Theme

The thread: The unit of cost is not one — page 2

Page 2 of 2, continuing the same thread in the same order.
Merge sortnone · 55% tiesMerge sort with an insertion cutoffnone · 47% tiesInsertion sortnone · 0% tiesBubble sortnone · 55% tiesShellsort2,014 · 71% tiesHeapsort2,403 · 32% tiesSelection sort1,209 · 24% tiesQuicksort, median-of-three pivot1,428 · 99% tiesequal pairs left out of the order they arrived in4,096 records, 8 distinct keysdark: the sorts that move them Counting

The order equal keys keep

Four of the eight sorts here leave every pair of equal keys in the order it arrived in and four move between 1,209 and 2,403 pairs, and none of the four counts every plate here reports can tell them apart. Decorating each record with its arrival position makes any of them stable, for four thousand words and between 0.05 and 1.64 times its comparisons — a charge of 64% on Shellsort and a saving of 95% on quicksort, because the ties stability has to break are the ties a two-way partition chokes on.

1234567characters addedoccurrencesgrown leftwardsgrown rightwards"tgatatt"4,000 characters · four symbols1 occurrences The index that replaces the text

An interval that grows at both ends

A backward search step is two ranks. A bidirectional step is two ranks and the width of the interval for every symbol that sorts before the one being added — 16.5 ranks on four symbols and 138.6 on twenty-six, which is a cost no account of the structure mentions.

reads and writescache missesRadix sort, 8-bit digitsno comparisonsMerge sort965,752 comparisonsHeapsort1,895,405 comparisonsQuicksort, median-of-three1,187,435 comparisons65,536 keys of 32 bitsdark: the sort that compares nothing Counting

The sort that makes none of them

Every count on this collection is a count of comparisons, swaps, reads or writes, and radix sort makes zero of the first. On 65,536 keys it moves five times less data than merge sort, misses the cache three times more, and sits 954,037 comparisons under the floor no comparison sort can go beneath — which is not an achievement, because the floor was never a statement about it.

ranks to compute D72extensions removed27,906extensions remaining12,051one search · k = 3 · 4,000 charactersand one more index: 17,033 bits4,000 characters · m = 16388 extensions a rank The other axis

A bound that has to be paid for

The pruning removes seventy per cent of a search tree for seventy-two rank operations. It also needs an FM-index of the reversed text — 17,033 bits against the forward index's 17,032 — which doubles the structure whose small size was the entire argument for walking an index.

weighted path lengthΣ wᵢdᵢ — what Huffman minimisescuts takenΣ over the mergesdamageworst error leftchaintreesmallest-firstlargest-first543k147k123k738k309254259388323148183403Misra-Gries · 32 shards · hashedleast path smallest · least damage balanced Structures

The fold that minimises the wrong thing

A fold charges per level and a survivor pays the cuts on its path, so the bill looks like a weighted external path length — and Huffman's construction minimises that quantity by proof. Built and measured on thirty-two uneven shards it does minimise it, 181,407 against a balanced tree's 200,000, and leaves more damage than the tree does.

run together38,400 bitsσ 21 · 5 bitsone separator38,470 bitsσ 22 · 5 bitsa separator each46,164 bitsσ 35 · 6 bits15 documents of 512 charactersthe separators are the only differenceEnglish-like · 15 documents1.20x for the distinct marks The other axis

One separator, or one for each

A shared separator costs one alphabet symbol and is free. Fifteen distinct ones take the alphabet from twenty-two to thirty-five, which crosses a power of two, so every character of every document costs a sixth bit — 1.2 times the packed collection, to tell the boundaries apart.

counter tablesworst error, ratio to bestquantile summariestuples kept, ratio to bestchain2.18× (323)1.10× (2,807)tree1.00× (148)1.26× (3,211)smallest-first1.24× (183)1.28× (3,278)largest-first2.72× (403)1.00× (2,556)32 shards · hashed · k = 32each column against its own best shape Structures

The shape one structure will not fold

Folding thirty-two shards largest-pair-first keeps 2,556 quantile tuples against a balanced tree's 3,211 — a fifth of the space saved. The same fold on the counter tables beside them leaves 403 counts of error against the tree's 148. A deployment holding both cannot fold once and be right twice.

32641282565121010010³10⁴elements sortedword operations a comparison must cost to break evenSelection sort, pairsSelection sort, closureBubble sort, pairsBubble sort, closureHeapsort, pairsHeapsort, closureQuicksort, median-of-three, pairsQuicksort, median-of-three, closuresolid: pairs asked · dashed: everything impliedrandom input · counted exactly Counting

The price of remembering an answer

Give selection sort a table of the pairs it has already compared and it makes 16,805 comparisons on 256 elements instead of 32,640. The table pays for itself once a comparison costs more than 3.1 word operations, and that price stays near three at every size measured. Give it the full closure of everything its answers imply and it makes 13,079. The closure pays only past 425 operations a comparison at that size, and past 1,659 at twice the size. The cheap memory is worth having on long keys. The complete one is worth having almost nowhere.

0%1.1%2.1%3.2%4.2%the whole filterblocks of 64blocks of 5120123456distinct 512-bit lines a lookup readsabsent keys answered yes16,384 bits, 2,048 keys, k = 6one line is what a block buys When the algorithm flips a coin

Positions confined to one line

A Bloom filter lookup reads 5.55 cache lines because its six positions are scattered across the whole filter. Confining them to a 512-bit block makes it exactly one, and costs 7% more false positives at eight bits a key. At sixteen bits a key the same block costs 91%, and the two-value trick that is free across a whole filter costs another 135% inside one — because a block is a small filter, and small filters are where the penalty lives.

6080100how far a full leaf looks for room, in siblingsmean leaf fill, per cent12481664randomascendingdescendingascending, 1% late131,072 keys · leaves of 64 · fan-out 64a reach of 1 is the adjacent-sibling rule When it does not fit

A key passed along the row

A full B+-tree leaf that offers a key to its immediate neighbours before splitting leaves an ascending stream 67.2% full. Let it look one sibling further and the same stream fills its leaves completely, 2,048 leaves where there were 3,048, for about the same key moves. On random keys each doubling of the reach adds a few points of fill and more writes than it saves, and at the whole parent a key travels sixteen leaves on average. On a stream that is mostly in order the same reach costs almost nothing.

48163264128110entry width, bytescache lines a lookup readsentries inlinea line of tags in front8,192 slots, load 0.9, buckets of 8keys the table holds What the machine does

A lookup that stops caring how wide an entry is

Buckets of eight entries aligned to a cache line read 1.20 lines a lookup when an entry is eight bytes and 19.25 when it is 128, because the bound was arithmetic about alignment and the arithmetic stops holding. Keeping one byte of each key's hash in a separate array and the entries in a parallel one reads 2.21 lines at every width from four bytes to sixty-four — and for a key the table does not hold, 2.05 against 31.98.

8121620240.000010.00010.0010.01bits a keyfalse-positive ratethe whole filterone block of 512 bitsthe emptier of two 512-bit blocksone block of 1,024 bits2,048 keys · 24 filters a pointdashed: no blocks When the algorithm flips a coin

Two blocks and the chances they add

Send each key to the emptier of two 512-bit blocks and the busiest block of a filter at sixteen bits a key holds 38 keys instead of 55. The false-positive rate does not move: 0.100% against 0.095%. At eight bits a key it doubles. A lookup cannot tell which block a key went to, so it has to ask both, and asking twice is two chances to be wrong. The repair that tames a hash table's worst bucket buys a filter nothing that a block twice as wide does not.

the whole table60,00060,000r · 0rkcounting filter24,84315,369r · 0rkseed filter17,7452,415r · 1,377rkindex walk00r · 177,046rkn = 3,000 · m = 20 · q = 4cells drawn · r = characters read, rk = index rankscells computed6 occurrences What is taught wrongly

Three savings in three currencies

The same six occurrences, found four ways. The table computes 60,000 cells and reads 60,000 characters. The counting filter computes 6,251 cells and reads 13,022 characters. The seed filter computes 3,325 and reads 550. The index walk computes none, reads none, and performs 18,645 ranks.

candidates, filtering4,355candidates, grid33rank and select, grid2,471occurrences32one query · same collection · same answer16,384 characters · 32 occurrencescandidates 132x · operations 1.76x What is taught wrongly

The operations a candidate count leaves out

The grid examines 33 candidates where the scan examines 4,355 — a factor of 132. Counted in the operations each of them performs, the same query is 2,471 against 4,355, and the factor is 1.8.

101001,00010100total length of the pattern settimes the published rule's precomputationwritten as a definitionbuilt from the linksthe published rulefour symbols · m = 10137x becomes 1.60x What is taught wrongly

The ratio that was an implementation

This collection published a factor of forty-two between two shift rules' preprocessing. Sixty-nine per cent of the denominator was a table the published rule never reads, and the numerator was a definition rather than a construction. The corrected ratio is 1.6.

level 0 · left335 keptlevel 1 · right+207 smallerlevel 2 · left61 keptlevel 3 · right+33 smallerlevel 4 · left8 keptpositions still in play, and the half the code leaves behindsmaller symbols before position 400: 240rank of "m": 8 — from the same 5 operationsσ 21 · code 010105 ranks, not 105 The index that replaces the text

Every child at once

A bidirectional extension counts the occurrences of every symbol smaller than the one being added, and this collection did it with one rank per symbol — 139.7 operations an extension on a twenty-six-letter alphabet. One walk down the tree gives the same number.

level 0 · left585 keptlevel 1 · right+351 smallerlevel 2 · right+113 smallerlevel 3 · right+39 smallerlevel 4 · right+40 smallerpositions still in play, and the half the code leaves behindsmaller symbols before position 700: 543rank of "s": 42 — from the same 5 operationsσ 21 · code 011115 ranks, not 105 What the machine does

The count that was already there

A bit vector's rank of one is the position minus its rank of zero. Every walk down a wavelet tree computes one of them at every level, and this collection asked for the other separately for years.

acdefghijklmnopqrstuvwyzroot24 of 26 symbols present54 ranks · 51 nodes What the machine does

A node costs two ranks

The left child's interval is the position minus the right child's. A descent that calls rank on both children returns exactly the same symbols at twice the cost, and nothing about the answer can see it.

Thompson's NFA, a set of states43.724 to buildthe subset DFA, built in full1.011,297 to buildthe subset DFA, built on demand41.824 to buildthe state set in machine words37.224 to buildoperations per character; the note is what was paid once, before the first character4,096 characters · k = 643.7 down to 1 What the machine does

What a character costs on four machines

Forty-four operations, one, forty-two, and a number that moves. Four machines for one language, with the construction charged separately from the steps, because a machine that is free per character paid eleven thousand operations before the first one.

rounds, if every ready cell ran at oncerow by row65,793column by column65,793anti-diagonal by anti-diagonal513reads that miss the cacherow by row6.3%column by column28.2%anti-diagonal by anti-diagonal31.1%fully associative · 32 lines × 8 elements · LRU263,169 reads and writes per order What the machine does

The order with the best depth

An edit-distance table can be filled row by row, column by column, or one anti-diagonal at a time, and the anti-diagonal order is the one that needs the fewest rounds — 513 against 65,793 on two strings of 256 characters, because every cell on an anti-diagonal is independent of the others. Stored the usual way, row by row, it also misses the cache on 31.1% of its reads, where row order misses 6.3%. The order that is best for parallel work is worst for the memory it runs on.

linear probingchainedcuckoo, two tables0240.10.20.30.40.50.60.70.80.9load factorentries read per lookup0120.10.20.30.40.50.60.70.80.9load factorcache misses per lookup8,192 slots, 64 cache lines of 8a probe is an entry read; a miss is a line fetched What the machine does

Two probes are two misses

Cuckoo hashing's lookup reads at most two slots, and at a load of 0.45 it reads 1.27 on average where linear probing reads 1.39. Replayed through a cache, it misses 1.18 times a lookup where linear probing misses 0.98. The table that wins the count the analysis uses loses the count the machine charges, because two slots in unrelated places are two cache lines, and a run of adjacent slots is usually one.

acdefghijklmnopqrstuvwyzroot24 of 26 symbols present54 ranks · 51 nodes The index that replaces the text

Asking about symbols that are not there

A search extends an interval by every character of the alphabet, and on a deep branch almost all of them produce an empty interval. That is a full rank walk whose entire result is the discovery that nothing was there.

linear probingcuckoo, two tablescuckoo, buckets of 8024680.30.450.60.750.850.95load factorentries read per lookup00.511.50.30.450.60.750.850.95load factorcache misses per lookup8,192 slots, 64 cache lines of 8a probe is an entry read; a miss is a line fetched What the machine does

The bucket that fits a line

Make each of a cuckoo table's two candidates a bucket of eight slots laid out on one cache line, and no lookup ever touches more than two lines, the table builds past a load of 0.95, and at that load it misses 1.21 times a lookup where linear probing misses 1.79. The prediction that it would lose to linear probing at low loads was wrong — it misses less at every load measured, 0.94 against 0.96 at 0.3 — because a key it holds almost never lives in its second bucket. The guarantee belongs to the alignment, not the bucket; eight slots on lines of four put a lookup on four lines.

extensions attempted9,206 dead7,594 livebit-vector ranksthe loop: 168,000the descent: 24,0868 patterns · 1 error · sigma 2054.8% dead · 6.98x Structures

The branches that find nothing

An approximate search over a twenty-symbol alphabet attempts sixteen thousand eight hundred extensions and nine thousand two hundred of them produce an empty interval. That is a full rank walk whose entire result is the discovery that nothing was there.

acdefghijklmnopqrstuvwyzroot24 of 26 symbols present54 ranks · 51 nodes What is taught wrongly

A walk that does not prune

Remove the emptiness test and the descent visits every node of the tree, returns exactly the same symbols with exactly the same intervals, and costs sixty per cent more. No test of the answer can see it.

124816326410step width, cellscache lines a stepstored by diagonalsby diagonals, each line-alignedstored by rows512 × 512, lines of 8 cellsone step of an anti-diagonal What the machine does

Eight cells at once

The anti-diagonal fill order exists because its cells do not depend on one another, and every table filled here has been walked one cell at a time anyway. Computed eight at a time, a step touches 5.71 cache lines on the layout that stores the table by diagonals and 10.87 on the one that stores it by rows — and per cell the first keeps falling to 0.42 while the second stops at 1.27. The prediction that a diagonal step would touch three or four lines was wrong, and line-aligning each diagonal only takes it to 4.94.

bits heldthe model3,591the structure3,905one rank, in operationsbinary search10 stepsselect and walk1 select + 0.7016,385 rows · one in 32 marked14x on the operation What is taught wrongly

A price with no structure under it

A class in this collection has charged Elias–Fano's price for several strands and stores an array of positions searched by binary search. The accounting is right about space to a bit per thousand and wrong about one operation by a factor of eight.

All threads