Parallelized data processing throughput scales with CPU cores

Drill-downs running over 113.2B quotes fully loaded and prepared

FOR CONTEXTWhere we measure: query compute Drill-downs Interactive aggs ClickHouse Cloud · Fresh-data pathQuery-ready dataMergeTree tableOrderedraw dataIncremental MVOrdered pre-aggregatedRead serviceQuerycomputeFastanswerscontinuouslyFreshdatacontinuouslyIngest service2 nodes2 CPUs / 8 GiB each
Drag to explore59 cores

Fixed 113.2-billion-quote dataset on one 59-vCPU, 236-GiB query node. Hot caches and cold data reads with query-condition caching disabled. Select 2, 4, 8, 16, 30 or 59 CPU cores. CPU utilization is measured during execution and normalized to the selected core budget. Memory shows the average of measured per-query peaks.