From 5dd4aef0f34adfd7ac17b0656f708ac8587552c7 Mon Sep 17 00:00:00 2001
From: Matthieu Schaller <matthieu.schaller@durham.ac.uk>
Date: Sun, 26 Aug 2018 11:49:20 +0100
Subject: [PATCH] Exchange the multipoles at the same time as the cells when
running with gravity.
---
src/cell.c | 17 +++--
src/cell.h | 13 ++--
src/engine.c | 24 ++++---
src/proxy.c | 196 ++++++++++++++++++++++++++-------------------------
src/proxy.h | 4 +-
5 files changed, 136 insertions(+), 118 deletions(-)
diff --git a/src/cell.c b/src/cell.c
index 78b9fdd2a2..2919839d50 100644
--- a/src/cell.c
+++ b/src/cell.c
@@ -167,14 +167,19 @@ int cell_link_sparts(struct cell *c, struct spart *sparts) {
* @param c The #cell.
* @param pc Pointer to an array of packed cells in which the
* cells will be packed.
+ * @param with_gravity Are we running with gravity and hence need
+ * to exchange multipoles?
*
* @return The number of packed cells.
*/
-int cell_pack(struct cell *restrict c, struct pcell *restrict pc) {
+int cell_pack(struct cell *restrict c, struct pcell *restrict pc,
+ const int with_gravity) {
#ifdef WITH_MPI
/* Start by packing the data of the current cell. */
+ if(with_gravity)
+ pc->multipole = *(c->multipole);
pc->h_max = c->h_max;
pc->ti_hydro_end_min = c->ti_hydro_end_min;
pc->ti_hydro_end_max = c->ti_hydro_end_max;
@@ -196,7 +201,7 @@ int cell_pack(struct cell *restrict c, struct pcell *restrict pc) {
for (int k = 0; k < 8; k++)
if (c->progeny[k] != NULL) {
pc->progeny[k] = count;
- count += cell_pack(c->progeny[k], &pc[count]);
+ count += cell_pack(c->progeny[k], &pc[count], with_gravity);
} else {
pc->progeny[k] = -1;
}
@@ -251,15 +256,19 @@ int cell_pack_tags(const struct cell *c, int *tags) {
* @param pc An array of packed #pcell.
* @param c The #cell in which to unpack the #pcell.
* @param s The #space in which the cells are created.
+ * @param with_gravity Are we running with gravity and hence need
+ * to exchange multipoles?
*
* @return The number of cells created.
*/
int cell_unpack(struct pcell *restrict pc, struct cell *restrict c,
- struct space *restrict s) {
+ struct space *restrict s, const int with_gravity) {
#ifdef WITH_MPI
/* Unpack the current pcell. */
+ if(with_gravity)
+ *(c->multipole) = pc->multipole;
c->h_max = pc->h_max;
c->ti_hydro_end_min = pc->ti_hydro_end_min;
c->ti_hydro_end_max = pc->ti_hydro_end_max;
@@ -304,7 +313,7 @@ int cell_unpack(struct pcell *restrict pc, struct cell *restrict c,
temp->parent = c;
c->progeny[k] = temp;
c->split = 1;
- count += cell_unpack(&pc[pc->progeny[k]], temp, s);
+ count += cell_unpack(&pc[pc->progeny[k]], temp, s, with_gravity);
}
/* Return the total number of unpacked cells. */
diff --git a/src/cell.h b/src/cell.h
index 356e03773c..c209f8d99b 100644
--- a/src/cell.h
+++ b/src/cell.h
@@ -77,6 +77,12 @@ struct link {
*/
struct pcell {
+ /*! This cell's gravity-related tensors */
+ struct gravity_tensors multipole;
+
+ /*! Relative indices of the cell's progeny. */
+ int progeny[8];
+
/*! Maximal smoothing length. */
double h_max;
@@ -116,9 +122,6 @@ struct pcell {
/*! Number of #spart in this cell. */
int scount;
- /*! Relative indices of the cell's progeny. */
- int progeny[8];
-
#ifdef SWIFT_DEBUG_CHECKS
/* Cell ID (for debugging) */
int cellID;
@@ -485,8 +488,8 @@ int cell_mlocktree(struct cell *c);
void cell_munlocktree(struct cell *c);
int cell_slocktree(struct cell *c);
void cell_sunlocktree(struct cell *c);
-int cell_pack(struct cell *c, struct pcell *pc);
-int cell_unpack(struct pcell *pc, struct cell *c, struct space *s);
+int cell_pack(struct cell *c, struct pcell *pc, const int with_gravity);
+int cell_unpack(struct pcell *pc, struct cell *c, struct space *s, const int with_gravity);
int cell_pack_tags(const struct cell *c, int *tags);
int cell_unpack_tags(const int *tags, struct cell *c);
int cell_pack_end_step(struct cell *c, struct pcell_step *pcell);
diff --git a/src/engine.c b/src/engine.c
index ab8c4dd691..6059839788 100644
--- a/src/engine.c
+++ b/src/engine.c
@@ -1670,10 +1670,11 @@ void engine_exchange_cells(struct engine *e) {
struct space *s = e->s;
const int nr_proxies = e->nr_proxies;
+ const int with_gravity = e->policy & engine_policy_self_gravity;
const ticks tic = getticks();
/* Exchange the cell structure with neighbouring ranks. */
- proxy_cells_exchange(e->proxies, e->nr_proxies, e->s);
+ proxy_cells_exchange(e->proxies, e->nr_proxies, e->s, with_gravity);
/* Count the number of particles we need to import and re-allocate
the buffer if needed. */
@@ -3948,18 +3949,14 @@ void engine_rebuild(struct engine *e, int clean_smoothing_length_values) {
/* If in parallel, exchange the cell structure, top-level and neighbouring
* multipoles. */
#ifdef WITH_MPI
- engine_exchange_cells(e);
-
if (e->policy & engine_policy_self_gravity) engine_exchange_top_multipoles(e);
+
+ engine_exchange_cells(e);
#endif
/* Re-build the tasks. */
engine_maketasks(e);
-#ifdef WITH_MPI
- if (e->policy & engine_policy_self_gravity) engine_exchange_proxy_multipoles(e);
-#endif
-
/* Make the list of top-level cells that have tasks */
space_list_cells_with_tasks(e->s);
@@ -5400,8 +5397,17 @@ void engine_makeproxies(struct engine *e) {
if (with_gravity) {
/* Are we too close for M2L? */
- if (!cell_can_use_pair_mm_rebuild(&cells[cid], &cells[cjd], e,
- s))
+ /* if (!cell_can_use_pair_mm_rebuild(&cells[cid], &cells[cjd], e, */
+ /* s)) */
+ if (((abs(ind[0] - ii) <= 5 ||
+ abs(ind[0] - ii - cdim[0]) <= 5 ||
+ abs(ind[0] - ii + cdim[0]) <= 5) &&
+ (abs(ind[1] - jj) <= 5 ||
+ abs(ind[1] - jj - cdim[1]) <= 5 ||
+ abs(ind[1] - jj + cdim[1]) <= 5) &&
+ (abs(ind[2] - kk) <= 5 ||
+ abs(ind[2] - kk - cdim[2]) <= 5 ||
+ abs(ind[2] - kk + cdim[2]) <= 5)))
proxy_type |= (int)proxy_cell_type_gravity;
}
diff --git a/src/proxy.c b/src/proxy.c
index 6156291378..a5d30848a3 100644
--- a/src/proxy.c
+++ b/src/proxy.c
@@ -159,103 +159,6 @@ void proxy_tags_exchange(struct proxy *proxies, int num_proxies,
#endif
}
-/**
- * @brief Exchange the cell structures with all proxies.
- *
- * @param proxies The list of #proxy that will send/recv cells.
- * @param num_proxies The number of proxies.
- * @param s The space into which the particles will be unpacked.
- */
-void proxy_cells_exchange(struct proxy *proxies, int num_proxies,
- struct space *s) {
-
-#ifdef WITH_MPI
-
- MPI_Request *reqs;
- if ((reqs = (MPI_Request *)malloc(sizeof(MPI_Request) * 2 * num_proxies)) ==
- NULL)
- error("Failed to allocate request buffers.");
- MPI_Request *reqs_in = reqs;
- MPI_Request *reqs_out = &reqs[num_proxies];
-
- /* Run through the cells and get the size of the ones that will be sent off.
- */
- int count_out = 0;
- int offset[s->nr_cells];
- for (int k = 0; k < s->nr_cells; k++) {
- offset[k] = count_out;
- if (s->cells_top[k].sendto)
- count_out +=
- (s->cells_top[k].pcell_size = cell_getsize(&s->cells_top[k]));
- }
-
- /* Allocate the pcells. */
- struct pcell *pcells = NULL;
- if (posix_memalign((void **)&pcells, SWIFT_CACHE_ALIGNMENT,
- sizeof(struct pcell) * count_out) != 0)
- error("Failed to allocate pcell buffer.");
-
- /* Pack the cells. */
- for (int k = 0; k < s->nr_cells; k++)
- if (s->cells_top[k].sendto) {
- cell_pack(&s->cells_top[k], &pcells[offset[k]]);
- s->cells_top[k].pcell = &pcells[offset[k]];
- }
-
- /* Launch the first part of the exchange. */
- for (int k = 0; k < num_proxies; k++) {
- proxy_cells_exchange_first(&proxies[k]);
- reqs_in[k] = proxies[k].req_cells_count_in;
- reqs_out[k] = proxies[k].req_cells_count_out;
- }
-
- /* Wait for each count to come in and start the recv. */
- for (int k = 0; k < num_proxies; k++) {
- int pid = MPI_UNDEFINED;
- MPI_Status status;
- if (MPI_Waitany(num_proxies, reqs_in, &pid, &status) != MPI_SUCCESS ||
- pid == MPI_UNDEFINED)
- error("MPI_Waitany failed.");
- // message( "request from proxy %i has arrived." , pid );
- proxy_cells_exchange_second(&proxies[pid]);
- }
-
- /* Wait for all the sends to have finished too. */
- if (MPI_Waitall(num_proxies, reqs_out, MPI_STATUSES_IGNORE) != MPI_SUCCESS)
- error("MPI_Waitall on sends failed.");
-
- /* Set the requests for the cells. */
- for (int k = 0; k < num_proxies; k++) {
- reqs_in[k] = proxies[k].req_cells_in;
- reqs_out[k] = proxies[k].req_cells_out;
- }
-
- /* Wait for each pcell array to come in from the proxies. */
- for (int k = 0; k < num_proxies; k++) {
- int pid = MPI_UNDEFINED;
- MPI_Status status;
- if (MPI_Waitany(num_proxies, reqs_in, &pid, &status) != MPI_SUCCESS ||
- pid == MPI_UNDEFINED)
- error("MPI_Waitany failed.");
- // message( "cell data from proxy %i has arrived." , pid );
- for (int count = 0, j = 0; j < proxies[pid].nr_cells_in; j++)
- count += cell_unpack(&proxies[pid].pcells_in[count],
- proxies[pid].cells_in[j], s);
- }
-
- /* Wait for all the sends to have finished too. */
- if (MPI_Waitall(num_proxies, reqs_out, MPI_STATUSES_IGNORE) != MPI_SUCCESS)
- error("MPI_Waitall on sends failed.");
-
- /* Clean up. */
- free(reqs);
- free(pcells);
-
-#else
- error("SWIFT was not compiled with MPI support.");
-#endif
-}
-
/**
* @brief Exchange cells with a remote node, first part.
*
@@ -350,6 +253,105 @@ void proxy_cells_exchange_second(struct proxy *p) {
#endif
}
+/**
+ * @brief Exchange the cell structures with all proxies.
+ *
+ * @param proxies The list of #proxy that will send/recv cells.
+ * @param num_proxies The number of proxies.
+ * @param s The space into which the particles will be unpacked.
+ * @param with_gravity Are we running with gravity and hence need
+ * to exchange multipoles?
+ */
+void proxy_cells_exchange(struct proxy *proxies, int num_proxies,
+ struct space *s, const int with_gravity) {
+
+#ifdef WITH_MPI
+
+ MPI_Request *reqs;
+ if ((reqs = (MPI_Request *)malloc(sizeof(MPI_Request) * 2 * num_proxies)) ==
+ NULL)
+ error("Failed to allocate request buffers.");
+ MPI_Request *reqs_in = reqs;
+ MPI_Request *reqs_out = &reqs[num_proxies];
+
+ /* Run through the cells and get the size of the ones that will be sent off.
+ */
+ int count_out = 0;
+ int offset[s->nr_cells];
+ for (int k = 0; k < s->nr_cells; k++) {
+ offset[k] = count_out;
+ if (s->cells_top[k].sendto)
+ count_out +=
+ (s->cells_top[k].pcell_size = cell_getsize(&s->cells_top[k]));
+ }
+
+ /* Allocate the pcells. */
+ struct pcell *pcells = NULL;
+ if (posix_memalign((void **)&pcells, SWIFT_CACHE_ALIGNMENT,
+ sizeof(struct pcell) * count_out) != 0)
+ error("Failed to allocate pcell buffer.");
+
+ /* Pack the cells. */
+ for (int k = 0; k < s->nr_cells; k++)
+ if (s->cells_top[k].sendto) {
+ cell_pack(&s->cells_top[k], &pcells[offset[k]], with_gravity);
+ s->cells_top[k].pcell = &pcells[offset[k]];
+ }
+
+ /* Launch the first part of the exchange. */
+ for (int k = 0; k < num_proxies; k++) {
+ proxy_cells_exchange_first(&proxies[k]);
+ reqs_in[k] = proxies[k].req_cells_count_in;
+ reqs_out[k] = proxies[k].req_cells_count_out;
+ }
+
+ /* Wait for each count to come in and start the recv. */
+ for (int k = 0; k < num_proxies; k++) {
+ int pid = MPI_UNDEFINED;
+ MPI_Status status;
+ if (MPI_Waitany(num_proxies, reqs_in, &pid, &status) != MPI_SUCCESS ||
+ pid == MPI_UNDEFINED)
+ error("MPI_Waitany failed.");
+ // message( "request from proxy %i has arrived." , pid );
+ proxy_cells_exchange_second(&proxies[pid]);
+ }
+
+ /* Wait for all the sends to have finished too. */
+ if (MPI_Waitall(num_proxies, reqs_out, MPI_STATUSES_IGNORE) != MPI_SUCCESS)
+ error("MPI_Waitall on sends failed.");
+
+ /* Set the requests for the cells. */
+ for (int k = 0; k < num_proxies; k++) {
+ reqs_in[k] = proxies[k].req_cells_in;
+ reqs_out[k] = proxies[k].req_cells_out;
+ }
+
+ /* Wait for each pcell array to come in from the proxies. */
+ for (int k = 0; k < num_proxies; k++) {
+ int pid = MPI_UNDEFINED;
+ MPI_Status status;
+ if (MPI_Waitany(num_proxies, reqs_in, &pid, &status) != MPI_SUCCESS ||
+ pid == MPI_UNDEFINED)
+ error("MPI_Waitany failed.");
+ // message( "cell data from proxy %i has arrived." , pid );
+ for (int count = 0, j = 0; j < proxies[pid].nr_cells_in; j++)
+ count += cell_unpack(&proxies[pid].pcells_in[count],
+ proxies[pid].cells_in[j], s, with_gravity);
+ }
+
+ /* Wait for all the sends to have finished too. */
+ if (MPI_Waitall(num_proxies, reqs_out, MPI_STATUSES_IGNORE) != MPI_SUCCESS)
+ error("MPI_Waitall on sends failed.");
+
+ /* Clean up. */
+ free(reqs);
+ free(pcells);
+
+#else
+ error("SWIFT was not compiled with MPI support.");
+#endif
+}
+
/**
* @brief Add a cell to the given proxy's input list.
*
diff --git a/src/proxy.h b/src/proxy.h
index 63f51846d0..fbc3f1a163 100644
--- a/src/proxy.h
+++ b/src/proxy.h
@@ -102,9 +102,7 @@ void proxy_parts_exchange_second(struct proxy *p);
void proxy_addcell_in(struct proxy *p, struct cell *c, int type);
void proxy_addcell_out(struct proxy *p, struct cell *c, int type);
void proxy_cells_exchange(struct proxy *proxies, int num_proxies,
- struct space *s);
-void proxy_cells_exchange_first(struct proxy *p);
-void proxy_cells_exchange_second(struct proxy *p);
+ struct space *s, int with_gravity);
void proxy_tags_exchange(struct proxy *proxies, int num_proxies,
struct space *s);
--
GitLab