SQream’s GPU-Acceleration Unlocks Linear Scalability in TPC-DS

NEW YORK, July 16, 2024 — SQream, the scalable GPU data analytics platform, today announced its groundbreaking results from the TPC-DS benchmark tests, showcasing the unmatched linear scalability of its GPU-accelerated SQL engine.

These results confirm SQream’s position as a leader in data analytics and AI/ML workload acceleration as the company’s solutions proved perfect linear scalability, when running both 1TB and 10TB using the same machine and maintaining the same ratio in runtime throughout.

SQream’s recent performance in the TPC-DS benchmark tests, the gold standard to evaluate decision support solutions, demonstrates the company’s exceptional capability to handle complex and massive (terabyte and petabyte) datasets at scale. Conducted using infrastructure with 4 Tesla V100 GPU processors and 4 Intel Xeon Gold 5220 (18 cores each), the tests included 1TB and 10TB datasets.

Highlights of the Benchmark Results:

• Loading Time: 1TB in 892 seconds (14:52 minutes) and 10TB in 8260 seconds (2:17:40 hours).
• Query Time: 1TB in 1374 seconds (22:54 minutes) and 10TB in 13083 seconds (3:38:03 hours).

Compared to competitors, SQream maintained linear scalability and consistent performance with significantly fewer resources. For instance, while others are using hundreds of nodes for 100TB-scale tests, SQream demonstrates similar performance with only 10-20 GPUs.

“Our TPC-DS benchmark results clearly illustrate the comparative advantage of SQream’s GPU-accelerated platform,” said Ami Gal, CEO and Co-Founder of SQream. “Leveraging our unique GPU-parallelizing technology, we achieve unprecedented data processing speeds, unlocking efficiency and resource conservation that outpaces the current industry standard by enormous margins. SQream’s best-in-class technical performance is the bedrock of our commitment to delivering cutting-edge solutions to meet the ever-growing data processing demands. We are committed to processing AI workloads to bring unprecedented insights to modern businesses in order to offer our customers unprecedented value when compared to traditional CPU-based platforms.”

James Pak, SQream’s Regional VP Engineering, added, “For years we have known the truly incredible power that our technology offers to the marketplace, but to see it put to the test and exceed even our own expectations is truly exciting. SQream’s ability to prove consistent performance excellence across the data pipeline is what cements it’s as the leading solution for modern enterprise, where rapid and accurate data processing at all magnitudes has become crucial.”

About SQream

SQream specializes in data processing and analytics acceleration, revolutionizing the way organizations approach big data analytics and AI/ML workloads with its unique GPU-patented SQL engine. SQream’s solutions are designed to meet the needs of enterprises grappling with massive or complex datasets, offering unparalleled performance, scalability, and cost-efficiency. Tailored for industries ranging from finance to telecommunications, SQream empowers businesses to unlock actionable insights from their data with unprecedented speed and efficiency.

Source: SQream