草榴社区

Optimizer

Autonomous Optimization Platform for Optimizing Applications to Silicon and Everything in Between

As a key component of the 草榴社区 integrated Silicon Lifecycle Management (SLM) family, the Optimizer Runtime and Optimizer Studio performance optimization software improve compute system performance, automatically and in real-time. The tools enable AI-powered dynamic and static optimization using autonomous software agents to continuously monitor the interactions between operating applications and the underlying system environment. This level of analytics and optimization is being driven by the growing demand for compute resources at the edge for applications such as cloud based data centers, automotive and Industrial IOT.

scatter chart with dots in multiple colors

The combination of this software-related dynamic optimization solution with the hardware-related optimization achieved with sensor-based analytics in the SLM family delivers a powerful, unique, and comprehensive system optimization solution covering reliability, resiliency and security.

EEJournal-ChalkTalk-YouTube-2023.jpg

Automated Benchmark Tuning

Nozar Nozarian explains the importance of benchmarking to measure the performance of the system and to understand and quantify how well it will perform under certain operating conditions.

Optimizer Runtime – Autonomous real time performance tuning

Optimizer Runtime is an AI-powered software agent that provides autonomous real time performance optimization, used for automatically optimizing system settings according to the currently running workload.

Key Features

  • Dynamically tunes system settings
  • Accelerates applications with no code changes
  • Applies to any stack using plugins
  • Fully autonomous

Optimizer Studio - Collaborative full-stack optimization platform

Static optimization tool for production and staging environments, used for optimizing a wide variety of goals by discovering the best-performing settings across the full stack.

Key Features

  • Built with native support for use cases such as full-stack application optimization, benchmarking, cloud cost optimization, and design space exploration
  • Fast search in the options space, using machine learning techniques such as reinforcement learning as well as evolution algorithms
  • Opinionated experimentation workflow methodology, which can be adopted fully or partially depending on the application
  • Collaborative experiment management system (Conductor) with full documentation and reporting capabilities
  • Advanced algorithms for parameter search, sensitivity analysis, invalid configuration exploration, and configuration refinement
  • Continuous Optimization (CO) via an API for integration with CICD and APM tools
  • Out-of-the-box plugins with 1000’s of prebuilt full-stack knob definitions such as compiler flags, operating systems, CPUs, databases and runtimes.