About this document
Parallelizing Data Stream Queries by berramla27 is a document available to read on EtoBox.
The document discusses challenges and opportunities in parallelizing algorithms for frequent elements and top-k queries over data streams on chip multiprocessor (CMP) architectures. It explores naive parallelization techniques, finding that intra-operator parallelism is difficult and requires redesign. The paper then proposes CoTS, an efficient and scalable framework based on cooperative thread scheduling. Experiments show CoTS achieves over 60 million element throughput and outperforms sequential implement
- Author
- berramla27
- Language
- EN