The thread: Measured, not assumed — page 16
The file is one draw too
Two errors in pricing a library merge sort's late tail survived averaging the calibration table, and were called systematic: the first round priced 14.7% and 17.1% high at a buffer of a 128th, and a partway miss at a 1,024th. A spread index for the table was proposed to fix both. The file's groups have no spread to index — every one of 73 sits within nine points of a random run's half. Drawn eight times instead of once, the same file's first-round error ranges from −7.4% to +23.5%, and the strand's file sits at the top of the eight at one reach and near the bottom at another. What is systematic is the table's reading: a straight line in log A across a doubling overprices the bend between its cells, and read at each group's own A the first round averages within 5% over eight files and the partway case is chosen right. What remains, about eight points either way on any one file, is the file's own draw.
Two parametersThe count asks and the price answers
A landmark watch on a weighted maze returns to its four corner landmarks when a count of the traffic's stretch ends reads fast, and every rule for when to switch off has traded the moderate pace against the fast one. The count reads the traffic and never the money, while the watch already prices placements on a sample of the queries it has served. Letting the count propose a return to the corners and the watch's last price confirm it was predicted to keep the margin rule's 117.8 cells a query at the moderate pace and recover the override's 142.6 at the fast pace. It costs 118.8 and 145.1. The price agrees with the count three times in four, a quarter of those because the watch had not yet left the corners; the tolerance, swept from nought to thirty cells, changes nothing; and pricing the corners at every look costs about a cell a query on every stream, including the slow ones where the count never asks.
What the machine doesA look whose answers are kept
A bit beside each resident of a cuckoo table, saying whether its other bucket had room, had to be bought by a read the blind chain never makes, and guided too few kicks to matter. The proposal joined it to the look: read a full bucket's residents' other buckets only when their answers are missing, keep the answers, and reuse them when a later chain arrives. Predicted: under 2% of keys in overflow at a cap of 32 for under six reads an insertion. Measured: 4.8% for 6.55. The answers a bucket keeps are the ones kicks write, and a kick can only write full, so 95% of kicks at that cap find no answer saying room. Re-reading the full answers empties the overflow, at 7.31 reads — three more than random kicks need for the same result. And the answers do not keep: one that says room is true a third of the time by the thousand insertions most of them wait, not much better than the 22% a random guess achieves.
What the libraries doA choice that knows its error bar
A rule that prices every number of merging rounds for a library merge sort's late tail was given an error bar on every estimate, built from the spread of its calibration table's draws, and run on 280 sorts. It was predicted that every wrong choice would fall inside the bars, and that a wrong choice would cost under 3% of the sort. The second half holds: 30 choices were wrong, the worst cost 2.1%, and the average over all 280 is 0.14%. The first half fails five times, all at one buffer and one reach, where the bars were near zero and the estimates of two options differed by six moves. They were not guesses the rule failed to flag. They were a step in the cost that the table reads as a ramp, and a bar built from draws cannot see an error that every draw shares.
Two parametersA price paid only when asked
A landmark watch returns to its corner landmarks when a count of the traffic's stretch ends reads fast, and confirming that proposal with a price of the corners cost a cell or two a query because the price was paid at every look. Paid only at the moment the count proposes, the confirmation was predicted to be free on six jump streams and, on traffic that drifts between jumps, to refuse the count's misreadings and beat the margin rule by more than a cell. It is nearly free: within a fifth of a cell of the margin rule on four streams and 0.6 at worst, its searching down from 0.4–2.1 cells a query to at most 0.1. It refuses nothing that matters. On drifting traffic it costs 119.8 against the margin rule's 119.7 while the watch left on costs 115.4, because a proposal comes just after the destination moves, and at that moment a placement fitted to where the traffic was looks far worse than the corners: 27 of 35 prices agree, 13 of them by more than ten cells a query.