Avesha Resources / Blogs
Veena Jayaram
We are excited to announce the release of Elastic GPU Service (EGS) version 1.12.0, which is live as of April, 10th, 2025. This latest version introduces several significant features and enhancements aimed at optimizing GPU utilization and streamlining workflows for AI projects.
The Elastic GPU Service (EGS ) platform is an innovative solution designed to optimize GPU utilization and efficiency for your AI projects. EGS leverages the power of Kubernetes to deliver optimized GPU resource management, GPU provisioning, and GPU fault identification.
Administrators can now create GPU Provisioning Request (GPR) templates directly within the Admin Portal and assign them to specific workspaces. This feature simplifies the GPR creation process for users by providing predefined configurations tailored to their needs. For detailed instructions on managing GPR templates, refer to Manage GPR Templates.
The new Auto GPR functionality allows administrators to enable automatic GPR creation for workspaces with a configured default GPR template. This automation ensures that GPU resources are provisioned seamlessly without manual intervention. Note that manual GPR creation is disabled for workspaces where Auto GPR is active. Learn more about enabling Auto GPR in Enable Auto GPR.
Version 1.12.0 introduces the API Key REST API and its corresponding Python SDK, providing functionalities equivalent to the API tokens managed through the Admin and User Portals. These additions facilitate programmatic interactions with EGS, enhancing automation capabilities. For more information, see API Key REST API and API Key Python SDK.
The Admin Dashboard has been updated with redesigned tiles in both the Overview and Cost Analysis tabs, offering a more intuitive and informative interface. These changes aim to provide administrators with clearer insights into system performance and cost metrics. Details can be found in the Admin Dashboard section.
A new dashboard has been introduced in the User Portal, featuring Overview and Cost Analysis sections specific to workspaces to which the user has access. This enhancement empowers users with detailed information about their resource utilization and associated costs. For a comprehensive overview, visit the User Dashboard page.
The EGS installation script now includes configurations for labeling and annotating user-defined namespaces, facilitating better organization and management of resources. For those opting for manual installation, these configurations can be set within the EGS Controller Helm chart. Detailed instructions are available in Install the EGS Controller.
We encourage all users to explore these new features and enhancements to fully leverage the capabilities of EGS 1.12.0. For a complete list of changes and detailed documentation,refer to the EGS product documentation and the official release notes.
Announcing Smart Scaler 2.13.0: Enhanced Kubernetes Scaling and Management
Transforming Cloud-Native Ecosystems: Unified Observability, AI, and Automation Take Center Stage
Avesha’s NVIDIA GTC 2025 Trip Report
Scaling AI Workloads Smarter: How Avesha's Smart Scaler Delivers Up to 3x Performance Gains over Traditional HPA
Copied