Tag
LiteTopK is a fused indexer-top-k kernel for long-context sparse attention that exploits the curse of dimensionality to reduce memory traffic and improve efficiency, accelerating GLM 5.2 by 1.2x during prefill.
GreptimeDB v1.0 integrates DataFusion's dynamic filter pushdown to accelerate TopK queries by pushing runtime bounds to the scan layer, reducing query time from 29 seconds to 0.21 seconds on a 5B-row trace table.