That might interest developers this month RAPIDS accelerator for Apache Spark v21.10 has been made available and the open source project is now available for download. The latest version contains a wealth of Community requests which are ideally suited for GPU acceleration. Such as new performance improvements and cost savings, as well as new I / O and nested data type qualification and profiling tool functionality along with updates to the Spark sample repository.
The development team has also announced that upcoming releases will introduce support for 128-bit decimal data types, inference support for the principal component analysis algorithm, and additional support for nested data types for multilevel structures and maps. As well as MIG support Per NVIDIA Amp architecture GPUs (A100 / A30) that can improve throughput when running multiple Spark jobs with A100. As always, we’d like to thank all of you for using RAPIDS Accelerator for Apache Spark and look forward to hearing from you. Contact us on GitHub and let us know how we can improve your experience with RAPIDS Accelerator on Apache Spark.
RAPIDS accelerator for Apache Spark
“RAPIDS Accelerator for Apache Spark is growing at a rapid pace in terms of both functionality and performance. Industry benchmarks are a great way to measure performance over time, but another barometer of measuring performance is to measure the performance of common operators used in the data preprocessing phase or in data analysis. Most Apache Spark users know that Spark 3.2 was released this October. Version 21.10 supports Spark 3.2 and CUDA 11.4. In this release, we’ve focused on expanding support for I / O, nested computing, and machine learning capabilities. RAPIDS Accelerator for Apache Spark v21.10 has released a new plug-in jar that supports machine learning in Spark. “
“In addition to the plug-in, several new features have been added to the RAPIDS Accelerator for the Apache Spark qualification and profiling tool. The qualification tool can now report the various nested data types and write existing data formats. It also now has support for adding conjunctive and disjunctive filters, as well as filter-based regular expressions and usernames. The qualifications tool isn’t the only one with new tricks: the profiling tool now offers a structured output format and supports scaling and running a large number of event logs. “
For more information on the new features and additions to RAPIDS Accelerator for Apache Spark v21.10, please visit the NVIDIA official website by following the link below.
Source : NVIDIA
Filed under: Gadget news
Latest geeky gadget deals
Disclosure: Some of our articles contain affiliate links. Geeky Gadgets can earn an affiliate commission for buying something through one of these links. Learn more.
