Scalable Simulation, Scientific Visualization and Professional Rendering with Atipa Polaris Select HPCV Clusters

Massive data sets generated in scientific computing, big data analytics, and machine learning are overwhelmingly complex and difficult to comprehend without visual analysis. Visualizing data sets enhances the scientist’s understanding of complex ideas while also facilitating peer-to-peer communication and public outreach of scientific results – sharing raw data tables is not as effective as colorful, photorealistic visualizations that appeal to business stakeholders as well as scientists and engineers.


Traditional simulation and data analysis workflows are based on an ad-hoc model where scientific simulations are performed on a high-performance compute cluster and the results are subsequently transferred to disk for eventual visual analysis on a dedicated GPU-based workstation or cluster. For truly big data, this data movement is inefficient, time consuming, and, at scale, simply impossible.


In the past year, Atipa Technologies has partnered with Intel® to deploy the Atipa Polaris Select High-Performance Computing and Visualization (HPCV) platform – the industry’s first verified Intel® Select Solution for Professional Visualization. The platform employs the software-defined visualization (or SDVis) approach through the use of applications utilizing the open source Intel® Rendering Framework, the most promising and practical way to manage the flood of data, complex simulation codes, and the convergence of machine learning and simulation workloads.  SDVis leverages the breadth of available graphics capabilities able to use both OpenGL rasterization and more recent and higher fidelity ray-tracing algorithms. In either graphics technology, a much larger memory pool is available to CPUs than to GPUs, and Intel Rendering Framework is optimized for Intel® Xeon® Scalable processors to visualize increasingly large data sets from HPC simulations, providing unprecedented detail.


High-fidelity ray-tracing’s embarrassingly parallel reputation makes it a natural candidate for visualizing large data sets on algorithm distributed memory clusters, especially for machines without specialized graphics hardware. This example shows a comparison between an OpenGL-based rasterization (top) and a ray-traced visualization using the Intel® Rendering Framework (bottom) of the same data set of a 3D flow simulation of groundwater. Data courtesy of Florida International University and Texas Advanced Computing Center.

Today, April 2nd of 2019, we are announcing the much-anticipated platform update using 2nd Generation Intel® Xeon® Scalable processors. The benefits to the Atipa Polaris Select HPCV platform for holistic scientific computing workflows that include robust, high fidelity visualization include:

  • Higher clock frequencies

  • Increased memory bandwidth with DDR4 2933MHz ECC RDIMM

  • Increased memory capacity (1TB base, up to 4.5TB on memory-optimized CPUs)

  • Intel® Deep Learning Boost with Vector Neural Network Instructions (VNNI)

Enabling visualization on the 2nd Generation Intel® Xeon® Scalable processors already capable of the largest number of simulation codes available also opens the door to in-situ simulation and visualization, where data is visualized with the same memory used by the simulation, bypassing I/O to and from a storage subsystem. This permits instantaneous insight and discovery into large scale simulations and including–interacting with and studying the results “live” before the simulation comes to an end.

Key Benefits of Atipa Polaris HPCV Clusters
  • Scalability: MPI-enabled simulation and visualization scales across an entire cluster of nodes

  • Fast performance: Generate and Visualize large data sets up to the size of the total aggregated cluster memory

  • Better quality: Higher fidelity up to photorealism with ray-tracing

  • Improved efficiency: Turn every cluster node into a dual-purpose compute and visual analysis node

  • Cost effective: No more need for specialized visualization hardware and infrastructure

  • Flexibility: Intel Rendering Framework optimized on the Intel® Xeon® Scalable processor family allows for straightforward integration into existing and new visualization applications via intuitive API’s, it’s open source implementation, and support from Intel and it’s SDVis community partners.

Intel Rendering Framework.png

Atipa Polaris Select HPCV clusters are composed of several key hardware and software components, optimized for high-performance computation and visualization. As a verified Intel® Select Solutions for Professional Visualization, customers get the benefit of the latest Intel® technologies and performance guarantees without the need to perform lengthy research to design the right system for simulation and visualization. The following Table summarizes the minimum specifications of Atipa Polaris Select HPCV clusters powered by 2nd Generation Intel® Xeon® Scalable processors:

Atipa Altezza SX226XXL-24
Atipa Altezza QX226XXL-16

The Atipa Altezza QX226XXL-16 offers four independent HPCV nodes in a compact 2U form factor. The QX226XXL-16 is easily maintainable with advanced RAS features including hot-swappable nodes, hard disks, and power supplies; out-of-band remote management with IPMI 2.0; command-line BIOS configuration utilities; and premium quick release rail kits.

Altezza QX225IS-16 Front + Node.png

The Intel® Rendering Framework for software-defined (i.e. CPU-based) visualization is already supported by more than 80 visualization applications such as ParaView, VisIT, VMD, Autodesk and many more. Using 2nd Generation Intel® Xeon® Scalable processors for visualization avoids costly data transfers between simulation and visualization systems as well as the acquisition and maintenance costs of disparate systems. Ultimately, the Atipa Polaris Select HPCV platform saves time and money, while helping scientists and engineers gain better and faster insights while solving their toughest problems.  


The updated Polaris Select HPCV platform is available for pre-order now and will start shipping in May 2019. Starting configurations ship with 1 head node and 4 HPCV nodes with dual Intel® Xeon® Gold 6248 processors, 192GB DDR4-2933MHz memory, and 100Gbps Intel® Omni-Path fabric interface.  To learn more contact for 1-on-1 consultations and evaluation requests.