Search papers, labs, and topics across Lattice.
This paper investigates the performance of Parquet, a columnar file format, on GPU-accelerated data processing systems, revealing that CPU-centric default configurations lead to significant underutilization of GPU parallelism. Through empirical evaluation, the authors demonstrate that Parquet's performance bottleneck on GPUs stems from suboptimal configurations rather than inherent limitations of the format. By applying GPU-aware configurations, they achieve a substantial increase in effective read bandwidth, reaching up to 125 GB/s, without altering the Parquet specification itself.
Stop crippling your GPU scans: CPU-centric Parquet defaults are likely the culprit, not the format itself.
Parquet is the de facto columnar file format in modern analytical systems, yet its configuration guidelines have largely been shaped by CPU-centric execution models. As GPU-accelerated data processing becomes increasingly prevalent, Parquet files generated with CPU-oriented defaults can severely underutilize GPU parallelism, turning GPU scans into a performance bottleneck. In this work, we systematically study how Parquet configurations affect GPU scan performance. We show that Parquet's poor GPU performance is not inherent to the format itself but rather a consequence of suboptimal configuration choices. By applying GPU-aware configurations, we increase effective read bandwidth up to 125 GB/s without modifying the Parquet specification.
