We are in the midst of a global transformation and it is touching every aspect of our world, our lives and our businesses. Many in the industry believe artificial intelligence (AI) is the key to fundamentally changing how organizations will derive insights from data. We know that AI will be a game-changer for our clients, but we also know that there is no “magic bullet” when it comes to AI. For most of our clients, AI will be a journey. This is demonstrated by the fact that most organizations are still in the early phases of AI adoption. Consider these AI adoption facts from IDC[i]:

• 31 percent of organizations are in discovery/evaluation
• 22 percent of organizations plan to implement AI in next 1-2 years
• 22 percent of organizations are running AI trials
• 4 percent of organizations have already deployed AI

Our experience with clients shows that as an organization embraces a holistic AI strategy to drive insights from their data, it is critical to look at their infrastructure as well as their data management strategy and AI capabilities. Organizations that are deriving the most value from data are building their data management and AI platforms close to where the data resides, thereby reducing latency. Additionally, they are using infrastructure specifically designed for data and compute-intensive workloads like advanced analytics and AI, as well as deploying software that is optimized to exploit it. Together this maximizes efficiency of deployment and the value of insights.

When it comes to implementing AI, organizations often struggle with server performance bottlenecks and open source software complexity. To address these challenges, today we are announcing software innovations that help make AI and deep learning easier along with new hardware platforms designed for the AI era.

Here’s a quick checklist to see how ready you are for the journey to enterprise AI:

1. Are you testing powerful AI frameworks that are optimized for the hardware upon which they’re running?
2. Is your hardware optimized as an AI-ready enterprise data platform?
3. Have you deployed modern data platforms like Hadoop and Spark that can help you organize unstructured data?

To make this optimized platform a reality, today IBM PowerAI, the best deep learning toolkit for the enterprise, is available on POWER9 and for the first time is also available on the Red Hat operating system. Now, Red Hat enterprises can leverage the industry’s leading open source frameworks and explore the potential of distributed deep learning with large model support. To support clients on this journey, IBM provides comprehensive enterprise support for deep learning frameworks. In addition, new updates to PowerAI’s AI Vision tool allow easy training of computer vision networks in an intuitive GUI, so enterprises can leverage the developer skills they have to get training right away with AI-enabled computer vision.

“IT optimization drives the need for more complete, enterprise-ready platforms that take the most popular open source innovations and back them with the reliability and support expected by modern enterprises,” said Tim Burke, Engineering Vice President, Cloud and Operating System Infrastructure, Red Hat. “POWER9-based servers, running Red Hat’s leading open technologies offer a more stable and performance optimized foundation for machine learning and AI frameworks, which is required for production deployments. Today, we are excited to extend our two decades-plus collaboration with IBM to include PowerAI, that has popular frameworks like Tensorflow and Caffe, as the first commercially supported AI software offering for our platform.”

To create an AI-optimized infrastructure, today we are also announcing the next in our line of IBM POWER9 servers: the IBM Power Systems LC922 and LC921. These new, balanced servers bring superior compute capabilities and up to 120 terabytes of data storage with hybrid storage options including HDD, SSD and NVMe for rapid access to vast amounts of valuable data.

These new servers join an updated version of our previously released AC922 server, which now features recently announced 32GB NVIDIA V100 GPUs and larger system memory which enables larger deep learning models to help improve the accuracy of AI workloads.

In this IT revolution, we are collaborating with industry leaders to bring innovation across the software and hardware stack that is specifically designed to advance enterprise AI capabilities.

”GPUs are at the foundation of major advances in AI and deep learning around the world,” said Paresh Kharya, group product marketing manager of Accelerated Computing at NVIDIA. “Through the tight integration of IBM POWER9 processors and NVIDIA V100 GPUs made possible by NVIDIA NVLink, enterprises can experience incredible increases in performance for compute- intensive workloads.”

Underpinning both servers is the IBM POWER9 CPU, which includes on-chip support for next-generation NVIDIA NVLink interconnect technology and PCI Express (PCIe) 4.0 for greater performance and expanded throughput for data transfers. These technologies give IBM POWER9 nearly 5.6x improved CPU to GPU bandwidth vs compared x86[ii], that can improve deep learning training times up to nearly 4x[iii][iv].

“Enterprises need a flexible, efficient and unified platform to put their data to work effectively,” says Scott Gnau, CTO of Hortonworks. “The winning platform includes a modern data architecture, including core Hadoop and streaming software, as well as hardware designed for data-intensive workloads which will ultimately drive real returns on AI. IBM is the one vendor that provides a unified infrastructure approach that can take AI to the next level.”

We know the journey to AI isn’t easy, but IBM is committed to collaborating with you every step of the way. To take the first steps on your enterprise AI journey, register for our upcoming webinar and learn more about our IBM Power Systems enterprise AI capabilities.

This story first appeared on the In The Making Blog

[i] IDC, When Computing Becomes Human: Automation, Innovation, and the Rise of the All-Powerful Service Provider, doc #DR2018_GS4_MB, February 2018

[ii] Results are based on IBM Internal Measurements running the CUDA H2D Bandwidth Test
Hardware: Power AC922; 32 cores (2 x 16c chips), POWER9 with NVLink 2.0; 2.25 GHz, 1024 GB memory, 4xTesla V100 GPU; Ubuntu 16.04. S822LC for HPC; 20 cores (2 x 10c chips), POWER8 with NVLink; 2.86 GHz, 512 GB memory, Tesla P100 GPU
Competitive HW: 2x Xeon E5-2640 v4; 20 cores (2 x 10c chips) / 40 threads; Intel Xeon E5-2640 v4; 2.4 GHz; 1024 GB memory, 4xTesla V100 GPU, Ubuntu 16.04

[iii] Results of 3.7X are based IBM Internal Measurements running 1000 iterations of Enlarged GoogleNet model (mini-batch size=5)  on Enlarged Imagenet Dataset (2560×2560). Hardware: Power AC922; 40 cores (2 x 20c chips), POWER9 with NVLink 2.0; 2.25 GHz, 1024 GB memory, 4xTesla V100 GPU; Red Hat Enterprise Linux 7.4 for Power Little Endian (POWER9) with CUDA 9.1/ CUDNN 7;. Competitive stack: 2x Xeon E5-2640 v4; 20 cores (2 x 10c chips) /  40 threads; Intel Xeon E5-2640 v4;  2.4 GHz; 1024 GB memory, 4xTesla V100 GPU, Ubuntu 16.04. with CUDA .9.0/ CUDNN 7 Software: Chainverv3 /LMS/Out of Core with patches found at https://github.com/cupy/cupy/pull/694 and https://github.com/chainer/chainer/pull/3762

[iv] Results of 3.8X are based IBM Internal Measurements running 1000 iterations of Enlarged GoogleNet model (mini-batch size=5) on Enlarged Imagenet Dataset (2240×2240). Power AC922; 40 cores (2 x 20c chips), POWER9 with NVLink 2.0; 2.25 GHz, 1024 GB memory, 4xTesla V100 GPU ; Red Hat Enterprise Linux 7.4 for Power Little Endian (POWER9) with CUDA 9.1/ CUDNN 7;. Competitive stack: 2x Xeon E5-2640 v4; 20 cores (2 x 10c chips) /  40 threads; Intel Xeon E5-2640 v4;  2.4 GHz; 1024 GB memory, 4xTesla V100 GPU, Ubuntu 16.04. with CUDA .9.0/ CUDNN 7.  Software: IBM Caffe with LMS Source code https://github.com/ibmsoe/caffe/tree/master-lms