		mul_complex_f(o->_fft_in[i * (1 + o->_N / 2) + j], o->_htilda[i * (1 + o->_N / 2) + j], scale);
}		}
}		}

static void ocean_compute_displacement_y(TaskPool *__restrict pool,		static void ocean_compute_displacement_y(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;

fftw_execute(o->_disp_y_plan);		fftw_execute(o->_disp_y_plan);
}		}

static void ocean_compute_displacement_x(TaskPool *__restrict pool,		static void ocean_compute_displacement_x(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
const float scale = osd->scale;		const float scale = osd->scale;
const float chop_amount = osd->chop_amount;		const float chop_amount = osd->chop_amount;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;
Show All 10 Lines	for (j = 0; j <= o->_N / 2; j++) {
0.0f :		0.0f :
o->_kx[i] / o->_k[i * (1 + o->_N / 2) + j]));		o->_kx[i] / o->_k[i * (1 + o->_N / 2) + j]));
init_complex(o->_fft_in_x[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));		init_complex(o->_fft_in_x[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));
}		}
}		}
fftw_execute(o->_disp_x_plan);		fftw_execute(o->_disp_x_plan);
}		}

static void ocean_compute_displacement_z(TaskPool *__restrict pool,		static void ocean_compute_displacement_z(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
const float scale = osd->scale;		const float scale = osd->scale;
const float chop_amount = osd->chop_amount;		const float chop_amount = osd->chop_amount;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;
Show All 10 Lines	for (j = 0; j <= o->_N / 2; j++) {
0.0f :		0.0f :
o->_kz[j] / o->_k[i * (1 + o->_N / 2) + j]));		o->_kz[j] / o->_k[i * (1 + o->_N / 2) + j]));
init_complex(o->_fft_in_z[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));		init_complex(o->_fft_in_z[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));
}		}
}		}
fftw_execute(o->_disp_z_plan);		fftw_execute(o->_disp_z_plan);
}		}

static void ocean_compute_jacobian_jxx(TaskPool *__restrict pool,		static void ocean_compute_jacobian_jxx(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
const float chop_amount = osd->chop_amount;		const float chop_amount = osd->chop_amount;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;

Show All 14 Lines	static void ocean_compute_jacobian_jxx(TaskPool __restrict pool, void UNUSED(taskdata))

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j < o->_N; j++) {		for (j = 0; j < o->_N; j++) {
o->_Jxx[i * o->_N + j] += 1.0;		o->_Jxx[i * o->_N + j] += 1.0;
}		}
}		}
}		}

static void ocean_compute_jacobian_jzz(TaskPool *__restrict pool,		static void ocean_compute_jacobian_jzz(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
const float chop_amount = osd->chop_amount;		const float chop_amount = osd->chop_amount;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;

Show All 14 Lines	static void ocean_compute_jacobian_jzz(TaskPool __restrict pool, void UNUSED(taskdata))

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j < o->_N; j++) {		for (j = 0; j < o->_N; j++) {
o->_Jzz[i * o->_N + j] += 1.0;		o->_Jzz[i * o->_N + j] += 1.0;
}		}
}		}
}		}

static void ocean_compute_jacobian_jxz(TaskPool *__restrict pool,		static void ocean_compute_jacobian_jxz(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
const float chop_amount = osd->chop_amount;		const float chop_amount = osd->chop_amount;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;

/* init_complex(mul_param, -scale, 0); */		/* init_complex(mul_param, -scale, 0); */
init_complex(mul_param, -1, 0);		init_complex(mul_param, -1, 0);

mul_complex_f(mul_param, mul_param, chop_amount);		mul_complex_f(mul_param, mul_param, chop_amount);
mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);		mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);
mul_complex_f(mul_param,		mul_complex_f(mul_param,
mul_param,		mul_param,
((o->_k[i * (1 + o->_N / 2) + j] == 0.0f) ?		((o->_k[i * (1 + o->_N / 2) + j] == 0.0f) ?
0.0f :		0.0f :
o->_kx[i] * o->_kz[j] / o->_k[i * (1 + o->_N / 2) + j]));		o->_kx[i] * o->_kz[j] / o->_k[i * (1 + o->_N / 2) + j]));
init_complex(o->_fft_in_jxz[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));		init_complex(o->_fft_in_jxz[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));
}		}
}		}
fftw_execute(o->_Jxz_plan);		fftw_execute(o->_Jxz_plan);
}		}

static void ocean_compute_normal_x(TaskPool *__restrict pool,		static void ocean_compute_normal_x(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;

init_complex(mul_param, 0.0, -1.0);		init_complex(mul_param, 0.0, -1.0);
mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);		mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);
mul_complex_f(mul_param, mul_param, o->_kx[i]);		mul_complex_f(mul_param, mul_param, o->_kx[i]);
init_complex(o->_fft_in_nx[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));		init_complex(o->_fft_in_nx[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));
}		}
}		}
fftw_execute(o->_N_x_plan);		fftw_execute(o->_N_x_plan);
}		}

static void ocean_compute_normal_z(TaskPool *__restrict pool,		static void ocean_compute_normal_z(TaskPool __restrict pool, void UNUSED(taskdata))
void *UNUSED(taskdata),
int UNUSED(threadid))
{		{
OceanSimulateData *osd = BLI_task_pool_userdata(pool);		OceanSimulateData *osd = BLI_task_pool_user_data(pool);
const Ocean *o = osd->o;		const Ocean *o = osd->o;
int i, j;		int i, j;

for (i = 0; i < o->_M; i++) {		for (i = 0; i < o->_M; i++) {
for (j = 0; j <= o->_N / 2; j++) {		for (j = 0; j <= o->_N / 2; j++) {
fftw_complex mul_param;		fftw_complex mul_param;

init_complex(mul_param, 0.0, -1.0);		init_complex(mul_param, 0.0, -1.0);
mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);		mul_complex_c(mul_param, mul_param, o->_htilda[i * (1 + o->_N / 2) + j]);
mul_complex_f(mul_param, mul_param, o->_kz[i]);		mul_complex_f(mul_param, mul_param, o->_kz[i]);
init_complex(o->_fft_in_nz[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));		init_complex(o->_fft_in_nz[i * (1 + o->_N / 2) + j], real_c(mul_param), image_c(mul_param));
}		}
}		}
fftw_execute(o->_N_z_plan);		fftw_execute(o->_N_z_plan);
}		}

void BKE_ocean_simulate(struct Ocean *o, float t, float scale, float chop_amount)		void BKE_ocean_simulate(struct Ocean *o, float t, float scale, float chop_amount)
{		{
TaskScheduler *scheduler = BLI_task_scheduler_get();
TaskPool *pool;		TaskPool *pool;

OceanSimulateData osd;		OceanSimulateData osd;

scale *= o->normalize_factor;		scale *= o->normalize_factor;

osd.o = o;		osd.o = o;
osd.t = t;		osd.t = t;
osd.scale = scale;		osd.scale = scale;
osd.chop_amount = chop_amount;		osd.chop_amount = chop_amount;

pool = BLI_task_pool_create(scheduler, &osd, TASK_PRIORITY_HIGH);		pool = BLI_task_pool_create(&osd, TASK_PRIORITY_HIGH);

BLI_rw_mutex_lock(&o->oceanmutex, THREAD_LOCK_WRITE);		BLI_rw_mutex_lock(&o->oceanmutex, THREAD_LOCK_WRITE);

/* Note about multi-threading here: we have to run a first set of computations (htilda one)		/* Note about multi-threading here: we have to run a first set of computations (htilda one)
* before we can run all others, since they all depend on it.		* before we can run all others, since they all depend on it.
* So we make a first parallelized forloop run for htilda,		* So we make a first parallelized forloop run for htilda,
* and then pack all other computations into a set of parallel tasks.		* and then pack all other computations into a set of parallel tasks.
* This is not optimal in all cases,		* This is not optimal in all cases,
▲ Show 20 Lines • Show All 947 Lines • Show Last 20 Lines