Dell Technologies has released a solution update that supports generative AI workloads.

Dell Technologies announced on the 12th that it has added new features to its Dell PowerScale storage to provide faster AI implementation and generative AI performance. It also added that it has released software updates and supports verification with NVIDIA DGX SuperPOD AI infrastructure.

Dell PowerScale, a scale-out file storage system, has updated its software and added new features to meet the increasing performance demands driven by the adoption of AI.

The updated PowerScale OneFS software enables faster preparation, training, fine-tuning, and inference of AI models. The new version of PowerScale All-Flash Storage, based on the latest generation of Dell PowerEdge servers, delivers up to 2x improved streaming read and write performance.

PowerScale has also added a new smart scale-out feature. It improves single computing node performance and increases GPU utilization to accelerate storage throughput for AI training, checkpointing, and inference.

Through collaboration with NVIDIA, Dell supports faster and more efficient implementation of AI storage by providing verified NVIDIA DGX systems, Dell PowerScale storage, NVIDIA Quantum-2 InfiniBand, and Spectrum Ethernet networking configurations.

Dell's solution is set to be the first Ethernet storage solution validated for the NVIDIA DGX SuperPOD.

As Dell PowerScale has completed validation for the NVIDIA DGX SuperPOD, a turnkey AI infrastructure solution, companies utilizing it can accelerate the development of AI and generative AI. The NVIDIA DGX SuperPOD includes the NVIDIA AI Enterprise Software Platform to provide full-stack AI supercomputing solutions.

Dell APEX File Storage for Microsoft Azure, a recent addition to the Dell APEX Storage portfolio for the public cloud, enables enterprise-grade file storage performance and management capabilities in the Azure environment. This allows you to easily meet the requirements of performance-intensive AI and machine learning applications such as Azure OpenAI Service and Azure AI Vision.

By adopting Dell Apex File Storage for AWS and Azure environments, you can leverage cloud-native AI and generative AI workflows for public cloud or on-premises data and improve data access and movement.

As Dell expands its collaboration with Databricks, customers can flexibly perform AI and analytics tasks across multi-cloud environments, such as selecting the necessary models from a variety of Large Language Models (LLMs) and retraining base models on their own data in Dell Apex File Storage using the 'Databricks MosaicML' library.

In addition, through the recently announced open 'Modern Data Lakehouse,' Dell integrates Starburst analytics software with Dell PowerEdge servers and file and object storage systems such as Dell PowerScale, Dell ECS, and Dell ObjectScale to enable the extraction of insights from data regardless of its location.

“Storage performance is a factor that has a massive impact on AI outcomes. There is a steady increase in customers seeking Dell storage to avoid data bottlenecks that act as obstacles to the throughput and scalability of compute-intensive applications.” "We plan to continuously pioneer the market so that more customers can successfully discover the potential of generative AI through fast, efficient, and secure data access," he said.

The Dell PowerScale oneFS software update will be available globally this month, while the all-flash storage Dell PowerScale System and new Smart Scale-out features are scheduled for global release in the first half of 2024. Dell PowerScale Storage is expected to complete validation for NVIDIA DGX Super PODs within the first quarter of 2024, and Dell Apex File Storage for Microsoft Azure is scheduled for release in the first half of 2024. Dell Apex File Storage integrated with Databricks and MosaicML is currently available on AWS and will be available on Microsoft Azure in the first quarter of 2024. Dell's open modern data lakehouse solution is scheduled for global release in the first half of 2024.