Fix second pass on map-by-obj

Computation of balance was using the wrong variable in the denominator. Signed-off-by: Ralph Castain <[email protected]>
openpmix · Aug 23, 2022 · 6ed6563 · 6ed6563
1 parent 75511d4
commit 6ed6563
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 18 deletions.
diff --git a/src/mca/rmaps/base/help-prte-rmaps-base.txt b/src/mca/rmaps/base/help-prte-rmaps-base.txt
@@ -29,7 +29,7 @@ slots that were requested by the application:
 
   %s
 
-Either request fewer slots for your application, or make more slots
+Either request fewer procs for your application, or make more slots
 available for use.
 
 A "slot" is the PRRTE term for an allocatable unit where we can

diff --git a/src/mca/rmaps/round_robin/rmaps_rr_mappers.c b/src/mca/rmaps/round_robin/rmaps_rr_mappers.c
@@ -103,6 +103,14 @@ int prte_rmaps_rr_byslot(prte_job_t *jdata,
             }
         }
 
+        if (!options->oversubscribe) {
+            /* since oversubscribe is not allowed, cap our usage
+             * at the number of available slots. */
+            if (node->slots_available < options->nprocs) {
+                options->nprocs = node->slots_available;
+            }
+        }
+
         /* if the number of procs is greater than the number of CPUs
          * on this node, but less or equal to the number of slots,
          * then we are not oversubscribed but we are overloaded. If
@@ -240,6 +248,14 @@ int prte_rmaps_rr_bynode(prte_job_t *jdata,
     {
         prte_rmaps_base_get_cpuset(jdata, node, options);
 
+        if (!options->oversubscribe) {
+            /* since oversubscribe is not allowed, cap our usage
+             * at the number of available slots. */
+            if (node->slots_available < options->nprocs) {
+                options->nprocs = node->slots_available;
+            }
+        }
+
         /* if the number of procs is greater than the number of CPUs
          * on this node, but less or equal to the number of slots,
          * then we are not oversubscribed but we are overloaded. If
@@ -259,19 +275,6 @@ int prte_rmaps_rr_bynode(prte_job_t *jdata,
             continue;
         }
 
-        /* if oversubscribe is specified, then just ignore the
-         * number of slots on each node and assign this number.
-         * Note that oversubscribe automatically dictates that
-         * we do not bind, so binding can also be ignored */
-
-        if (!options->oversubscribe) {
-            /* since oversubscribe is not allowed , cap our usage
-             * at the number of available slots */
-            if (node->slots_available < options->nprocs) {
-                options->nprocs = node->slots_available;
-            }
-        }
-
         PRTE_OUTPUT_VERBOSE((10, prte_rmaps_base_framework.framework_output,
                              "%s NODE %s ASSIGNING %d PROCS",
                              PRTE_NAME_PRINT(PRTE_PROC_MY_NAME),
@@ -374,6 +377,14 @@ int prte_rmaps_rr_bycpu(prte_job_t *jdata, prte_app_context_t *app,
             }
         }
 
+        if (!options->oversubscribe) {
+            /* oversubscribe is not allowed, so cap our usage
+             * at the number of available slots. */
+            if (node->slots_available < options->nprocs) {
+                options->nprocs = node->slots_available;
+            }
+        }
+
         /* if the number of procs is greater than the number of CPUs
          * on this node, but less or equal to the number of slots,
          * then we are not oversubscribed but we are overloaded. If
@@ -576,8 +587,14 @@ int prte_rmaps_rr_byobj(prte_job_t *jdata, prte_app_context_t *app,
                 }
             }
         }
-        prte_output_verbose(2, prte_rmaps_base_framework.framework_output,
-                            "mca:rmaps:rr: assigning nprocs %d", nprocs);
+
+        if (!options->oversubscribe) {
+            /* since oversubscribe is not allowed, cap our usage
+             * at the number of available slots. */
+            if (node->slots_available < nprocs) {
+                nprocs = node->slots_available;
+            }
+        }
 
         /* if the number of procs is greater than the number of CPUs
          * on this node, but less or equal to the number of slots,
@@ -592,6 +609,9 @@ int prte_rmaps_rr_byobj(prte_job_t *jdata, prte_app_context_t *app,
             jdata->map->binding = PRTE_BIND_TO_NONE;
         }
 
+        prte_output_verbose(2, prte_rmaps_base_framework.framework_output,
+                            "mca:rmaps:rr: assigning nprocs %d", nprocs);
+
         nodefull = false;
         if (span) {
             /* if we are mapping spanned, then we loop over
@@ -700,8 +720,7 @@ int prte_rmaps_rr_byobj(prte_job_t *jdata, prte_app_context_t *app,
      * handling the oversubscription. Figure out how many procs
      * to add to each of them.
      */
-    balance = (float) ((int) app->num_procs - nprocs_mapped)
-    / (float) total_nobjs;
+    balance = (float) ((int) app->num_procs - nprocs_mapped) / (float) options->total_nobjs;
     extra_procs_to_assign = (int) balance;
     if (0 < (balance - (float) extra_procs_to_assign)) {
         /* compute how many nodes need an extra proc */