Our Blog

BY : Sigin Jacob George 0 comment

NVIDIA simplifies GPU-accelerated computing with Red Hat OpenShift

NVIDIA is known for inventing the graphical processing unit (GPU), and is now helping enterprise customers adopt GPU-accelerated computing for artificial intelligence (AI) and high-performance computing (HPC) applications. Through its partnership with Red Hat, NVIDIA helps customers run GPU-accelerated computing on Red Hat OpenShift. A GPU operator, developed initially by Red Hat and now owned by NVIDIA, simplifies GPU-accelerated computing on an enterprise level container platform.

Facilitating processing-intensive computing in the enterprise

NVIDIA invented the GPU, a specialized processor that can process many pieces of data simultaneously, and helps enterprise customers adopt the processor for running processing-intensive HPC, AI, and cloud operations. GPU-accelerated computing—where the compute-intensive portion of a workload runs on GPUs—is reshaping transportation, healthcare, manufacturing, and many other industries. NVIDIA developed Compute Unified Device Architecture (CUDA)—a parallel computing platform and programming model for general computing on GPUs—to simplify the development of GPUaccelerated applications. The framework includes libraries, a toolkit, runtime, and plugins that communicate with the GPU. 

Customers who initially wanted to take advantage of running Kubernetes on top of GPUs had to manually write containers for CUDA and all the software needed to run GPU-accelerated applications on Kubernetes. Developers also had to write additional code to tell Kubernetes which nodes contained GPUs. The process was time-consuming and prone to errors, but is now greatly simplified using Red Hat OpenShift.

Partnering for an optimal solution 

While NVIDIA caters to all Kubernetes distributions, Red Hat OpenShift is seen as a priority. “Red Hat OpenShift is very important to NVIDIA as it allows our customers to develop, deploy, and deliver new apps faster and easier,” said Akins. “When we adapted CUDA for Kubernetes, Red Hat OpenShift was top of mind.” NVIDIA produced a series of Red Hat OpenShift techniques for CUDA and the software needed by GPU-accelerated applications with guidance from Red Hat. “Working with the Red Hat team, we wrapped all these different plugins up into a single operator, to provide a better way to communicate with the container platform,” said Akins.

Making AI and HPC accessible 

When a customer deploys Red Hat OpenShift on top of a server with GPUs, the GPU operator automatically containerizes CUDA and all the software needed before deploying to Red Hat OpenShift. More than 100 customers are currently using the GPU operator to help them implement and run GPU-accelerated workloads across a wide range of application types, including AI, machine learning, model training, and inferencing. Red Hat provided NVIDIA with the initial code for the GPU operator, and NVIDIA now maintains it. “The Red Hat engineering team played an integral role in helping us develop the GPU operator,” said Akins.

Building on each others’ strengths 

Offered crucial insight for ongoing development of the operator

NVIDIA uses Red Hat’s expertise and influence regarding Kubernetes, helping NVIDIA understand the container platform’s future direction so they can build critical evolutionary advancements into the GPU operator. 

Ensured optimal compute efficiency for customers’ AI and HPC workloads 

The GPU operator allows NVIDIA to optimize compute efficiency for its customers. A process orchestrated by Red Hat OpenShift uses node-labeling techniques, so workloads can automatically find the specific type of GPU they need. 

Saved customers time and avoided manual errors because of automation

The GPU operator makes it easier for customers to use CUDA to take advantage of GPU technology for running HPC and AI workloads on Red Hat OpenShift. Automation saves customers time and helps them avoid errors. 

Provided customers with support and expertise from the right partner 

With the NVIDIA and Red Hat teams aligned, any customer facing an issue with the GPU operator can submit a ticket to Red Hat. The partners then triage the ticket together and have an escalation path before escalating it to either NVIDIA or Red Hat experts, ensuring customers have access to the best support. 

About NVIDIA 

NVIDIA’s invention of the GPU sparked the PC gaming market. The company’s pioneering work in accelerated computing—a supercharged form of computing at the intersection of computer graphics, high performance computing and AI—is reshaping trillion-dollar industries, such as transportation, healthcare, manufacturing, and fueling the growth of many others.

If you would like to learn more on Red Hat OpenShift Courses and Certifications please visit : 

List of Authors

Global Relations Manager at ipsr solutions limited, Technological Specialist with 20+ years experience in IT Training, Sales and Management. Technical Areas of Interest - Open Source, Cloud, DevOps

Leave a Reply

Your email address will not be published.