X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fgovern.c;h=9f525e60c67c22b695328d873ba43cf0fa711306;hb=8c8736b23823acce9df684b85c6320cc36e43f76;hp=388fac4a4dff8c9a359d5e74f92c95bde46f5b62;hpb=ae3892dc4f3c3a759315969e2ab5156f034707b2;p=libucw.git

diff --git a/lib/sorter/govern.c b/lib/sorter/govern.c
index 388fac4a..9f525e60 100644
--- a/lib/sorter/govern.c
+++ b/lib/sorter/govern.c
@@ -19,29 +19,27 @@
 
 #define F_BSIZE(b) stk_fsize(sbuck_size(b))
 
-static u64
-sorter_clock(void)
+static void
+sorter_start_timer(struct sort_context *ctx)
 {
-  struct timeval tv;
-  gettimeofday(&tv, NULL);
-  return (u64)tv.tv_sec * 1000 + tv.tv_usec / 1000;
+  init_timer(&ctx->start_time);
 }
 
 static void
-sorter_start_timer(struct sort_context *ctx)
+sorter_stop_timer(struct sort_context *ctx, uns *account_to)
 {
-  ctx->start_time = sorter_clock();
+  ctx->last_pass_time = get_timer(&ctx->start_time);
+  *account_to += ctx->last_pass_time;
 }
 
 static uns
 sorter_speed(struct sort_context *ctx, u64 size)
 {
-  u64 stop_time = sorter_clock();
   if (!size)
     return 0;
-  if (stop_time <= ctx->start_time)
+  if (!ctx->last_pass_time)
     return -1;
-  return (uns)((double)size / (1<<20) * 1000 / (stop_time-ctx->start_time));
+  return (uns)((double)size / (1<<20) * 1000 / ctx->last_pass_time);
 }
 
 static int
@@ -106,6 +104,7 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
       ins[0] = sbuck_new(ctx);
       if (!sorter_presort(ctx, b, ins[0], join ? : ins[0]))
 	{
+	  sorter_stop_timer(ctx, &ctx->total_pre_time);
 	  SORT_XTRACE(((b->flags & SBF_SOURCE) ? 1 : 2), "Sorted in memory");
 	  if (join)
 	    sbuck_drop(ins[0]);
@@ -120,6 +119,7 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
       while (sorter_presort(ctx, b, ins[i], ins[i]))
 	i = 1-i;
       sbuck_drop(b);
+      sorter_stop_timer(ctx, &ctx->total_pre_time);
       SORT_TRACE("Presorting pass (%d+%d runs, %s+%s, %dMB/s)",
 		 ins[0]->runs, ins[1]->runs,
 		 F_BSIZE(ins[0]), F_BSIZE(ins[1]),
@@ -146,6 +146,7 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
 	ASSERT(join->runs == 2);
 	join->runs--;
 	join_size = sbuck_size(join) - join_size;
+	sorter_stop_timer(ctx, &ctx->total_ext_time);
 	SORT_TRACE("Mergesort pass %d (final run, %s, %dMB/s)", pass, stk_fsize(join_size), sorter_speed(ctx, join_size));
 	sbuck_drop(ins[0]);
 	sbuck_drop(ins[1]);
@@ -155,6 +156,7 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
     outs[1] = sbuck_new(ctx);
     outs[2] = NULL;
     ctx->twoway_merge(ctx, ins, outs);
+    sorter_stop_timer(ctx, &ctx->total_ext_time);
     SORT_TRACE("Mergesort pass %d (%d+%d runs, %s+%s, %dMB/s)", pass,
 	       outs[0]->runs, outs[1]->runs,
 	       F_BSIZE(outs[0]), F_BSIZE(outs[1]),
@@ -171,27 +173,29 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
 static uns
 sorter_radix_bits(struct sort_context *ctx, struct sort_bucket *b)
 {
-  if (!b->hash_bits || !ctx->radix_split || (sorter_debug & SORT_DEBUG_NO_RADIX))
+  if (!b->hash_bits || !ctx->radix_split ||
+      (b->flags & SBF_CUSTOM_PRESORT) ||
+      (sorter_debug & SORT_DEBUG_NO_RADIX))
     return 0;
 
   u64 in = sbuck_size(b);
-  u64 mem = ctx->internal_estimate(ctx, b);
-  if (in < mem)
+  u64 mem = ctx->internal_estimate(ctx, b) * 0.8;	// FIXME: Magical factor for hash non-uniformity
+  if (in <= mem)
     return 0;
 
-  uns n;
-  for (n = sorter_min_radix_bits; n < sorter_max_radix_bits && n < b->hash_bits; n++)
-    if ((in >> n) < mem)
-      break;
+  uns n = sorter_min_radix_bits;
+  while (n < sorter_max_radix_bits && n < b->hash_bits && (in >> n) > mem)
+    n++;
   return n;
 }
 
 static void
-    sorter_radix(struct sort_context *ctx, struct sort_bucket *b, uns bits)
+sorter_radix(struct sort_context *ctx, struct sort_bucket *b, uns bits)
 {
   uns nbuck = 1 << bits;
-  SORT_XTRACE(2, "Running radix sort on %s with %d bits of %d (expected size %s)",
+  SORT_XTRACE(2, "Running radix split on %s with hash %d bits of %d (expecting %s buckets)",
 	      F_BSIZE(b), bits, b->hash_bits, stk_fsize(sbuck_size(b) / nbuck));
+  sorter_free_buf(ctx);
   sorter_start_timer(ctx);
 
   struct sort_bucket **outs = alloca(nbuck * sizeof(struct sort_bucket *));
@@ -215,6 +219,7 @@ static void
 	sbuck_swap_out(outs[i]);
     }
 
+  sorter_stop_timer(ctx, &ctx->total_ext_time);
   SORT_TRACE("Radix split (%d buckets, %s min, %s max, %s avg, %dMB/s)", nbuck,
 	     stk_fsize(min), stk_fsize(max), stk_fsize(sum / nbuck), sorter_speed(ctx, sum));
   sbuck_drop(b);
@@ -227,8 +232,6 @@ sorter_run(struct sort_context *ctx)
   clist_init(&ctx->bucket_list);
   sorter_prepare_buf(ctx);
 
-  /* FIXME: Remember to test sorting of empty files */
-
   // Create bucket containing the source
   struct sort_bucket *bin = sbuck_new(ctx);
   bin->flags = SBF_SOURCE | SBF_OPEN_READ;
@@ -237,7 +240,7 @@ sorter_run(struct sort_context *ctx)
   else
     bin->fb = ctx->in_fb;
   bin->ident = "in";
-  bin->size = ctx->in_size;		/* FIXME: Sizes should be either sh_off_t or u64, not both; beware of ~0U */
+  bin->size = ctx->in_size;
   bin->hash_bits = ctx->hash_bits;
   clist_add_tail(&ctx->bucket_list, &bin->n);
   SORT_XTRACE(2, "Input size: %s", F_BSIZE(bin));
@@ -269,5 +272,7 @@ sorter_run(struct sort_context *ctx)
   sorter_free_buf(ctx);
   sbuck_write(bout);		// Force empty bucket to a file
   SORT_XTRACE(2, "Final size: %s", F_BSIZE(bout));
+  SORT_XTRACE(2, "Final timings: %.3fs external sorting, %.3fs presorting, %.3fs internal sorting",
+	      ctx->total_ext_time/1000., ctx->total_pre_time/1000., ctx->total_int_time/1000.);
   ctx->out_fb = sbuck_read(bout);
 }