Cloud GPUs Aimed at Data Scientists
Core Scientific, an AI and cloud infrastructure vendor, is teaming with GPU-accelerated analytics specialist SQream Technologies to deliver what the partners tout as a “GPU Cloud for Data Scientists.”
The goal of the collaboration is accelerating data analytics using GPUs for machine and deep learning. The partners said Thursday (April 23) they are focusing on data science projects in the financial services, health care, pharmaceutical and telecommunications sectors.
Core Scientific, Bellevue, Wash., specializes in AI hosting and blockchain technologies used for transaction processing and application development. The two-year-old startup released its data science cloud earlier this year. The partnership with SQream adds an GPU accelerator to Core Scientific’s cloud platform to enable data scientists to crunch much larger data sets used for training and inference.
The combination is billed as allowing data scientists to load terabytes of data while analyzing petabyte-size data stores. The data science cloud also is promoted as combining the convenience of public clouds with the advantaged of co-located infrastructure.
The data science cloud would leverage SQreamDB, the company’s GPU-accelerated data warehouse that supports SQL. The database is intended to load, store and analyze large data sets.
The partnership with Core Scientific “will make SQreamDB’s accelerated analytics platform accessible to more organizations, enabling them to rapidly analyze significantly more data,” said SQream CEO Ami Gal.
Awhile back, New York-based SQream announced its data warehouse would support the Power9 multicore design along 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.
A host of emerging database technologies are leveraging GPU accelerators as vendors like Nvidia (NASDAQ: NVDA) target data scientists looking to crunch ever-larger data sets using parallel computing schemes that use graphics processor to alongside CPUs.
The Core Scientific-SQream partnership appears to take that approach a step further by hosting those analytics engines in the cloud or, in the case of Core Scientific, co-located infrastructure.