Dell Project Helix :Democratizing AI
We know today that GPT applications that taken off and it’s caused an inflection point of IT DMs. A Gartner study polled their customers, almost half of them had already […]
Dell Storage, PowerStore, PowerFlex PowerMax & PowerScale, Virtualization & Containers Technologies
We know today that GPT applications that taken off and it’s caused an inflection point of IT DMs. A Gartner study polled their customers, almost half of them had already […]
We know today that GPT applications that taken off and it’s caused an inflection point of IT DMs. A Gartner study polled their customers, almost half of them had already increased their AI investment just because of the fact on the potential of chat GPT and generative AI applications. We know that large enterprises are already using AI improve their own efficiency and quality from their results. And three years from now, Gartner expects over sixty percent of applications like phone apps, web apps or Windows apps to be generated automatically from AI coding, programming generating functions. That’s only three years from now which means companies need to jump on the bandwagon today because AI and Generative AI potential is changing things for the enterprise. Furthermore, a quote from a Bloomberg call where the CEO of ArK Invest Funds was mentioning that the killer app for any kind of AI is really automating and delivering productivity gains that businesses are looking for. She said that the real winners are those that really have these huge masses of proprietary data kind of like the Bloomberg GPT application (https://www.bloomberg.com/company/press/bloomberggpt-50-billion-parameter-llm-tuned-finance/) and high quality data that can turn be turned into value quickly. Adopting this approach is going to put a lot of these companies who are already developing and deploying AI further into a leadership position. And AI leaders have already spent the time and effort to leverage AI, build AI teams with skillsets and more quickly plan for converting their data into value, into competitive advantage. The challenge for many companies on-the-fence with regards to AI is to jump on the AI bandwagon to increase their business transformation and boost productivity while minimizing the learning curve, trial-and-error adoption of AI and now other forms of AI including Generative AI to farm value from their data. Not doing so risks remaining behind AI leaders.
enabling customers to accelerate their business outcomes so they can move quickly past GenAI deployment and start extracting value from their data
Enabling them to drive and streamline their AI journey, from defining the data that needs to be modeled, finding out the models,, and streamlining, making repeatable processes and work streams based on generative AI.
securing and trusting, making trusted decisions and actions from a basis of a trusted infrastructure, taking all the trusted proprietary company data to provide the right outcomes and recommendations that impacts business decision makers to change business direction and strategies.
Innovative approaches require the enterprise to rethink their AI status quo. With new ways, new models and much larger data sets, businesses are able to adopt large language model training and benefit from Generative AI applications and methodology. Their traditional AI approaches can now be bolstered and complemented with Enterprise-grade Generative AI foundations.
Dell Technologies and NVIDIA and are excited to announce Project Helix, a unique collaboration that will make the promise of Generative AI real for the enterprise.
Project Helix will deliver a full-stack solution, built on Dell hardware and utilizing the award-winning software stack and accelerator technology of Nvidia. Bringing together the deep knowledge and creativity of Nvidia with the global customer knowledge and technology expertise of Dell, Project Helix will:
Deliver full-stack scalable solutions: Deliver full-stack generative AI solutions built on the best of Dell infrastructure and software, in combination with the latest NVIDIA accelerators, AI software and expertise.
Provide expertise and deployment guidance: Provide expertise and guidance for enterprises to build, train, and model-tune Generative AI on-premise to help solve specific business challenges.
Assist from development to multi-site deployment: Assist enterprises with the entire GenAI lifecycle, from infrastructure provisioning, large model development and training, pre-trained model fine tuning, to multi-site, multi-cloud model deployment (and consumption) and large model inferencing.
Deliver trusted AI outcomes with integrity: Helix is aimed at helping deliver Trusted AI outcomes, that the decision outcomes are made from objective data inputs (assessed and sanitized according to a defined set of “guardrails”) which assures data integrity. These decisions elevate the quality of those results to being trusted as well as providing the rationale for the decisions/recommendations provided by the model.
Types of guardrails
Topical guardrails prevent apps from veering off into undesired areas.
Safety guardrails can filter out unwanted language
Security guardrails enforce that confidential information stays that way.
Ensure security and privacy of business data: Ensure security and privacy of sensitive and proprietary company data, as well as compliance with government regulations.
Flexible Capex/Opex consumption with APEX flex consumption approaches
To that end, project Helix is about
and is our no compromise AI Acceleration product. The XE9680 is Dell’s first entry of a 8 way GPU SXM server that will offer both Nvidia’s next generation H100 sxm GPU’s as well as their current A100 offering, providing our customers optionality of performance and power. This server can handle the largest and most demanding AI, ML, DL training/HPC modeling & simulation initiatives, and now Generative AI, all in a 6U air cooled package. Once again this will feature next gen technologies with Intel’s 4th Generation Xeon processors, DDR5 memory, E3 storage but also has 10 PCIe Gen 5 slots enabling 1:1 GPU: NIC ratio which means you get max GPU utilization with smooth data transfer in and out resulting in minimizing latency. This product is great for HPC super computing clusters and the largest data sets for AI, used for large scale natural language processing, recommendation engines and neural network. Also great for conversational AI, chatbots, voice recognition and audio applications Verticals: Generative AI/ChatGPT, CRISP, CSP Finance, HPC, Healthcare, Federal, Research/Universities 1- Dell MLCommons MLPerf Inference v3.0 vs v2.0, BERT 99.9 v3.0 Offline scenario renders 843 percent more improvement compared to PowerEdge XE8545 four-way NVIDIA A100 SXM. March 2023, https://infohub.delltechnologies.com/p/dell-servers-excel-in-mlperf-tm-inference-3-0-performance/ 2- Dell PowerEdge XE9680 with 8x Nvidia H100 GPUs delivers 6.4x (+543%) more TeraFLOPS than the PowerEdge XE8545 with 4x Nvidia A100 GPUs in HPL benchmark. Testing conducted by Dell in March of 2023. Performed on PowerEdge XE9680 with 8x Nvidia H100 SXM5-80GB and PowerEdge XE8545 with 4x Nvidia A100-SXM-80GB. 3-NVIDIA, LLM training test, For a cluster of 4K GPUs: 1T parameter model on A100 takes 1 month, while only 1 week for same size H100 cluster. NVIDIA conducted test, 2023.
4-Dell white paper, Launching & Running Large Language Models on a single Dell server produces outstanding results, 2023. Throughput in tokens per second of the BLOOM-176B large language model (LLM) on the Dell Technologies PowerEdge XE9680 server vs. the Jean Zay supercomputer using the same benchmarking code and same model instantiation with HuggingFace Accelerate and float16 precision. In this instance, the XE9680 demonstrated a 70% increase in throughput as compared and the Jean Zay supercomputer, regardless of batch size. https://www.delltechnologies.com/asset/en-us/products/servers/industry-market/launching-llms-on-poweredge-xe9680.pdf 1. Based on Dell Technologies internal analysis comparing the max bandwidth of the ECS EXF900 (511 MB/s) to the max bandwidth of the ECS EX300 (24 MB/s) for 10KB writes, November 2020. Actual performance will vary. 2. Based on an Enterprise Strategy Group Technical Review commissioned by Dell EMC, “Accelerating the Artificial Intelligence Journey with Dell EMC Ready Solutions for AI”, August 2018. 3-Dell spec sheet. Raw capacity up to exabytes (ECS) or up to 186PB in a 252-node cluster (PowerScale)
CloudIQ delivers intelligent infrastructure insights by combining machine intelligence and human intelligence, enabling algorithms to be continuously updated leveraging Dell Technologies product and subject matter expertise, combined with industry best practices to address the most potentially impactful issues. CloudIQ helps organizations:
The PowerEdge integrated Dell Remote Access Controller (iDRAC) delivers advanced, agent-free local and remote server administration.
OpenManage Enterprise (OME), 1: many on-prem systems management and automation.
OpenManage Enterprise Power Manager plugin: Analyze and prevent a server-related power and/or thermal outage
Address sustainability
Get a server into data center production faster
Storage For nearly two decades, we’ve been successfully helping our customers solve their unstructured data challenges. The market data and industry analyst commentary reflect that commitment and proven track record. To date, we’ve shipped over 17 exabytes of capacity, including a single customer managing more than 1EB of unstructured data on OneFS. The bottom-line: We have unmatched experience in addressing these challenges, and every day we think about how to help our customers overcome them in a sustainable way. OneFS has always been a highly flexible software platform which was designed from the ground up to provide agility when managing data at scale. In fact, OneFS has historically provided many of the benefits you would usually associate with a “software defined” solution, such as: single pool aggregation of multiple types of media, policy-based management and automation, dynamic scaling of performance and capacity, and a distributed architecture – just to name a few. We provide cloud consumption options not available on scalable file solutions until now. Our multi-cloud offering delivers the full fidelity experience of OneFS in a hosted, managed service with simultaneous connectivity on the same IP range with Google, AWS and Azure. Our native integration with Google enables a cloud-native experience where
Google handles all provisioning, management and billing. All these solutions can be run as cloud-only or interoperate with on-prem instances of OneFS for a hybrid deployment
.
Dell Technologies Services helps customers drive their businesses forward by meeting them where they are in their digital journey. We collaborate with customers on IT solutions that give them an advantage today and prepare them for what’s ahead.
Services offerings:
Dell Technologies is a leader and helps customers with solutions that can drive their performance, harness AI and accelerate their initiatives. With the right foundation, customers can accelerate their time to results with building blocks from high performance servers, fast GPU accelerators to complete turnkey, certified and validated solutions for their environment. Deploying technology, innovation starting with the compute helps customers with the ample horsepower needed to solve larger and greater complexity problems, within HPC as well all other AI initiatives. With the continuous increase in data generation across the organization, more compute is needed to extract value for business initiatives. With the best AI technology foundation as well as a framework of operations and suites to help virtualize AI, businesses can readily make available resources and democratize business wide. Suites like VMware vSphere and NVIDIA AI for Enterprise rapidly helps drive AI resources to those who need the access. IDC found that AI leaders recognize that many of the demanding AI workloads require on-premise compute, shift AI-intensive algorithms to that infrastructure and thereby own their infrastructure for optimal on-premise performance. While businesses can also utilize the cloud, for massive data and compute resources, the operations remain on-premise for lowest latency and fastest performance. These benefits help organizations be more competitive in their business, from faster insights which foster faster decision-making which drives more value sooner. Time-to-value is quickened furthering AI leaders as well as AI evaluators in their transformation Want to read more? We have just published a white paper which you can download, by clicking the screenshot below
To understand how the new Dell solution is superior, watch the interview with Andrea Booker, Director of Product Management, Accelerator Portfolio, Dell Technologies & Steen Graham, CEO & Founder, Scalers AI, sit with Lisa Martin & Dave Vellante at Dell Technologies World 2023