X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;ds=sidebyside;f=lib%2Fsorter%2Fgovern.c;h=2a8bea0b27da7330c18f7b4f65d91fadcb96bc53;hb=7bddf3297ee2592efb3ca02b822683371e64e340;hp=b6dea908ee3ebc0686bd53ad5f0f1a67982fdebd;hpb=9d9c5f728fc131aff8b450da6bce4a0d4414b8ff;p=libucw.git

diff --git a/lib/sorter/govern.c b/lib/sorter/govern.c
index b6dea908..2a8bea0b 100644
--- a/lib/sorter/govern.c
+++ b/lib/sorter/govern.c
@@ -10,15 +10,14 @@
 #include "lib/lib.h"
 #include "lib/fastbuf.h"
 #include "lib/mempool.h"
+#include "lib/stkstring.h"
 #include "lib/sorter/common.h"
 
 #include <string.h>
 #include <sys/time.h>
 #include <time.h>
-#include <alloca.h>
 
-#define F_SIZE(x) ({ byte *buf = alloca(16); format_size(buf, x); buf; })
-#define F_BSIZE(b) F_SIZE(sbuck_size(b))
+#define F_BSIZE(b) stk_fsize(sbuck_size(b))
 
 static u64
 sorter_clock(void)
@@ -147,7 +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;
-	SORT_TRACE("Mergesort pass %d (final run, %s, %dMB/s)", pass, F_SIZE(join_size), sorter_speed(ctx, join_size));
+	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]);
 	return;
@@ -169,20 +168,32 @@ sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
   clist_insert_after(&ins[0]->n, list_pos);
 }
 
-static int
-sorter_radix_p(struct sort_context *ctx, struct sort_bucket *b)
+static uns
+sorter_radix_bits(struct sort_context *ctx, struct sort_bucket *b)
 {
-  return b->hash_bits && ctx->radix_split &&
-    !(sorter_debug & SORT_DEBUG_NO_RADIX) &&
-    sbuck_size(b) > (sh_off_t)sorter_bufsize;
+  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) * 0.8;	// FIXME: Magical factor for hash non-uniformity
+  if (in <= mem)
+    return 0;
+
+  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)
+sorter_radix(struct sort_context *ctx, struct sort_bucket *b, uns bits)
 {
-  uns bits = MIN(b->hash_bits, 4);	/* FIXME */
   uns nbuck = 1 << bits;
-  SORT_XTRACE(2, "Running radix sort on %s with %d bits of %d", F_BSIZE(b), bits, b->hash_bits);
+  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 *));
@@ -202,10 +213,12 @@ sorter_radix(struct sort_context *ctx, struct sort_bucket *b)
       min = MIN(min, s);
       max = MAX(max, s);
       sum += s;
+      if (nbuck > 4)
+	sbuck_swap_out(outs[i]);
     }
 
   SORT_TRACE("Radix split (%d buckets, %s min, %s max, %s avg, %dMB/s)", nbuck,
-	     F_SIZE(min), F_SIZE(max), F_SIZE(sum / nbuck), sorter_speed(ctx, sum));
+	     stk_fsize(min), stk_fsize(max), stk_fsize(sum / nbuck), sorter_speed(ctx, sum));
   sbuck_drop(b);
 }
 
@@ -214,8 +227,7 @@ sorter_run(struct sort_context *ctx)
 {
   ctx->pool = mp_new(4096);
   clist_init(&ctx->bucket_list);
-
-  /* FIXME: Remember to test sorting of empty files */
+  sorter_prepare_buf(ctx);
 
   // Create bucket containing the source
   struct sort_bucket *bin = sbuck_new(ctx);
@@ -225,7 +237,7 @@ 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));
@@ -240,6 +252,7 @@ sorter_run(struct sort_context *ctx)
   clist_add_head(&ctx->bucket_list, &bout->n);
 
   struct sort_bucket *b;
+  uns bits;
   while (bout = clist_head(&ctx->bucket_list), b = clist_next(&ctx->bucket_list, &bout->n))
     {
       SORT_XTRACE(2, "Next block: %s, %d hash bits", F_BSIZE(b), b->hash_bits);
@@ -247,8 +260,8 @@ sorter_run(struct sort_context *ctx)
 	sbuck_drop(b);
       else if (b->runs == 1)
 	sorter_join(b);
-      else if (sorter_radix_p(ctx, b))
-	sorter_radix(ctx, b);
+      else if (bits = sorter_radix_bits(ctx, b))
+	sorter_radix(ctx, b, bits);
       else
 	sorter_twoway(ctx, b);
     }