Addition of schedule(simd:static)

[tlwilmar]

Summary

This patch implements the new kmp_sch_static_balanced_chunked schedule kind that the compiler will generate when it encounters schedule(simd: static). It just adds the new constant and the new switch case __kmp_for_static_init.

Patch by Alex Duran.

[AndreyChurbanov]

LGTM

openmp/trunk/runtime/src/kmp.h

enum sched_type {
     kmp_sch_static_greedy             = 40,
     kmp_sch_static_balanced           = 41,
     /* accessible only through KMP_SCHEDULE environment variable */
     kmp_sch_guided_iterative_chunked  = 42,
     kmp_sch_guided_analytical_chunked = 43,
 
     kmp_sch_static_steal              = 44,   /**< accessible only through KMP_SCHEDULE environment variable */
 
+#if OMP_41_ENABLED
+    kmp_sch_static_balanced_chunked   = 45,   /**< static with chunk adjustment (e.g., simd) */
+#endif
+
     /* accessible only through KMP_SCHEDULE environment variable */
-    kmp_sch_upper                     = 45,   /**< upper bound for unordered values */
+    kmp_sch_upper                     = 46,   /**< upper bound for unordered values */
 
     kmp_ord_lower                     = 64,   /**< lower bound for ordered values, must be power of 2 */
     kmp_ord_static_chunked            = 65,
     kmp_ord_static                    = 66,   /**< ordered static unspecialized */
     kmp_ord_dynamic_chunked           = 67,
     kmp_ord_guided_chunked            = 68,
     kmp_ord_runtime                   = 69,
     kmp_ord_auto                      = 70,   /**< ordered auto */

openmp/trunk/runtime/src/kmp_sched.cpp

case kmp_sch_static_chunked:
             span = chunk * incr;
             *pstride = span * nth;
             *plower = *plower + (span * tid);
             *pupper = *plower + span - incr;
             if( plastiter != NULL )
                 *plastiter = (tid == ((trip_count - 1)/( UT )chunk) % nth);
             break;
         }
+#if OMP_41_ENABLED
+    case kmp_sch_static_balanced_chunked:
+        {
+            register T old_upper = *pupper;
+            // round up to make sure the chunk is enough to cover all iterations
+            register UT span = (trip_count+nth-1) / nth;
+
+            // perform chunk adjustment
+            chunk = (span + chunk - 1) & ~(chunk-1);
+
+            span = chunk * incr;
+            *plower = *plower + (span * tid);
+            *pupper = *plower + span - incr;
+            if ( incr > 0 ) {
+              if ( *pupper > old_upper ) *pupper = old_upper;
+            } else
+              if ( *pupper < old_upper ) *pupper = old_upper;
+
+            if( plastiter != NULL )
+                *plastiter = ( tid == ((trip_count - 1)/( UT )chunk) );
+            break;
+        }
+#endif
     default:
         KMP_ASSERT2( 0, "__kmpc_for_static_init: unknown scheduling type" );
         break;
     }
 
 #if USE_ITT_BUILD
     // Report loop metadata
     if ( KMP_MASTER_TID(tid) && __itt_metadata_add_ptr && __kmp_forkjoin_frames_mode == 3 &&