Follow Datanami:
October 3, 2018

SQream Adds Power9 to GPU Data Warehouse

Data warehouse vendor SQream Technologies, which recently claimed a 15-fold performance increase with the latest release of its SQreamDB, has now added support for IBM’s Power9 architecture to its GPU-accelerated database.

The New York-based database specialist with R&D operations in Tel Aviv, Israel, announced during an IBM OpenPower event in Europe on Wednesday (Oct. 3) that its SQream data warehouse will now combine the Power9 multicore design with its existing data warehouse accelerated with Nvidia’s GPUs. The combination would yield SQL query performance improvements of as much as 150 percent for Power9 users, the company claimed.

As more database users are turning to GPU acceleration to handle data growth and real-time analytics, coupling them with a range of CPUs is seen as one way of scaling and boosting overall performance for applications like machine learning and business intelligence.

SQream said it is using the Power9 architecture “as the base computing platform for significantly reducing the time of complex SQL queries while greatly increasing the amount of data analyzed.”

The GPU-accelerated data warehouse is optimized for multicore processors like Power9 as a way of handling data stores that continue to expand exponentially. The multicore architecture also incorporates Nvidia’s emerging NVLink interconnect to bridge GPUs with Power9 processors. The result is said to be increased throughput performance for both data loading and SQL query execution.

SQream said its database customers also could incorporate IBM flash storage to boost data and query performance.

The partnerships with IBM (NYSE: IBM) and Nvidia (NASDAQ: NVDA) would enable “significantly improved analytics,” claimed SQream CEO Ami Gal.

SQream is among a handful of GPU-powered relational databases that have emerged in the past few years. Buoyed by the ever-growing processing capacity of Nvidia GPUs, the database vendor claims it can process tough SQL queries against tens or hundreds of terabytes of data in a fraction of the amount of time it takes traditional MPP data warehouses.

Meanwhile, infrastructure vendors are attempting to fill development gaps with software tools, reference architectures and alternatives to traditional on-chip processing. Others, like IBM, are looking to combine new these AI development frameworks with customized hardware such as servers based on its Power9 processors.

SQream recently completed a Series B funding round that raised more than $26 million. Chinese e-commerce giant Alibaba (NYSE: BABA) is among SQream’s early investors.

Recent items:

SQream Boasts 15x Speedup for GPU Data Warehouse

GPUs Seen Lifting SQL Analytic Constraints