Accelerating the Opportunities of Generative AI with Lenovo AI Workstations and NVIDIA Accelerated Computing

It’s great to be back at NVIDIA® GTC! We’re proud and grateful for our extended partnership with NVIDIA, and we continue to be energized by the possibilities we can jointly provide industry professional and data scientists who are exploring the immense opportunities provided by GPU accelerated LenovoThinkStation desktop and ThinkPad mobile workstation computing, AI and Generative AI.  

The evolving workflows of data science and Gen AI, especially with large language models (LLMs), require robust NVIDIA GPUs like the NVIDIA A800 40GB Active, designed specifically for AI workflows, or the NVIDIA RTX™ 6000 Ada Generation for unprecedented multi-purpose performance. Available in Lenovo ThinkStation and ThinkPad systems, these GPUs enable secure, personal supercomputing environments for organizations working with all kinds of AI Workflows. For example, the ThinkPad P16 Gen 2 featuring NVIDIA RTX 5000 Ada Generation Laptop GPU exemplifies cutting-edge technology suited for demanding professional workloads. These powerful laptops are engineered to enhance mobile AI development and offer LLM fine-tuning with unparalleled performance for tasks requiring substantial computational power.

Built in collaboration with NVIDIA, here are a few innovations we are excited to share at NVIDIA GTC.

Lenovo Workstations for Generative AI

From the ThinkPad P Series Mobile Workstations, which are perfect for a data scientist on–the-go, to the ThinkStation PX, which is the world’s fastest, most powerful and most advanced workstation1 for AI workflows, the Lenovo Workstations portfolio is the largest collection of AI-Ready2 products on the market and provides highly versatile solutions for today’s AI workloads. Furthermore, Lenovo’s commitment to open-source operating system development is unwavering, investing resources annually into Linux development and certifications. Its dedication is evident in having the industry’s broadest range of Linux-certified workstations3, supported by comprehensive pre- and post-sales Linux services. Lenovo’s proactive approach not only highlights its technical expertise but also underscores its commitment to meeting the needs of a diverse clientele seeking reliable and customizable solutions.

To help AI developers increase productivity through automated workflows, Lenovo is proud to announce Lenovo Workstations with NVIDIA AI Enterprise and NVIDIA AI Workbench – Built on the NVIDIA CUDA® compute platform and leveraging CUDA-X libraries to speed up data processing for a variety of data types. This includes the NVIDIA RAPIDS™ cuDF library, which by using an NVIDIA RTX™ 6000 Ada Generation GPU can accelerate the work of nearly 10 million data scientists using pandas software by up to 110x4 without requiring any code changes.

RAPIDS cuDF, NVIDIA AI Workbench and NVIDIA AI Enterprise will be available as part of the Lenovo AI workstation solution portfolio; delivering an enterprise class full stack AI developer software platform that is built to enhance the productivity and efficiency of data science, generative AI & LLM training and fine-tuning workflows, no matter where or when they are executed.

With NVIDIA AI Enterprise, Lenovo workstations now also support NVIDIA NIM and NeMo Retriever microservices for inference and retrieval-augmented generation (RAG). This enables developers to work on Generative AI applications on their powerful local systems.

ThinkStation PX for Data Scientists

Today, many organizations are moving to privately hosted and developed LLMs and small-language models (SLMs), especially given security concerns around leveraging proprietary data in these models and the rising costs associated with training data and LLMs. Many companies are finding that, to get moving with Generative AI, it is easier and more beneficial to leverage existing models and augment with one’s own data in a more secure sandbox environment. By leveraging Lenovo Workstations with NVIDIA AI Enterprise, AI practitioners and data scientists are provided with proactive LLM and SLM recommendations based on preferred use cases, such as building chatbots, voice to text, image manipulation and other emerging applications.

Lenovo’s dedication to enhancing Open-source technologies shines through its substantial investments that affirm Lenovo’s unwavering commitment to rigorous certification and qualification processes across a broad range of NVIDIA Certified Platforms, spanning workstations, servers, and edge devices in their extensive portfolio. By actively expanding support for Linux and Open-source initiatives, Lenovo demonstrates a forward-thinking approach in driving technological progress and meeting the evolving needs of its customers.

The proactive curation of high-performance configurations tailored for diverse requirements in Generative AI underscores Lenovo’s innovation focus. Through strategic collaborations with NVIDIA AI Enterprise and NVIDIA AI Workbench on their renowned ThinkStation desktop and ThinkPad P series mobile workstation lineup, Lenovo excels in empowering groundbreaking outcomes within the realm of AI. These meticulously designed configurations elevate performance and highlight Lenovo’s commitment to anticipating and fulfilling the demands of users engaged in many Generative AI applications and complex datasets. The collaboration with NVIDIA highlights Lenovo’s leadership in delivering top-tier solutions for cutting-edge AI projects, cementing its position as an industry trailblazer.

Turnkey Digital Twin Solutions: Start Your Digital Twin Journey Now

Digital twins are performance-hungry and existing hardware structures are not built to support the needed performance requirements of these applications. To further help customers fully realize the potential of digital twin applications, Lenovo partnered with NVIDIA to create turnkey solutions for customers to start their NVIDIA Omniverse Enterprise and digital twin journey.

By combining the power, performance and unique capabilities of the Lenovo ThinkStation PX with up to 4x NVIDIA RTX 6000 Ada Generation GPUs plus NVIDIA ConnextX-6 high-performance networking, customers can easily scale their digital transformation journey with NVIDIA Omniverse Enterprise by reducing the complexities, costs and business concerns around digitizing workflows. This turnkey digital twin platform is poised to make quick and impactful business outcomes to customers across a wide number of industry verticals, allowing customers to start small and scale quickly and easily to larger and more complex digital twin projects on a single workstation. The Lenovo ThinkStation Digital Twin solution can start at your desk and being fully rack mountable, scale all the way to the datacenter, with NVIDIA OVX, NVIDIA’s deployment model for large scale digital twins.

Realizing End-to-End Spatial Computing

The digitization of certain industries is happening at a very fast pace today and companies are looking for ways to digitize their businesses. However, organizations across a broad number of verticals are struggling to adopt real-world use cases of digital twins with spatial computing due to the complex layers of computer hardware and software that are required from multiple different vendors.

Spatial computing can be defined as the digitization of activities involving machines, people, objects and their environments through immersive technologies like AR, VR, MR and AI & ML to deliver a fully interactive and collaborative environment and platform to view digital twins. Spatial computing can be very CPU-, GPU- and system resource intensive, demanding a large amount of physical CPU cores, GPUs, memory and network bandwidth. While gluing these requirements together has proved difficult, Lenovo’s enterprise class workstation and virtual reality portfolio provide the capabilities needed for a high-performance solution. The ThinkStation PX, ThinkReality VRX HMD and ThinkReality MDM software are the perfect recipe for spatial computing in the enterprise.

Configured together with leading software like NVIDIA Omniverse Enterprise (OVE) and NVIDIA CloudXR, Lenovo and NVIDIA can deliver a fully validated reference architecture for spatial computing, something others in the industry cannot do today[1]. The architecture is composed of:

  • ThinkStation PX Workstation
  • ThinkReality VRX HMD
  • ThinkReality MDM Software
  • NVIDIA Cloud XR
  • NVIDIA Omniverse Enterprise

This end-to-end solution, the Lenovo Spatial Computing Appliance, can truly allow organizations to realize the possibilities of digital twin technology. The standalone workstation appliance can be rack-mounted in an organization’s office, and by using professional software such as Autodesk VRED or NVIDIA Omniverse Enterprise, multi-user collaborative VR experiences can be easily pixel-streamed from a single ThinkStation PX workstation to four separate ThinkReality VRX headsets simultaneously – Enabling an easy to manage, easy to host platform for collaborative design review across many different industries including aerospace, automotive, architectural, engineering, construction, manufacturing, and product design . By having one centralized on-prem resource for immersive experiences reduces latency and delivers a higher quality, more reproducible experience for the end customer, and with VR experiences streamed to ThinkReality VRX headsets using NVIDIA CloudXR users do not have to be tethered to a physical workstation.

The appliance will support four ThinkReality VRX headsets, with each headset having full access to a dedicated NVIDIA RTX6000 Ada GPU for the highest performing professional VR experience, regardless of project dataset size or complexity.

 See You in San Jose!

If you are attending GTC, running through March 21 in San Jose, Calif., we look forward to seeing you there! We will be demoing the Spatial Computing Appliance and have four ThinkReality VRX headsets ready for attendees to experience a high-performance digital twin reality, this will take place in the XR Zone Booth #1740. At the Lenovo Booth #840, visitors can experience NVIDIA AI Enterprise, AI Workbench cuDF and NVIDIA Omniverse/Digital Twin solutions running on Lenovo workstations. Visitors will also discover new products, technologies, and initiatives shaping the future of AI from Lenovo’s Infrastructure Solutions Group.

For full details of Lenovo speaking sessions at NVIDIA GTC visit here. For more information on how Lenovo Workstations are working toward a smarter, faster future, visit https://techtoday.lenovo.com/us/en/workstations.

 

1 Based on March 2024 internal analysis around a combination of simultaneous CPU & GPU workloads.
2 Based on March 2024 internal analysis of top 5 global vendors on the number of professional ISV certified workstations that have both NVCerts & NVQual for NVIDIA AI Enterprise
3 Based on March 2024 internal analysis of the top 5 global vendors on the number of professional Linux certified workstations available
4 Instead of CPU only workflows

Source link

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.