From 76d49c72d6ca263e109df9358ff6fdcbe1d0eef8 Mon Sep 17 00:00:00 2001
From: Matthieu Schaller <matthieu.schaller@durham.ac.uk>
Date: Sun, 13 May 2018 16:22:37 +0200
Subject: [PATCH] In the gravity force checks read the Ewald correction table
from the file if the file exists.
---
src/gravity.c | 336 ++++++++++++++++++++++++++++++--------------------
src/gravity.h | 2 +
2 files changed, 206 insertions(+), 132 deletions(-)
diff --git a/src/gravity.c b/src/gravity.c
index 06a1e94167..42cf1dc4bb 100644
--- a/src/gravity.c
+++ b/src/gravity.c
@@ -73,155 +73,228 @@ float ewald_fac;
void gravity_exact_force_ewald_init(double boxSize) {
#ifdef SWIFT_GRAVITY_FORCE_CHECKS
- const ticks tic = getticks();
- message("Computing Ewald correction table...");
-
- /* Level of correction (Hernquist et al. 1991)*/
- const float alpha = 2.f;
-
- /* some useful constants */
- const float alpha2 = alpha * alpha;
- const float factor_exp1 = 2.f * alpha / sqrt(M_PI);
- const float factor_exp2 = -M_PI * M_PI / alpha2;
- const float factor_sin = 2.f * M_PI;
- const float factor_cos = 2.f * M_PI;
- const float factor_pot = M_PI / alpha2;
+
const float boxSize_inv2 = 1.f / (boxSize * boxSize);
- /* Ewald factor to access the table */
- ewald_fac = (double)(2 * Newald) / boxSize;
+ int use_file = 0;
- /* Zero everything */
- bzero(fewald_x, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
- bzero(fewald_y, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
- bzero(fewald_z, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
- bzero(potewald, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
+#ifdef HAVE_HDF5
+ if (access("Ewald.hdf5", R_OK) != -1) use_file = 1;
+#endif
- potewald[0][0][0] = 2.8372975f;
+ /* Can we use the stored HDF5 file? */
+ if (use_file) {
- /* Compute the values in one of the octants */
- for (int i = 0; i <= Newald; ++i) {
- for (int j = 0; j <= Newald; ++j) {
- for (int k = 0; k <= Newald; ++k) {
+ const ticks tic = getticks();
+ message("Reading Ewald correction table from file...");
- if (i == 0 && j == 0 && k == 0) continue;
-
- /* Distance vector */
- const float r_x = 0.5f * ((float)i) / Newald;
- const float r_y = 0.5f * ((float)j) / Newald;
- const float r_z = 0.5f * ((float)k) / Newald;
-
- /* Norm of distance vector */
- const float r2 = r_x * r_x + r_y * r_y + r_z * r_z;
- const float r_inv = 1.f / sqrtf(r2);
- const float r_inv3 = r_inv * r_inv * r_inv;
-
- /* Normal gravity potential term */
- float f_x = r_x * r_inv3;
- float f_y = r_y * r_inv3;
- float f_z = r_z * r_inv3;
- float pot = r_inv + factor_pot;
-
- for (int n_i = -4; n_i <= 4; ++n_i) {
- for (int n_j = -4; n_j <= 4; ++n_j) {
- for (int n_k = -4; n_k <= 4; ++n_k) {
-
- const float d_x = r_x - n_i;
- const float d_y = r_y - n_j;
- const float d_z = r_z - n_k;
-
- /* Discretised distance */
- const float r_tilde2 = d_x * d_x + d_y * d_y + d_z * d_z;
- const float r_tilde_inv = 1.f / sqrtf(r_tilde2);
- const float r_tilde = r_tilde_inv * r_tilde2;
- const float r_tilde_inv3 =
- r_tilde_inv * r_tilde_inv * r_tilde_inv;
-
- const float val_pot = erfcf(alpha * r_tilde);
-
- const float val_f =
- val_pot + factor_exp1 * r_tilde * expf(-alpha2 * r_tilde2);
-
- /* First correction term */
- const float f = val_f * r_tilde_inv3;
- f_x -= f * d_x;
- f_y -= f * d_y;
- f_z -= f * d_z;
- pot -= val_pot * r_tilde_inv;
+#ifdef HAVE_HDF5
+ const hid_t h_file = H5Fopen("Ewald.hdf5", H5F_ACC_RDONLY, H5P_DEFAULT);
+ if (h_file < 0) error("Error opening the old 'Ewald.hdf5' file.");
+
+ /* Check the table size */
+ const hid_t h_grp = H5Gopen1(h_file, "Info");
+ const hid_t h_attr = H5Aopen(h_grp, "Ewald_size", H5P_DEFAULT);
+ int size;
+ H5Aread(h_attr, H5T_NATIVE_INT, &size);
+ if (size != Newald)
+ error("File 'Ewald.hdf5' contains arrays of the wrong size");
+ H5Aclose(h_attr);
+ H5Gclose(h_grp);
+
+ /* Now read the tables themselves */
+ hid_t h_data;
+ h_data = H5Dopen(h_file, "Ewald_x", H5P_DEFAULT);
+ H5Dread(h_data, H5T_NATIVE_FLOAT, H5S_ALL, H5S_ALL, H5P_DEFAULT,
+ &(fewald_x[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dopen(h_file, "Ewald_y", H5P_DEFAULT);
+ H5Dread(h_data, H5T_NATIVE_FLOAT, H5S_ALL, H5S_ALL, H5P_DEFAULT,
+ &(fewald_y[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dopen(h_file, "Ewald_z", H5P_DEFAULT);
+ H5Dread(h_data, H5T_NATIVE_FLOAT, H5S_ALL, H5S_ALL, H5P_DEFAULT,
+ &(fewald_z[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dopen(h_file, "Ewald_pot", H5P_DEFAULT);
+ H5Dread(h_data, H5T_NATIVE_FLOAT, H5S_ALL, H5S_ALL, H5P_DEFAULT,
+ &(potewald[0][0][0]));
+ H5Dclose(h_data);
+
+ /* Done */
+ H5Fclose(h_file);
+#endif
+
+ /* Report time this took */
+ message("Ewald correction table read in (took %.3f %s). ",
+ clocks_from_ticks(getticks() - tic), clocks_getunit());
+ } else {
+
+ /* Ok.. let's recompute everything */
+ const ticks tic = getticks();
+ message("Computing Ewald correction table...");
+
+ /* Level of correction (Hernquist et al. 1991)*/
+ const float alpha = 2.f;
+
+ /* some useful constants */
+ const float alpha2 = alpha * alpha;
+ const float factor_exp1 = 2.f * alpha / sqrt(M_PI);
+ const float factor_exp2 = -M_PI * M_PI / alpha2;
+ const float factor_sin = 2.f * M_PI;
+ const float factor_cos = 2.f * M_PI;
+ const float factor_pot = M_PI / alpha2;
+
+ /* Ewald factor to access the table */
+ ewald_fac = (double)(2 * Newald) / boxSize;
+
+ /* Zero everything */
+ bzero(fewald_x, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
+ bzero(fewald_y, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
+ bzero(fewald_z, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
+ bzero(potewald, (Newald + 1) * (Newald + 1) * (Newald + 1) * sizeof(float));
+
+ potewald[0][0][0] = 2.8372975f;
+
+ /* Compute the values in one of the octants */
+ for (int i = 0; i <= Newald; ++i) {
+ for (int j = 0; j <= Newald; ++j) {
+ for (int k = 0; k <= Newald; ++k) {
+
+ if (i == 0 && j == 0 && k == 0) continue;
+
+ /* Distance vector */
+ const float r_x = 0.5f * ((float)i) / Newald;
+ const float r_y = 0.5f * ((float)j) / Newald;
+ const float r_z = 0.5f * ((float)k) / Newald;
+
+ /* Norm of distance vector */
+ const float r2 = r_x * r_x + r_y * r_y + r_z * r_z;
+ const float r_inv = 1.f / sqrtf(r2);
+ const float r_inv3 = r_inv * r_inv * r_inv;
+
+ /* Normal gravity potential term */
+ float f_x = r_x * r_inv3;
+ float f_y = r_y * r_inv3;
+ float f_z = r_z * r_inv3;
+ float pot = r_inv + factor_pot;
+
+ for (int n_i = -4; n_i <= 4; ++n_i) {
+ for (int n_j = -4; n_j <= 4; ++n_j) {
+ for (int n_k = -4; n_k <= 4; ++n_k) {
+
+ const float d_x = r_x - n_i;
+ const float d_y = r_y - n_j;
+ const float d_z = r_z - n_k;
+
+ /* Discretised distance */
+ const float r_tilde2 = d_x * d_x + d_y * d_y + d_z * d_z;
+ const float r_tilde_inv = 1.f / sqrtf(r_tilde2);
+ const float r_tilde = r_tilde_inv * r_tilde2;
+ const float r_tilde_inv3 =
+ r_tilde_inv * r_tilde_inv * r_tilde_inv;
+
+ const float val_pot = erfcf(alpha * r_tilde);
+
+ const float val_f =
+ val_pot + factor_exp1 * r_tilde * expf(-alpha2 * r_tilde2);
+
+ /* First correction term */
+ const float f = val_f * r_tilde_inv3;
+ f_x -= f * d_x;
+ f_y -= f * d_y;
+ f_z -= f * d_z;
+ pot -= val_pot * r_tilde_inv;
+ }
}
}
- }
- for (int h_i = -4; h_i <= 4; ++h_i) {
- for (int h_j = -4; h_j <= 4; ++h_j) {
- for (int h_k = -4; h_k <= 4; ++h_k) {
+ for (int h_i = -4; h_i <= 4; ++h_i) {
+ for (int h_j = -4; h_j <= 4; ++h_j) {
+ for (int h_k = -4; h_k <= 4; ++h_k) {
- const float h2 = h_i * h_i + h_j * h_j + h_k * h_k;
+ const float h2 = h_i * h_i + h_j * h_j + h_k * h_k;
- if (h2 == 0.f) continue;
+ if (h2 == 0.f) continue;
- const float h2_inv = 1.f / (h2 + FLT_MIN);
- const float h_dot_x = h_i * r_x + h_j * r_y + h_k * r_z;
+ const float h2_inv = 1.f / (h2 + FLT_MIN);
+ const float h_dot_x = h_i * r_x + h_j * r_y + h_k * r_z;
- const float common = h2_inv * expf(h2 * factor_exp2);
+ const float common = h2_inv * expf(h2 * factor_exp2);
- const float val_pot =
- (float)M_1_PI * common * cosf(factor_cos * h_dot_x);
+ const float val_pot =
+ (float)M_1_PI * common * cosf(factor_cos * h_dot_x);
- const float val_f = 2.f * common * sinf(factor_sin * h_dot_x);
+ const float val_f = 2.f * common * sinf(factor_sin * h_dot_x);
- /* Second correction term */
- f_x -= val_f * h_i;
- f_y -= val_f * h_j;
- f_z -= val_f * h_k;
- pot -= val_pot;
+ /* Second correction term */
+ f_x -= val_f * h_i;
+ f_y -= val_f * h_j;
+ f_z -= val_f * h_k;
+ pot -= val_pot;
+ }
}
}
- }
- /* Save back to memory */
- fewald_x[i][j][k] = f_x;
- fewald_y[i][j][k] = f_y;
- fewald_z[i][j][k] = f_z;
- potewald[i][j][k] = pot;
+ /* Save back to memory */
+ fewald_x[i][j][k] = f_x;
+ fewald_y[i][j][k] = f_y;
+ fewald_z[i][j][k] = f_z;
+ potewald[i][j][k] = pot;
+ }
}
}
- }
/* Dump the Ewald table to a file */
#ifdef HAVE_HDF5
- hid_t h_file =
- H5Fcreate("Ewald.hdf5", H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
- if (h_file < 0) error("Error while opening file for Ewald dump.");
-
- /* Create dataspace */
- hsize_t dim[3] = {Newald + 1, Newald + 1, Newald + 1};
- hid_t h_space = H5Screate_simple(3, dim, NULL);
- hid_t h_data;
- h_data = H5Dcreate(h_file, "Ewald_x", H5T_NATIVE_FLOAT, h_space, H5P_DEFAULT,
- H5P_DEFAULT, H5P_DEFAULT);
- H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
- &(fewald_x[0][0][0]));
- H5Dclose(h_data);
- h_data = H5Dcreate(h_file, "Ewald_y", H5T_NATIVE_FLOAT, h_space, H5P_DEFAULT,
- H5P_DEFAULT, H5P_DEFAULT);
- H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
- &(fewald_y[0][0][0]));
- H5Dclose(h_data);
- h_data = H5Dcreate(h_file, "Ewald_z", H5T_NATIVE_FLOAT, h_space, H5P_DEFAULT,
- H5P_DEFAULT, H5P_DEFAULT);
- H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
- &(fewald_z[0][0][0]));
- H5Dclose(h_data);
- h_data = H5Dcreate(h_file, "Ewald_pot", H5T_NATIVE_FLOAT, h_space,
- H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
- H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
- &(potewald[0][0][0]));
- H5Dclose(h_data);
- H5Sclose(h_space);
- H5Fclose(h_file);
+ hid_t h_file =
+ H5Fcreate("Ewald.hdf5", H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
+ if (h_file < 0) error("Error while opening file for Ewald dump.");
+
+ /* Write the Ewald table size */
+ const int size = Newald;
+ const hid_t h_grp = H5Gcreate1(h_file, "Info", 0);
+ const hid_t h_aspace = H5Screate(H5S_SCALAR);
+ hid_t h_att =
+ H5Acreate1(h_grp, "Ewald_size", H5T_NATIVE_INT, h_aspace, H5P_DEFAULT);
+ H5Awrite(h_att, H5T_NATIVE_INT, &size);
+ H5Aclose(h_att);
+ H5Gclose(h_grp);
+ H5Sclose(h_aspace);
+
+ /* Create dataspace and write arrays */
+ hsize_t dim[3] = {Newald + 1, Newald + 1, Newald + 1};
+ hid_t h_space = H5Screate_simple(3, dim, NULL);
+ hid_t h_data;
+ h_data = H5Dcreate(h_file, "Ewald_x", H5T_NATIVE_FLOAT, h_space,
+ H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+ H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
+ &(fewald_x[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dcreate(h_file, "Ewald_y", H5T_NATIVE_FLOAT, h_space,
+ H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+ H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
+ &(fewald_y[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dcreate(h_file, "Ewald_z", H5T_NATIVE_FLOAT, h_space,
+ H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+ H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
+ &(fewald_z[0][0][0]));
+ H5Dclose(h_data);
+ h_data = H5Dcreate(h_file, "Ewald_pot", H5T_NATIVE_FLOAT, h_space,
+ H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+ H5Dwrite(h_data, H5T_NATIVE_FLOAT, h_space, H5S_ALL, H5P_DEFAULT,
+ &(potewald[0][0][0]));
+ H5Dclose(h_data);
+ H5Sclose(h_space);
+ H5Fclose(h_file);
#endif
+ /* Report time this took */
+ message("Ewald correction table computed (took %.3f %s). ",
+ clocks_from_ticks(getticks() - tic), clocks_getunit());
+ }
+
/* Apply the box-size correction */
for (int i = 0; i <= Newald; ++i) {
for (int j = 0; j <= Newald; ++j) {
@@ -233,16 +306,11 @@ void gravity_exact_force_ewald_init(double boxSize) {
}
}
}
-
- /* Say goodbye */
- message("Ewald correction table computed (took %.3f %s). ",
- clocks_from_ticks(getticks() - tic), clocks_getunit());
#else
error("Gravity checking function called without the corresponding flag.");
#endif
}
-#ifdef SWIFT_GRAVITY_FORCE_CHECKS
/**
* @brief Compute the Ewald correction for a given distance vector r.
*
@@ -255,9 +323,10 @@ void gravity_exact_force_ewald_init(double boxSize) {
* @param corr_f (return) The Ewald correction for the force.
* @param corr_p (return) The Ewald correction for the potential.
*/
-__attribute__((always_inline)) INLINE static void
-gravity_exact_force_ewald_evaluate(double rx, double ry, double rz,
- double corr_f[3], double *corr_p) {
+void gravity_exact_force_ewald_evaluate(double rx, double ry, double rz,
+ double corr_f[3], double *corr_p) {
+
+#ifdef SWIFT_GRAVITY_FORCE_CHECKS
const double s_x = (rx < 0.) ? 1. : -1.;
const double s_y = (ry < 0.) ? 1. : -1.;
@@ -327,8 +396,11 @@ gravity_exact_force_ewald_evaluate(double rx, double ry, double rz,
*corr_p += potewald[i + 1][j + 0][k + 1] * dx * ty * dz;
*corr_p += potewald[i + 1][j + 1][k + 0] * dx * dy * tz;
*corr_p += potewald[i + 1][j + 1][k + 1] * dx * dy * dz;
-}
+
+#else
+ error("Gravity checking function called without the corresponding flag.");
#endif
+}
/**
* @brief Checks whether the file containing the exact accelerations for
diff --git a/src/gravity.h b/src/gravity.h
index 33bef8ba32..073b0b0532 100644
--- a/src/gravity.h
+++ b/src/gravity.h
@@ -37,6 +37,8 @@ struct space;
void gravity_exact_force_ewald_init(double boxSize);
void gravity_exact_force_ewald_free();
+void gravity_exact_force_ewald_evaluate(double rx, double ry, double rz,
+ double corr_f[3], double *corr_p);
void gravity_exact_force_compute(struct space *s, const struct engine *e);
void gravity_exact_force_check(struct space *s, const struct engine *e,
float rel_tol);
--
GitLab