Reduce size of sort_key from k + 1 to k bits in phases 2 and 3

k bits are sufficient to encode table entry position after removing unused entries in phase 2.
SkaZun · Apr 2, 2021 · 7bacb3a · 7bacb3a
1 parent a3c390f
commit 7bacb3a
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 10 deletions.
diff --git a/src/entry_sizes.hpp b/src/entry_sizes.hpp
@@ -60,8 +60,8 @@ class EntrySizes {
                     //    a:  sort_key, pos, offset        or
                     //    b:  line_point, sort_key
                     return Util::ByteAlign(
-                               std::max(static_cast<uint32_t>(k + 1 + (k) + kOffsetSize),
-                                   static_cast<uint32_t>(2 * k + k + 1))) /
+                               std::max(static_cast<uint32_t>(2 * k + kOffsetSize),
+                                   static_cast<uint32_t>(3 * k))) /
                            8;
             case 7:
             default:

diff --git a/src/phase2.hpp b/src/phase2.hpp
@@ -52,11 +52,11 @@ Phase2Results RunPhase2(
     uint32_t const log_num_buckets,
     bool const show_progress)
 {
-    // An extra bit is used, since we may have more than 2^k entries in a table. (After pruning,
-    // each table will have 0.8*2^k or fewer entries).
+    // After pruning each table will have 0.865 * 2^k or fewer entries on
+    // average
     uint8_t const pos_size = k;
     uint8_t const pos_offset_size = pos_size + kOffsetSize;
-    uint8_t const write_counter_shift = 128 - (k + 1);
+    uint8_t const write_counter_shift = 128 - k;
     uint8_t const pos_offset_shift = write_counter_shift - pos_offset_size;
     uint8_t const f7_shift = 128 - k;
     uint8_t const t7_pos_offset_shift = f7_shift - pos_offset_size;
@@ -160,7 +160,7 @@ Phase2Results RunPhase2(
             uint16_t(entry_size),
             tmp_dirname,
             filename + ".p2.t" + std::to_string(table_index),
-            uint32_t(k) + 1,
+            uint32_t(k),
             0,
             strategy_t::quicksort_last);
 

diff --git a/src/phase3.hpp b/src/phase3.hpp
@@ -173,9 +173,10 @@ Phase3Results RunPhase3(
         Disk& right_disk = res2.disk_for_table(table_index + 1);
         Disk& left_disk = res2.disk_for_table(table_index);
 
-        // Sort key for table 7 is just y, which is k bits. For all other tables it can
-        // be higher than 2^k and therefore k+1 bits are used.
-        uint32_t right_sort_key_size = table_index == 6 ? k : k + 1;
+        // Sort key is k bits for all tables. For table 7 it is just y, which
+        // is k bits, and for all other tables the number of entries does not
+        // exceed 0.865 * 2^k on average.
+        uint32_t right_sort_key_size = k;
 
         uint32_t left_entry_size_bytes = EntrySizes::GetMaxEntrySize(k, table_index, false);
         right_entry_size_bytes = EntrySizes::GetMaxEntrySize(k, table_index + 1, false);
@@ -302,7 +303,7 @@ Phase3Results RunPhase3(
                 } else {
                     // k+1 bits in case it overflows
                     left_new_pos[current_pos % kCachedPositionsSize] =
-                        Util::SliceInt64FromBytes(left_entry_disk_buf, k + 1, k + 1);
+                        Util::SliceInt64FromBytes(left_entry_disk_buf, right_sort_key_size, k + 1);
                 }
             }