LLM-synthesized GPU query kernels outperform engineered engines by 7.4x, but a portable SYCL engine with lifted optimizations closes the gap to 1.27x, suggesting engineering remains preferable on GPUs.
Patel, and Rodrigo Aramburú
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
Online analytical processing of queries on datasets in the many-terabyte range is only possible with costly distributed computing systems. To decrease the cost and increase the throughput, systems can leverage accelerators such as GPUs, which are now ubiquitous in the compute infrastructure. This introduces many challenges, the majority of which are related to when, where, and how to best move data around the system. We present Theseus -- a production-ready enterprise-scale distributed accelerator-native query engine designed to balance data movement, memory utilization, and computation in an accelerator-based system context. Specialized asynchronous control mechanisms are tightly coupled to the hardware resources for the purpose of network communication, data pre-loading, data spilling across memories and storage, and GPU compute tasks. The memory subsystem contains a mechanism for fixed-size page-locked host memory allocations to increase throughput and reduce memory fragmentation. For the TPC-H benchmarks at scale factors ranging from 1k to 30k on cloud infrastructure, Theseus outperforms Databricks Photon by up to $4\times$ at cost parity. Theseus is capable of processing all queries of the TPC-H and TPC-DS benchmarks at scale factor 100k (100 TB scale) with as few as 2 DGX A100 640GB nodes.
fields
cs.DB 4representative citing papers
PystachIO is a PyTorch-based distributed OLAP engine that delivers up to 3x end-to-end speedups for storage-resident queries by combining fast RDMA networks, NVMe storage, and I/O-computation overlap optimizations.
Presto is extended to GPU-aware execution using cuDF experiments on TPC-H, delivering up to 6x cost/performance gains over CPU Presto via optimized data paths and inter-operator communication.
Relational engines achieve faster SQL+vector-search queries on GPU than CPU when using compact vector indexes and fast interconnects, reversing the CPU-only design in current systems.
citing papers explorer
-
From Custom-Fit to Portable: Bridging the Gap Between Synthesized and Engineered GPU Query Execution
LLM-synthesized GPU query kernels outperform engineered engines by 7.4x, but a portable SYCL engine with lifted optimizations closes the gap to 1.27x, suggesting engineering remains preferable on GPUs.
-
PystachIO: Efficient Distributed GPU Query Processing with PyTorch over Fast Networks & Fast Storage
PystachIO is a PyTorch-based distributed OLAP engine that delivers up to 3x end-to-end speedups for storage-resident queries by combining fast RDMA networks, NVMe storage, and I/O-computation overlap optimizations.
-
Accelerating Presto with GPUs
Presto is extended to GPU-aware execution using cuDF experiments on TPC-H, delivering up to 6x cost/performance gains over CPU Presto via optimized data paths and inter-operator communication.
-
To GPU or Not to GPU: Vector Search in Relational Engines
Relational engines achieve faster SQL+vector-search queries on GPU than CPU when using compact vector indexes and fast interconnects, reversing the CPU-only design in current systems.