Rescale’s platform intelligence helps organizations easily find and adopt the most efficient computing resources for powering advanced engineering simulation, machine learning, and generative AI applications
Rescale, the leader in high performance computing (HPC) cloud management, announces the availability of Performance Profiles, a revolutionary capability that offers simple and automated benchmarking across any computing architecture or infrastructure configuration available in the cloud.
With Performance Profiles, organizations can quickly and easily capture significant cost savings and performance improvements for computational engineering, machine learning, and generative AI workloads.
Engineering and R&D-driven organizations must spend significant portions of their budgets on powering computational-intensive research efforts. Identifying the optimal hardware and software configurations can save businesses and institutions thousands or even millions of dollars while dramatically reducing run times, helping speed the pace of innovation.
Though manual benchmarking has been a common practice, the explosive growth of specialized microprocessor chips and high performance computing (HPC) infrastructure services from cloud providers has made it extremely difficult to effectively identify the best configurations for different kinds of intensive computational workloads.
Rescale Performance Profiles automates benchmarking of workloads and datasets to help organizations understand the performance, cost, energy consumption, and carbon footprint of their computing tasks on any infrastructure available from Rescale’s cloud provider network, which includes traditional CPUs, GPUs, and HPC specialized architectures.
“Our initial use of Rescale Performance Profiles has been eye-opening,” says Brian Jackson, lead fluid dynamics engineer for Kairos Power, a cutting-edge clean energy start-up. “With Performance Profiles’ benchmarking capabilities, we quickly identified how we can save as much as 30 percent on our computing costs by switching from our current HPC architecture to other types of specialized chips. That knowledge is helping us be much smarter about our HPC cloud spending.”
Picking the Right HPC Architecture, Made Easy
Rescale Performance Profiles work in conjunction with the platform’s AI-driven Compute Recommendation Engine, which suggests the best computing architectures for benchmarking based on the user’s application. A user can then benchmark applications across multiple types of microprocessors and core configurations using their own simulation and models.
With Performance Profiles, organizations can comprehensively assess the best architecture options for their needs through an intuitive console that’s easy to operate, even for non-technical users. Promising configurations can be saved to job templates used by engineers and researchers to deploy the best infrastructure configuration, every time.
“With Rescale Performance Profiles, organizations now have near real-time awareness of the costs and performance of any cloud computing architecture,” said Edward Hsu, Rescale’s chief product officer. “Performance management becomes a simple, seamless, and continuous process, rather than a time-consuming, painstaking, manual effort carried out every few years. With such information at their fingertips, organizations can easily manage costs while still driving innovation.”
Performance Profiles provide yet another way organizations can harness the intelligence of the Rescale Platform to more effectively manage their computing investments through comprehensive, centralized control and complete visibility into how their teams are using these critical resources.
With just a couple of clicks in the Rescale console, customers can compare as many hardware infrastructures as necessary to determine the best computing architecture for any of their digital design, research, or engineering projects.
With the cloud’s pay-per-use model, the insights provided by Performance Profiles are critical for helping organizations ensure they are using the highest performing and least expensive option for driving key innovations initiatives in computational R&D, as well as ensuring the best computational power for their generative AI initiatives.
Visit AITechPark for cutting-edge Tech Trends around AI, ML, Cybersecurity, along with AITech News, and timely updates from industry professionals!