Major optimization, rewrite to avoid deep recursion if possible.

[SXSI/xpathcomp.git] / main.ml

diff --git a/main.ml b/main.ml
index 26b445e..76f8089 100644 (file)
--- a/main.ml
+++ b/main.ml
@@ -9,10 +9,12 @@ open Ata
 INCLUDE "utils.ml"
 let () = init_timer();;
 
-let enabled_gc = Gc.get()
-let disabled_gc = { Gc.get() with
-                     Gc.max_overhead = 1000000; 
-                     Gc.space_overhead = 100 }
+let default_gc = Gc.get()
+let tuned_gc = { Gc.get() with
+                  Gc.minor_heap_size = 4*1024*1024;
+                  Gc.major_heap_increment = 1024*1024;
+                  Gc.max_overhead = 1000000;
+                 }
 let hash x = 131*x/(x-1+1)
 
 let test_loop tree tag =
@@ -54,6 +56,28 @@ let test_loop2 tree tag =
   Hashtbl.add f (hash 101) `Foo;
   g t' Tree.root
 
+let test_text doc = 
+  let _ = Printf.eprintf "Contains(bree)" in
+  let _ = time (Tree.test_contains doc) "bree" in
+  let _ = Printf.eprintf "Contains(brain)" in
+  let _ = time (Tree.test_contains doc) "brain" in
+  let _ = Printf.eprintf "Contains(brain)" in
+  let i = time (Tree.test_contains doc) "brain" in
+  let _ = Printf.eprintf "%i\nContains(Australia)" i in
+  let i = time (Tree.test_contains doc) "AUSTRALIA" in
+  let _ = Printf.eprintf "%i\n Contains(1930)" i in
+  let i = time (Tree.test_contains doc) "1930" in
+  let _ = Printf.eprintf "%i\n startswith(bar)" i in
+  let i = time (Tree.test_prefix doc) "bar" in
+  let _ = Printf.eprintf "%i\n endswith(LAND)" i in
+  let i = time (Tree.test_suffix doc) "LAND" in
+  let _ = Printf.eprintf "%i\n =(2001)" i in
+  let i = time (Tree.test_equals doc) "2001" in
+  let _ = Printf.eprintf "%i\n =(Nguyen)" i in
+  let i = time (Tree.test_equals doc) "Nguyen" in
+  Printf.eprintf "%i\n" i ;
+  ()
+
 type pointers
 external build_pointers : Tree.t -> pointers = "caml_build_pointers"
 external iter_pointers : pointers -> int = "caml_iter_pointers"
@@ -71,15 +95,15 @@ let main v query_string output =
          Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
       in
       let _ = Printf.eprintf "Number of nodes %i\n%!" (Tree.size v) in
-      let _ = Tree.stats v in
-      let _ = Printf.eprintf "Timing first_child/next_sibling %!" in
-      let _ = time (Tree.benchmark_fcns)  v in
-      let _ = Printf.eprintf "Timing last_child/prev_sibling %!" in
+(*      let _ = test_text v in *)
+(*      let _ = Tree.stats v in *)
+(*      let _ = Printf.eprintf "Timing first_child/next_sibling %!" in  *)
+(*      let _ = time ~count:1 (Tree.benchmark_fcns)  v in  *)
+   (*   let _ = Printf.eprintf "Timing last_child/prev_sibling %!" in
       let _ = time (Tree.benchmark_lcps)  v in
       let _ = Printf.eprintf "Timing jump to a %!" in
-      let _ = time (Tree.benchmark_jump v) (Tag.tag "a")  in
-      
-(*      let _ = Printf.eprintf "Timing pointer allocation %!" in
+      let _ = time (Tree.benchmark_jump v) (Tag.tag "a")  in      
+      let _ = Printf.eprintf "Timing pointer allocation %!" in
       let pointers = time (build_pointers) v  in
       let _ = Printf.eprintf "Timing pointer iteration %!" in
       let i = time (iter_pointers) pointers  in
@@ -143,7 +167,7 @@ let main v query_string output =
          begin
            let _ = Gc.full_major();Gc.compact() in
            let _ = Printf.eprintf "%!" in
-(*         let _ = Gc.set (disabled_gc) in *)
+           let _ = Gc.set (tuned_gc) in 
              if !Options.backward && ((snd test_list) != `NOTHING )then 
                if !Options.count_only then
                let r = time_mem (bottom_up_count auto v )(snd test_list)  in
@@ -173,12 +197,12 @@ let main v query_string output =
                  if !Options.backward then Printf.eprintf "WARNING: couldn't find a jumping point, running top-down\n" 
                in
                if !Options.count_only then
-                 let r = time_mem ( top_down_count auto ) v in 
+                 let r = time ~count:5 ( top_down_count1 auto ) v in 
                  let _ = Printf.eprintf "Number of nodes in the result set : %i\n%!" r
                  in ()
                else      
                let module GR = Ata.Test(struct let doc = v end) in
-                 let result = time_mem (GR.top_down auto) v in
+                 let result = time ~count:5 (GR.top_down1 auto) v in
                  let _ = Printf.eprintf "Counting results " in
                  let rcount = time (GR.Results.length) result in
                    Printf.eprintf "Number of nodes in the result set : %i\n" rcount;
@@ -211,8 +235,7 @@ let main v query_string output =
                                                     ) result) ();
                    end;
          end;
-         let _ = Gc.set enabled_gc in
-           Printf.eprintf "Total running time : %fms\n%!" (total_time())
+         Printf.eprintf "Total running time : %fms\n%!" (total_time())
 ;;
 
 Options.parse_cmdline();;
@@ -240,3 +263,5 @@ let v =
 in
   main v !Options.query !Options.output_file;;
 
+
+