ikarus/benchmarks.larceny/src/gcbench.sml

(*
;  This is adapted from a benchmark written by John Ellis and Pete Kovac
;  of Post Communications.
;  It was modified by Hans Boehm of Silicon Graphics.
;  It was translated into Scheme by William D Clinger of Northeastern Univ,
;    and modified for compatibility with the Gambit benchmark suite.
;  It was translated into Standard ML by William D Clinger.
;  Last modified 6 July 1999.
; 
;       This is no substitute for real applications.  No actual application
;       is likely to behave in exactly this way.  However, this benchmark was
;       designed to be more representative of real applications than other
;       Java GC benchmarks of which we are aware.
;       It attempts to model those properties of allocation requests that
;       are important to current GC techniques.
;       It is designed to be used either to obtain a single overall performance
;       number, or to give a more detailed estimate of how collector
;       performance varies with object lifetimes.  It prints the time
;       required to allocate and collect balanced binary trees of various
;       sizes.  Smaller trees result in shorter object lifetimes.  Each cycle
;       allocates roughly the same amount of memory.
;       Two data structures are kept around during the entire process, so
;       that the measured performance is representative of applications
;       that maintain some live in-memory data.  One of these is a tree
;       containing many pointers.  The other is a large array containing
;       double precision floating point numbers.  Both should be of comparable
;       size.
; 
;       The results are only really meaningful together with a specification
;       of how much memory was used.  It is possible to trade memory for
;       better time performance.  This benchmark should be run in a 32 MB
;       heap, though we don't currently know how to enforce that uniformly.

; In the Java version, this routine prints the heap size and the amount
; of free memory.  There is no portable way to do this in Scheme; each
; implementation needs its own version.
*)

datatype Tree = Dummy
              | Node of { left: Tree ref, right: Tree ref, i: int, j: int }

fun make_empty_node () =
  Node { left= ref Dummy, right= ref Dummy, i= 0, j= 0 }

fun make_node (l, r) =
  Node { left= ref l, right= ref r, i= 0, j= 0 }

fun PrintDiagnostics () = ()

fun gcbench kStretchTreeDepth =

  let open Int

      fun expt (m:int, n:int) =
        if n = 0 then 1 else m * expt (m, n - 1)
  
      (*  Nodes used by a tree of a given size  *)
      fun TreeSize i =
        expt (2, i + 1) - 1
  
      (*  Number of iterations to use for a given tree depth  *)
      fun NumIters i =
        (2 * (TreeSize kStretchTreeDepth)) div (TreeSize i)

      (*
      ;  Parameters are determined by kStretchTreeDepth.
      ;  In Boehm's version the parameters were fixed as follows:
      ;    public static final int kStretchTreeDepth    = 18;  // about 16Mb
      ;    public static final int kLongLivedTreeDepth  = 16;  // about 4Mb
      ;    public static final int kArraySize  = 500000;       // about 4Mb
      ;    public static final int kMinTreeDepth = 4;
      ;    public static final int kMaxTreeDepth = 16;
      ;  In Larceny the storage numbers above would be 12 Mby, 3 Mby, 6 Mby.
      *)

      val kLongLivedTreeDepth = kStretchTreeDepth - 2
      val kArraySize          = 4 * (TreeSize kLongLivedTreeDepth)
      val kMinTreeDepth       = 4
      val kMaxTreeDepth       = kLongLivedTreeDepth
    
      (*  Build tree top down, assigning to older objects.  *)
      fun Populate (iDepth, Node { left=lr, right=rr, i, j }) =
        if iDepth <= 0
            then false
            else let val iDepth = iDepth - 1
                 in
                 (
                     lr := make_empty_node();
                     rr := make_empty_node();
                     Populate (iDepth, !lr);
                     Populate (iDepth, !rr)
                 )
                 end
      
      (*  Build tree bottom-up  *)
      fun MakeTree iDepth =
        if iDepth <= 0
            then make_empty_node()
            else make_node (MakeTree (iDepth - 1),
                            MakeTree (iDepth - 1))
      
      fun TimeConstruction depth =
        let val iNumIters = NumIters depth
        in
        (
          print (concat ["Creating ",
                         toString iNumIters,
                         " trees of depth ",
                         toString depth,
                         "\n"]);
          let fun loop i =
                if i < iNumIters
                  then (Populate (depth, make_empty_node());
                        loop (i + 1))
                  else ()
          in loop 0
          end;
          let fun loop i =
                if i < iNumIters
                  then (MakeTree depth;
                        loop (i + 1))
                  else ()
          in loop 0
          end
        )
        end
      
      fun main () =
        (
        print "Garbage Collector Test\n";
        print (concat [" Stretching memory with a binary tree of depth ",
                       toString kStretchTreeDepth,
                       "\n"]);
        PrintDiagnostics();
        (*  Stretch the memory space quickly  *)
        MakeTree kStretchTreeDepth;
                         
        (*  Create a long lived object  *)
        print (concat[" Creating a long-lived binary tree of depth ",
                      toString kLongLivedTreeDepth,
                      "\n"]);
        let val longLivedTree = make_empty_node()
        in
        (
          Populate (kLongLivedTreeDepth, longLivedTree);
          
          (*  Create long-lived array, filling half of it  *)
          print (concat [" Creating a long-lived array of ",
                         toString kArraySize,
                         " inexact reals\n"]);
          let open Array
              val arr = array (kArraySize, 0.0)
              fun loop1 i =
                if i < (kArraySize div 2)
                  then (update (arr, i, 1.0/(Real.fromInt(i)));
                        loop1 (i + 1))
                  else ()
              fun loop2 d =
                if d <= kMaxTreeDepth
                  then (TimeConstruction d;
                        loop2 (d + 2))
                  else ()
          in
          (
            loop1 0;
            PrintDiagnostics();

            loop2 kMinTreeDepth;

            if (longLivedTree = Dummy)
               orelse
               let val n = min (1000, (length(arr) div 2) - 1)
               in Real.!= (sub (arr, n), (1.0 / Real.fromInt(n)))
               end
              then print "Failed\n"
              else ()
            (*  fake reference to LongLivedTree
                and array to keep them from being optimized away
            *)
          )
          end)
        end;
        PrintDiagnostics())
  in main()
  end

fun main () =
  run_benchmark ("gcbench",
                 1,
                 fn () => gcbench 18,
                 fn (result) => true)