ClickHouse's Hash Table Marks Empty Cells With Zero Bytes: No Tombstone, and Floats Compare Bitwise
ClickHouse’s hash table has no occupancy bit, no tombstone, and no sentinel pointer.
Posts
All posts published on this site.
Everything published here, newest first.
ClickHouse's uniq Estimates Distinct Counts by Thinning: The Wide Set That Lifts the 32-Bit Ceiling
uniq() is a hash table that deliberately throws data away. When it fills up, it halves its sample rate, keeps the survivors, and multiplies the count back up by the…
ClickHouse's Query Result Cache Never Invalidates: A Hash of the AST and a TTL
ClickHouse caches query results by hashing the parsed query, not the text you typed, plus the settings and database that could change the answer.
CPython's Garbage Collector: Generational Collection at Work
CPython frees most objects the instant their refcount hits zero; the cyclic GC is the safety net for reference cycles, using three generations and a gc_ref copy of the refcount.
ClickHouse JIT-Compiles Scalar Expressions with LLVM: From ActionsDAG to an RX Page
ClickHouse compiles hot scalar expressions to native code with LLVM.