IBM Bare Metal Cloud Targets AI with New P100 GPUs

By Tiffany Trader

April 5, 2017

IBM announced today that it will be adding Nvidia P100 graphics processors to its Bluemix cloud later this month, becoming the “first major global cloud vendor” to provide the high-end “Pascal” GPUs. Big Blue is targeting the new hardware at customers who run compute-heavy workloads, such as artificial intelligence, deep learning, data analytics and high-performance computing.

Unlike Nimbix, the heterogeneous cloud vendor that began offering NVLink’d Nvidia P100 GPUs on the IBM “Minsky” Power8 platform last October (2016), IBM will be using PCIe form factor cards within an Intel x86 server. This is not really a surprise since IBM operates most of its cloud servers on Intel-based chip sets. Customers will be able to add up to two Nvidia P100 cards to a dual Xeon E5-2690 v3 machine (24-core CPUs running at 2.6 GHz).

The IBM cloud does have some Power server options for specific big data workloads but it does not have an expanded assortment of Power, says Jay Jubran, Global Offering Management for Compute at IBM Cloud. A plan to integrate Power8 based systems with NVIDIA P100 GPUs into the IBM cloud portfolio is underway. “We are are working side by side with the Power Systems team to ensure that IBM Cloud will deliver access to the best of IBM technology to allow customers to run HPC and AI workloads,” Jubran told us.

The Power8 “Minsky” platform enables tight coupling of the Power CPU and P100 GPU over Nvidia’s proprietary NVLink interconnect. The mezzanine form factor P100 also provides nearly 13 percent better raw performance than the PCIe card, 5.3 double-precision teraflops versus 4.7. Both versions provide 16 gigabytes of HBM2 stacked memory. Networking on the IBM cloud stands at 10 Gigabit Ethernet today with IBM stating that future platforms might go up to 25 Gigabit Ethernet.

IBM will be first to the P100 punch in terms of major cloud providers, but as we have seen, other cloud purveyors are advancing with P100 plays of their own. Here’s a rundown:

Nimbix – As mentioned above, Nimbix added IBM Power S822LC for HPC systems (codenamed “Minsky”) to its heterogeneous HPC cloud platform last October. Target markets include high-performance computing, data analytics, in-memory databases, and machine learning.

Cirrascale – On its GPU-driven deep learning infrastructure as a service, San Diego, Calif.-based Cirrascale offers a number of P100-based server configurations, including four-way and eight-way Intel-based GPU servers and IBM Power8 Systems with two and four GPU options.

Google – The Google Cloud platform website states that P100s are “coming soon.” Google will also be incorporating AMD FirePro S9300 x2 GPUS into its infrastructure. Google began offering K80 GPU-equipped virtual machines (as a beta release) in February of this year.

Microsoft – Microsoft last month revealed blueprints for a new open source P100-based accelerator – HGX-1 – developed under Project Olympus. It’s an accelerator box with eight Tesla P100s, connected in the same hypercube mesh as the Nvidia DGX-1 server and also leveraging the NVLink interconnect. The HGX-1 hooks to servers via PCIe interface. We’re to assume the boxes, being manufactured by Ingrasys, will show up on Azure but Microsoft hasn’t indicated when that will be. The company has had some notable delays in GPU rollouts – announcing a planned K80 instance in September 2015, and AWS beating them to  general availability a year later.

Tencent – Two weeks ago, Chinese cloud giant Tencent said it will offer a range of cloud products that will include GPU cloud servers incorporating Nvidia Tesla P100, P40 and M40 GPU accelerators and Nvidia deep learning software. Tencent Cloud launched GPU servers based on Nvidia Tesla M40 GPUs and NVIDIA deep learning software in December; it expects to integrate cloud servers with up to eight Pascal-based GPUs each by mid-year.

Reigning cloud king Amazon does not yet offer Nvidia’s Pascal-based silicon (the P100 or the P40 inferencing engine). Amazon’s most recent P2 instance family is backed by Kepler-generation K80 parts, rolled out last September (2016).

IBM emphasized the advantage of its bare metal cloud offering, compared to the multi-tenant environments of AWS and the other mega-cloud providers, especially for HPC workloads. “The main reason why people come to IBM cloud, other than the global presence, is the performance and consistency of having access to the bare metal. The bare metal allows us to give better performance than any other virtualized environment with the same specification because we do not have the hypervisor tax which is roughly 10-15 percent of the CPU power,” said Jubran.

“We find HPC workloads typically find their way to the IBM cloud. If the customer is looking to run HPC on an hourly basis sometimes you’ll see them go to other clouds, but in terms of monthly consumption we have the best offering in terms of performance and price value,” he added.

The bare metal infrastructure is also attractive to the graphics community, for gaming, especially a subset called cognitive gaming, and for engineering, said Jubran. Financial services, healthcare, and retail are all target verticals.

Customers that prioritize highly elastic resources and pay-by-the-sip pricing typically go to IBM’s competitors, Jubran noted, but their core customers are the ones who understand the performance metrics that IBM offers.

“We are attracting both digital customers looking for performance, gaming customers and born on the web type customers who are looking for bare metal performance, but scalability of the cloud. And we also get in the higher end of the spectrum in terms of enterprise and that is because of IBM obviously being an enterprise-focused company from day one and they put trust in IBM to bring their workload to our datacenters. So having both aspects of the spectrum keeps us on the innovative side in terms of digital and keeps us on the high-performance secure side for the enterprise,” said Jubran.

Aside from the advantage of this enterprise trust factor, IBM’s distributed model of 50 datacenters (built up since the Softlayer acquisition in 2013 for a reported $2 billion) gives them the geo-precision to provide local data sovereignty for their customers and is a natural fit for edge computing (important for AI training workflows and for IoT). For many customers, proximity of compute and data are far more important than saving on compute cost offered by the greater elasticity of mega-datacenters. A typical IBM datacenter unit consists of roughly 20,000 servers; in the hyperscaler world, that’s pretty small.

The Tesla P100 joins Nvidia’s portfolio of GPU offerings on the IBM Cloud, including the older Tesla K2 GPU, the Tesla M60 for virtualized graphics and the Tesla K80, which IBM added in 2015, about a year ahead of the competition. IBM expects most of its K80 customers will be migrating over to the P100 servers as they begin adding the parts later this month. “We also expect newcomers into the AI platform as the P100 is the most powerful GPU in terms of AI workloads that are based on TensorFlow, Caffe, Nvidia SDK or any of the AI SDKs available out today,” said Jubran. “With so much focus from all the different industries in AI, I think you will see more and more of those workloads coming to IBM cloud and the P100 will enable that. If you look at the Nvidia material for P100 it is the most powerful GPU for both training and inferencing, the two aspects of AI.”

“With all key deep learning frameworks GPU-accelerated and over 400 HPC applications in a broad range of domains, including the top 10 high performance computing applications, IBM Cloud customers can quickly tap into the power of the our GPU platform to boost performance, accelerate time to results and save money,” Nvidia’s Vice President of Accelerated Computing Ian Buck wrote in a blog post.

The cost for the new Pascal-based hardware is $750 per month per P100 GPU card, tacked on to the price of the server. This adds a 50 percent premium over the cost of the K80s ($500 per card) but the P100 card offers a 60 percent additional performance improvement over the K80. That should make switching a no-brainer and while IBM won’t be forcing customers with active workloads off the K80, they are planning to sunset the older Teslas as inventory depletes.

Editor’s note — April 6, 2017: In an earlier version of this article, we reported (based on information IBM shared with us) that the Power8 “Minsky” platform was not on IBM’s cloud roadmap. After the article was published, IBM contacted us to let us know that it does have plans to incorporate Power8 based systems with Nvidia P100 GPUs into its cloud portfolio. We have amended the story to include this updated information.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire