X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=main.ml;h=8bb6b4465d23ce8cf39115b3a5cfcb119addc2af;hb=55ee2680ca6d4e952cb800c529d45e064a2eff1c;hp=342bb2f1ed6f556fb8bce1b548d01fecb68f8137;hpb=cad5e2e2831477cba1f6211c57b9a4cc5b58bd55;p=SXSI%2Fxpathcomp.git diff --git a/main.ml b/main.ml index 342bb2f..8bb6b44 100644 --- 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 +56,34 @@ 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" +external free_pointers : pointers -> unit = "caml_free_pointers" + + let main v query_string output = let _ = Tag.init (Tree.tag_pool v) in @@ -65,11 +95,20 @@ 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 _ = Printf.eprintf "Timing first_child/next_sibling %!" in - let _ = time (Tree.benchmark_fsns) v in +(* let _ = test_text v in *) +(* let _ = Tree.stats v in *) + let _ = Printf.eprintf "Timing first_child/next_sibling %!" in + let _ = time ~count:3 (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 _ = 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 + let _ = Printf.eprintf "Traversed %i pointers\n\nTiming pointer deallocation %!" i in + let _ = time (free_pointers) pointers in *) (* let _ = Printf.eprintf "Timing //keyword :" in let r = time (test_loop v) (Tag.tag "keyword") in let _ = Printf.eprintf "Count is %i\n%!" r in @@ -128,7 +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 @@ -158,14 +197,14 @@ 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 module GR = Ata(*.Test(struct let doc = v end) *) 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 + let rcount = time (IdSet.length) result in Printf.eprintf "Number of nodes in the result set : %i\n" rcount; Printf.eprintf "\n%!"; begin @@ -173,31 +212,19 @@ let main v query_string output = | None -> () | Some f -> Printf.eprintf "Serializing results : "; - time( fun () -> - (*let oc = open_out f in *) - let oc = Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 in - (*output_string oc "\n";*) - let t1 = ref (Unix.gettimeofday()) in - let count = ref 1 in - let old_count = ref 1 in - GR.Results.iter (fun t -> - incr count; - begin - if (!count mod 15) == 0 - then - let t2 = Unix.gettimeofday() in - let _ = Printf.eprintf "Printing %i elements in %f ms\n" - (!count - !old_count) (1000. *.(t2 -. !t1)) - in - ( old_count := !count; t1 := Unix.gettimeofday()) - end; - Tree.print_xml_fast3 v t oc; - (*output_char oc '\n'; *) - ) result) (); - end; + let oc = + Unix.openfile f [ Unix.O_WRONLY;Unix.O_TRUNC;Unix.O_CREAT] 0o644 + in + time( + IdSet.iter ( + fun t -> + + Tree.print_xml_fast3 v t oc; + + )) 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();; @@ -225,3 +252,5 @@ let v = in main v !Options.query !Options.output_file;; + +