X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fs-internal.h;h=35ba1f50dde6ab00ddb1c98495f5900c5ebf542f;hb=b20998a1656dafd2cf694b7f4d5807d8321db43a;hp=a2f9e06bc27fd2bcf337dd48b9429c6e97e1dd0f;hpb=c0544008a494710146ac6cf88b61b5fd2d46f51b;p=libucw.git

diff --git a/lib/sorter/s-internal.h b/lib/sorter/s-internal.h
index a2f9e06b..35ba1f50 100644
--- a/lib/sorter/s-internal.h
+++ b/lib/sorter/s-internal.h
@@ -115,7 +115,7 @@ static int P(internal)(struct sort_context *ctx, struct sort_bucket *bin, struct
 #ifdef SORT_VAR_DATA
   if (sizeof(key) + 2*CPU_PAGE_SIZE + SORT_DATA_SIZE(key) + P(internal_workspace)(&key) > bufsize)
     {
-      SORT_XTRACE(3, "s-internal: Generating a giant run");
+      SORT_XTRACE(4, "s-internal: Generating a giant run");
       struct fastbuf *out = sbuck_write(bout);
       P(copy_data)(&key, in, out);
       bout->runs++;
@@ -123,7 +123,7 @@ static int P(internal)(struct sort_context *ctx, struct sort_bucket *bin, struct
     }
 #endif
 
-  SORT_XTRACE(4, "s-internal: Reading");
+  SORT_XTRACE(5, "s-internal: Reading");
   P(internal_item_t) *item_array = ctx->big_buf, *item = item_array, *last_item;
   byte *end = (byte *) ctx->big_buf + bufsize;
   size_t remains = bufsize - CPU_PAGE_SIZE;
@@ -165,23 +165,24 @@ static int P(internal)(struct sort_context *ctx, struct sort_bucket *bin, struct
 
   uns count = last_item - item_array;
   void *workspace UNUSED = ALIGN_PTR(last_item, CPU_PAGE_SIZE);
-  SORT_XTRACE(3, "s-internal: Read %u items (%s items, %s workspace, %s data)",
+  SORT_XTRACE(4, "s-internal: Read %u items (%s items, %s workspace, %s data)",
 	count,
 	stk_fsize((byte*)last_item - (byte*)item_array),
 	stk_fsize(end - (byte*)last_item - remains),
 	stk_fsize((byte*)ctx->big_buf + bufsize - end));
   timestamp_t timer;
   init_timer(&timer);
-  item_array = P(array_sort)(item_array, count,
+  item_array = P(array_sort)(item_array, count
 #ifdef SORT_INTERNAL_RADIX
-    workspace, bin->hash_bits
-#else
-    NULL, 0
+    , workspace, bin->hash_bits
 #endif
     );
+  if ((void *)item_array != ctx->big_buf)
+    workspace = ctx->big_buf;
+  last_item = item_array + count;
   ctx->total_int_time += get_timer(&timer);
 
-  SORT_XTRACE(4, "s-internal: Writing");
+  SORT_XTRACE(5, "s-internal: Writing");
   if (!ctx->more_keys)
     bout = bout_only;
   struct fastbuf *out = sbuck_write(bout);
@@ -219,7 +220,7 @@ static int P(internal)(struct sort_context *ctx, struct sort_bucket *bin, struct
 #endif
     }
 #ifdef SORT_UNIFY
-  SORT_XTRACE(3, "Merging reduced %u records", merged);
+  SORT_XTRACE(4, "Merging reduced %u records", merged);
 #endif
 
   return ctx->more_keys;
@@ -228,17 +229,24 @@ static int P(internal)(struct sort_context *ctx, struct sort_bucket *bin, struct
 static u64
 P(internal_estimate)(struct sort_context *ctx, struct sort_bucket *b UNUSED)
 {
+  // Most of this is just wild guesses
 #ifdef SORT_VAR_KEY
-  uns avg = ALIGN_TO(sizeof(P(key))/4, CPU_STRUCT_ALIGN);	// Wild guess...
+  uns avg = ALIGN_TO(sizeof(P(key))/4, CPU_STRUCT_ALIGN);
 #else
   uns avg = ALIGN_TO(sizeof(P(key)), CPU_STRUCT_ALIGN);
 #endif
-  // We ignore the data part of records, it probably won't make the estimate much worse
-  size_t bufsize = ctx->big_buf_size;
-#ifdef SORT_UNIFY_WORKSPACE		// FIXME: Or if radix-sorting
-  bufsize /= 2;
+  uns ws = 0;
+#ifdef SORT_UNIFY
+  ws += sizeof(void *);
 #endif
-  return (bufsize / (avg + sizeof(P(internal_item_t))) * avg);
+#ifdef SORT_UNIFY_WORKSPACE
+  ws += avg;
+#endif
+#ifdef SORT_INTERNAL_RADIX
+  ws = MAX(ws, sizeof(P(internal_item_t)));
+#endif
+  // We ignore the data part of records, it probably won't make the estimate much worse
+  return (ctx->big_buf_size / (avg + ws + sizeof(P(internal_item_t))) * avg);
 }
 
 #undef SORT_COPY_HASH