micro optimizations for the evaluation engine

doc/gc.txt

@@ -22,6 +22,13 @@ delta-subgraph of the heap that changed since its last run.
 The GC is only triggered in a memory allocation operation. It it not triggered
 by some timer and does not run in a background thread.
 
+To force a full collection call ``GC_fullCollect``. Note that it is generally
+better to let the GC do its work and not enforce a full collection.
+
+
+Cycle collector
+===============
+
 The cycle collector can be en-/disabled independently from the other parts of
 the GC with ``GC_enableMarkAndSweep`` and ``GC_disableMarkAndSweep``. The
 compiler analyses the types for their possibility to build cycles, but often
@@ -30,11 +37,10 @@ it is necessary to help this analysis with the ``acyclic`` pragma (see
 
 You can also use the ``acyclic`` pragma for data that is cyclic in reality and 
 then break up the cycles explicitly with ``GC_addCycleRoot``. This can be a
-very good optimization; the Nimrod compiler itself relies on this optimization
-trick to improve performance.
-
-To force a full collection call ``GC_fullCollect``. Note that it is generally
-better to let the GC do its work and not enforce a full collection.
+very valuable optimization; the Nimrod compiler itself relies on this 
+optimization trick to improve performance. Note that ``GC_addCycleRoot`` is
+a quick operation; the root is only registered for the next run of the 
+cycle collector.
 
 
 Realtime support