X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fcommon.h;h=92423046d4f62bf7dbfc0ef61c17b1e6220ec97d;hb=a5ff98a53789157a6c96e58b2385bb898d688a22;hp=52949657ad411c2face099d6d3b4fd0b4ea85d36;hpb=e65a1a48088af1a5731f64b50ba933f4c29d5040;p=libucw.git

diff --git a/lib/sorter/common.h b/lib/sorter/common.h
index 52949657..92423046 100644
--- a/lib/sorter/common.h
+++ b/lib/sorter/common.h
@@ -13,18 +13,25 @@
 #include "lib/clists.h"
 
 /* Configuration, some of the variables are used by the old sorter, too. */
-extern uns sorter_trace, sorter_presort_bufsize, sorter_stream_bufsize;
-extern uns sorter_debug;
-extern u64 sorter_bufsize;
+extern uns sorter_trace, sorter_stream_bufsize;
+extern uns sorter_debug, sorter_min_radix_bits, sorter_max_radix_bits, sorter_add_radix_bits;
+extern uns sorter_min_multiway_bits, sorter_max_multiway_bits;
+extern uns sorter_threads, sorter_thread_threshold, sorter_thread_chunk;
+extern uns sorter_radix_threshold;
+extern u64 sorter_bufsize, sorter_small_input;
+extern struct fb_params sorter_fb_params, sorter_small_fb_params;
 
-#define SORT_TRACE(x...) do { if (sorter_trace) log(L_DEBUG, x); } while(0)
-#define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) log(L_DEBUG, x); } while(0)
+#define SORT_TRACE(x...) do { if (sorter_trace) msg(L_DEBUG, x); } while(0)
+#define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) msg(L_DEBUG, x); } while(0)
 
 enum sort_debug {
   SORT_DEBUG_NO_PRESORT = 1,
   SORT_DEBUG_NO_JOIN = 2,
   SORT_DEBUG_KEEP_BUCKETS = 4,
   SORT_DEBUG_NO_RADIX = 8,
+  SORT_DEBUG_NO_MULTIWAY = 16,
+  SORT_DEBUG_ASORT_NO_RADIX = 32,
+  SORT_DEBUG_ASORT_NO_THREADS = 64
 };
 
 struct sort_bucket;
@@ -34,19 +41,29 @@ struct sort_context {
   struct fastbuf *out_fb;
   uns hash_bits;
   u64 in_size;
+  struct fb_params *fb_params;
 
   struct mempool *pool;
   clist bucket_list;
-  void *big_buf, *big_buf_half;
-  size_t big_buf_size, big_buf_half_size;
+  void *big_buf;
+  size_t big_buf_size;
 
   int (*custom_presort)(struct fastbuf *dest, void *buf, size_t bufsize);
+
   // Take as much as possible from the source bucket, sort it in memory and dump to destination bucket.
   // Return 1 if there is more data available in the source bucket.
   int (*internal_sort)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only);
+
+  // Estimate how much input data from `b' will fit in the internal sorting buffer.
+  u64 (*internal_estimate)(struct sort_context *ctx, struct sort_bucket *b);
+
   // Two-way split/merge: merge up to 2 source buckets to up to 2 destination buckets.
   // Bucket arrays are NULL-terminated.
   void (*twoway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket **outs);
+
+  // Multi-way merge: merge an arbitrary number of source buckets to a single destination bucket.
+  void (*multiway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket *out);
+
   // Radix split according to hash function
   void (*radix_split)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket **outs, uns bitpos, uns numbits);
 
@@ -55,7 +72,9 @@ struct sort_context {
   int more_keys;
 
   // Timing
-  u64 start_time;
+  timestamp_t start_time;
+  uns last_pass_time;
+  uns total_int_time, total_pre_time, total_ext_time;
 };
 
 void sorter_run(struct sort_context *ctx);
@@ -63,6 +82,7 @@ void sorter_run(struct sort_context *ctx);
 /* Buffers */
 
 void *sorter_alloc(struct sort_context *ctx, uns size);
+void sorter_prepare_buf(struct sort_context *ctx);
 void sorter_alloc_buf(struct sort_context *ctx);
 void sorter_free_buf(struct sort_context *ctx);
 
@@ -99,8 +119,28 @@ struct fastbuf *sbuck_read(struct sort_bucket *b);
 struct fastbuf *sbuck_write(struct sort_bucket *b);
 void sbuck_swap_out(struct sort_bucket *b);
 
-#define F_SIZE(x) ({ byte buf[16]; format_size(buf, x); buf; })
-#define F_BSIZE(b) F_SIZE(sbuck_size(b))
-void format_size(byte *buf, u64 x);
+/* Contexts and helper functions for the array sorter */
+
+struct asort_context {
+  void *array;				// Array to sort
+  void *buffer;				// Auxiliary buffer (required when radix-sorting)
+  uns num_elts;				// Number of elements in the array
+  uns elt_size;				// Bytes per element
+  uns hash_bits;			// Remaining bits of the hash function
+  uns radix_bits;			// How many bits to process in a single radix-sort pass
+  void (*quicksort)(void *array_ptr, uns num_elts);
+  void (*quicksplit)(void *array_ptr, uns num_elts, int *leftp, int *rightp);
+  void (*radix_count)(void *src_ptr, uns num_elts, uns *cnt, uns shift);
+  void (*radix_split)(void *src_ptr, void *dest_ptr, uns num_elts, uns *ptrs, uns shift);
+
+  // Used internally by array.c
+  struct rs_work **rs_works;
+  struct work_queue *rs_work_queue;
+  struct eltpool *eltpool;
+};
+
+void asort_run(struct asort_context *ctx);
+void asort_start_threads(uns run);
+void asort_stop_threads(void);
 
 #endif