mirror of
https://github.com/hanzoai/datastore.git
synced 2026-09-26 13:55:07 +00:00
murphy-4oandClaude
1c439ee68e
OR disjuncts with nullable keys and an ON-section mask, forced through 1024-row chunks: the shape where each resumed chunk re-prepares the skip bytes of its remaining tail. Measured +2.5% vs master at this 64x-smaller- than-default cap and below noise at the default; the query pins that it stays that way. Requested in review. Co-Authored-By: Claude <[email protected]>
90 lines
4.4 KiB
XML
90 lines
4.4 KiB
XML
<test>
|
|
<!-- Probe-dominated hash joins against small (cache-resident) build sides with Nullable
|
|
keys. With a tiny hash table the probe loop iteration is only a few cycles, so any
|
|
extra per-row work in `HashJoinMethods::joinRightColumns` shows up here first; the
|
|
Nullable-key shape exercises the null-map handling of the probe loop, which is not
|
|
covered by the other join tests (their Nullable cases use large build sides). -->
|
|
|
|
<query>
|
|
SELECT * FROM (SELECT toNullable(number) AS k FROM numbers_mt(5e8)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs USING (k)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash'
|
|
</query>
|
|
|
|
<query>
|
|
SELECT * FROM (SELECT if(number % 10 = 0, NULL, number) AS k FROM numbers_mt(5e8)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs USING (k)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash'
|
|
</query>
|
|
|
|
<query>
|
|
SELECT count() FROM (SELECT toNullable(number) AS k FROM numbers_mt(1e8)) AS lhs
|
|
INNER JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs USING (k)
|
|
FORMAT Null
|
|
SETTINGS max_threads=1, join_algorithm='hash'
|
|
</query>
|
|
|
|
<!-- Non-nullable keys with a join mask of kind `Unknown`: a left-side-only condition in the
|
|
ON section stays inside the JOIN as the probe-side mask column (it cannot be pushed to
|
|
WHERE without changing LEFT JOIN semantics). -->
|
|
<query>
|
|
SELECT * FROM numbers_mt(5e8) AS lhs
|
|
LEFT JOIN (SELECT number AS k FROM numbers_mt(1e3)) AS rhs ON lhs.number = rhs.k AND (lhs.number % 2 = 0)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash'
|
|
</query>
|
|
|
|
<!-- Nullable keys combined with an ON-section mask: both skip sources merged. -->
|
|
<query>
|
|
SELECT * FROM (SELECT toNullable(number) AS k FROM numbers_mt(5e8)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs ON lhs.k = rhs.k AND (lhs.k % 2 = 0)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash'
|
|
</query>
|
|
|
|
<!-- Multiple disjuncts (OR) with nullable keys: the multi-clause probe loop. -->
|
|
<query>
|
|
SELECT count() FROM (SELECT toNullable(number) AS k FROM numbers_mt(2e8)) AS lhs
|
|
LEFT JOIN (SELECT number*2 AS a, number*3 AS b FROM numbers_mt(1e3)) AS rhs ON lhs.k = rhs.a OR lhs.k = rhs.b
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash'
|
|
</query>
|
|
|
|
<!-- The same nullable shape probed through scattered blocks (index selector). -->
|
|
<query>
|
|
SELECT * FROM (SELECT toNullable(number) AS k FROM numbers_mt(5e8)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs USING (k)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='parallel_hash'
|
|
</query>
|
|
|
|
<!-- Masked shapes through scattered blocks: each shard merges the skip bytes over the
|
|
source block, which must stay cheaper than the per-row mask reads it replaces. -->
|
|
<query>
|
|
SELECT * FROM numbers_mt(5e8) AS lhs
|
|
LEFT JOIN (SELECT number AS k FROM numbers_mt(1e3)) AS rhs ON lhs.number = rhs.k AND (lhs.number % 2 = 0)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='parallel_hash'
|
|
</query>
|
|
|
|
<query>
|
|
SELECT * FROM (SELECT toNullable(number) AS k FROM numbers_mt(5e8)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number) AS k FROM numbers_mt(1e3)) AS rhs ON lhs.k = rhs.k AND (lhs.k % 2 = 0)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='parallel_hash'
|
|
</query>
|
|
|
|
<!-- Continuation chunks: a probe stopped early by max_joined_block_size_rows re-prepares
|
|
the skip bytes of its remaining tail on every resumed chunk; this stays bounded by
|
|
the selector and must remain within noise of the per-row checks it replaced. -->
|
|
<query>
|
|
SELECT count() FROM (SELECT if(number % 7 = 0, NULL, number % 50) AS k, toUInt8(number % 3) AS cond FROM numbers_mt(2000000)) AS lhs
|
|
LEFT JOIN (SELECT toNullable(number % 25) AS a, toNullable(number % 30) AS b FROM numbers(2000)) AS rhs
|
|
ON (lhs.k = rhs.a AND lhs.cond) OR (lhs.k = rhs.b)
|
|
FORMAT Null
|
|
SETTINGS max_threads=8, join_algorithm='hash', join_use_nulls=1, max_joined_block_size_rows=1024
|
|
</query>
|
|
</test>
|