IBM And NVIDIA Power New Scale-Out Gear For AI

Accelerating deep learning (DL) training – on GPUs, TPUs, FPGAs or other accelerators – is in the early days scale-out architecture, like the server market was in the mid-2000s. DL training enables the advanced pattern recognition behind modern artificial intelligence (AI) based services. NVIDIA GPUs have been a major driver for DL development and commercialization, but IBM just made an important contribution to scale-out DL acceleration. Understanding what IBM did and how that work advances AI deployments takes some explanation.

Scale Matters

TIRIAS Research

Key Definitions

Inference scales-out. Trained DL models can be simplified for faster processing with good enough pattern recognition to create profitable services. Inference can scale-out as small individual tasks running on multiple inexpensive servers. There is a lot of industry investment aimed at lowering the cost of delivering inference, we’ll discuss that in the future.

The immediate challenge for creating deployable inference models is that, today, training scales-up. Training requires large data sets and high numeric precision; aggressive system designs are needed to meet real-world training times and accuracies. But cloud economics are driven by scale-out.

The challenge for cloud companies deploying DL-based AI services, such as Microsoft’s Cortana, Amazon’s Alexa and Google Home, is that DL training has not scaled well. Poor off-the-shelf scaling is mostly due to the immature state of DL acceleration, forcing service providers to invest (in aggregate) hundreds of millions of dollars in research and development (R&D), engineering and deployment of proprietary scale-out systems.

NVLink Scales-Up in Increments of Eight GPUs

GPU evolution has been a key part of DL success over recent years. General purpose processors were, and still are, too slow at processing DL math with large training data sets. NVIDIA invested early in leveraging GPUs for DL acceleration, in both new GPU architectures to further accelerate DL and in DL software development tools to enable easy access to GPU acceleration.

An important part of NVIDIA’s GPU acceleration strategy is NVLink. NVLink is a scale-up high-speed direct GPU-to-GPU interconnect architecture that directly connects two to eight GPU sockets. NVLink enables GPUs to train together with minimum processor intervention. Prior to NVLink, GPUs did not have the low-latency interconnect, data flow control sophistication, or unified memory space needed to scale-up by themselves. NVIDIA implements NVLink using its SXM2 socket instead of PCIe.

NVIDIA’s DGX-1, Microsoft’s Open Compute Project (OCP) Project Olympus HGX-1 GPU chassis and Facebook’s “Big Basin” server contribution to OCP are very similar designs that each house eight NVIDIA Tesla SXM2 GPUs. The DGX-1 design includes a dual-processor x86 server node in the chassis, while the HGX-1 and Big Basin designs must be paired with separate server chassis.

Microsoft’s HGX-1 can bridge four GPU chassis by using its PCIe switch chips to connect the four NVLink domains to one to four server nodes. While all three designs are significant feats of server architecture, the HGX-1’s 32-GPU design limit presents a practical upper limit for directly connected scale-up GPU systems.

TIRIAS Research

Microsoft HGX-1 motherboard with eight SXM2 sockets (four populated)

The list price for each DGX-1 is $129,000 using NVIDIA’s P100 SXM2 GPU and $149,000 using its V100 SXM2 GPU (including the built-in dual-processor x86 server node). While this price range is within reach of some high-performance computing (HPC) cluster bids, it is not a typical cloud or academic purchase.

Original Design Manufacturers (ODMs) like Quanta Cloud Technology (QCT) manufacture variants of OCP’s HGX-1 and Big Basin chassis, but do not publish pricing. NVIDIA P100 modules are priced from about $5,400 to $9,400 each. Because NVIDIA’s SXM2 GPUs account for most of the cost of both Big Basin and HGX-1, we believe that system pricing for both is in the range of $50,000 to $70,000 per chassis unit (not including matching x86 servers), in cloud-sized purchase quantities.

Facebook’s Big Basin Performance Claims

Facebook published a paper in June describing how it connected 32 Big Basin systems over its internal network to aggregate 256 GPUs and train a ResNet-50 image recognition model in under an hour with about 90% scaling efficiency and 72% accuracy.

While 90% scaling efficiency is an impressive achievement for state-of-the-art, there are several challenges with Facebook’s paper.

The eight-GPU Big Basin chassis is the largest possible scale-up NVIDIA NVLink instance. It is expensive, even if you could buy OCP gear as an enterprise buyer. Plus, Facebook’s paper does not mention which OCP server chassis design and processor model they used for their benchmarks. Which processor it used may be a moot point, because if you are not a cloud giant, it is very difficult to buy a Big Basin chassis or any of the other OCP servers that Facebook uses internally. Using different hardware, your mileage is guaranteed to vary.

Facebook also does not divulge the operating system or development tools used in the paper, because Facebook has its own internal cloud instances and development environments. No one else has access to them.

The net effect is that it is nearly impossible to replicate Facebook’s achievement if you are not Facebook.

TIRIAS Research

Facebook Big Basin Server

IBM Scales-Out with Four GPUs in a System

IBM recently published a paper as a follow-up to the Facebook paper. IBM’s paper describes how to train a Resnet-50 model in under an hour at 95% scaling efficiency and 75% accuracy, using the same data sets that Facebook used for training. IBM’s paper is notable in several ways:

  1. Not only did IBM beat Facebook on all the metrics, but 95% efficiency is very linear scaling.
  2. Anyone can buy the equipment and software to replicate IBM’s work. Equipment, operating systems and development environments are called out in the paper.
  3. IBM used smaller scale-out units than Facebook. Assuming Facebook used their standard dual-socket compute chassis, IBM has half the ratio of GPUs to CPUs – Facebook uses a 4:1 ratio and IBM uses a 2:1 ratio.

IBM sells its OpenPOWER “Minsky” deep learning reference design as the Power Systems S822LC for HPC. IBM’s PowerAI software platform with Distributed Deep Learning (DDL) libraries includes IBM-Caffe and “topology aware communication” libraries. PowerAI DDL is specific to OpenPOWER-based systems, so it will run on similar POWER8 Minsky-based designs and upcoming POWER9 “Zaius”-based systems (Zaius was designed by Google and Rackspace), such as those shown at various events by Wistron, E4, Inventec and Zoom.

PowerAI DDL enables creating large scale-out systems out of smaller, more affordable, GPU-based scale-up servers. It optimizes communications between GPU-based servers based on network topology, the capabilities of each network link, and the latencies for each phase of a DL model.

IBM used 64 Power System S822LC systems, each with four NVIDIA Tesla P100 SXM2-connected GPUs and two POWER8 processors, for a total of 256 GPUs – matching Facebook’s paper. Even with twice as many IBM GPU-accelerated chassis required to host the same number of GPUs as in Facebook’s system, IBM achieved a higher scaling efficiency than Facebook. That is no small feat.

TIRIAS Research

IBM Power System S822LC with two POWER8 processors (silver heat sinks) and four NVIDIA Tesla P100 SXM2 modules

Commercial availability of IBM’s S822LC for low volume buyers will be a key element enabling academic and enterprise researchers to buy a few systems and test IBM’s hardware and software scaling efficiencies. The base price for an IBM S822LC for Big Data (without GPUs) is $6,400, so the total price of a S822LC for High Performance Computing should be in the $30,000 to $50,000 ballpark (including the dual-processor POWER8 server node), depending on which P100 model is installed and other options.

Half the battle is knowing that something can be done. We believe IBM’s paper and product availability will spur a lot of DL development work by other hardware and software vendors.

— The author and members of the TIRIAS Research staff do not hold equity positions in any of the companies mentioned. TIRIAS Research tracks and consults for companies throughout the electronics ecosystem from semiconductors to systems and sensors to the cloud.

[“Source-forbes”]

Nike, NBA unveil innovative new jerseys, gear

The NBA and Nike unveil the ‘Statement’ edition uniforms for the 2017-18 season.

LOS ANGELES, Calif. –Ushering in a radical evolution of the fan experience, NIKE, Inc. and the NBA today unveiled three new innovations: the Nike NBA Connected Jersey, the first NBA jersey that gives the wearer an all-access pass into the world of their favorite team and players; Statement Edition uniforms; and the Nike Therma Flex Showtime Warm-up Jacket, the first hooded jacket to be worn on the court during gametime. The official announcements took place at an event in Los Angeles attended by players from all 30 NBA teams.

“When you think about basketball, you think of Nike and the NBA,” said Mark Parker, Chairman, CEO and President of NIKE, Inc. “After helping fuel the sport’s impact for years, we are thrilled to now be able to push the boundaries of what’s possible directly with the NBA. From modernizing the look of the game to revolutionizing how fans can be a part of it, we can’t wait for the season to start.”

“Our Nike uniforms set a new standard for connecting our most passionate fans with their favorite NBA teams and players,” said Adam Silver, NBA Commissioner. “Nike has once again taken the game of basketball to a new level, this time with one of the most advanced jerseys in sports.”

Using new NikeConnect technology, each adult-sized Nike NBA Connected Jersey will have an embedded NFC (near field communication) chip that will launch real-time team and player content such as pregame arrival footage, highlight packages and top players’ favorite music playlists – all on the jersey owner’s mobile device. Throughout the season, a wealth of exclusive offers and experiences will bring fans closer to the game they love.

To unlock it all, fans will download the NikeConnect app and tap the jersey with their smartphone. These jerseys become available online globally on Sept. 29, rewarding the game’s fans with a revolutionary experience that creates a new point of connection between fans and the game.

“In our new connected era, Nike will lead with products that build a relationship with our consumers,” said Trevor Edwards, President, Nike Brand. “With NikeConnect, we deliver to basketball fans everywhere the personal and premium experiences we know they love.”

Nike and the NBA also unveiled the third of four core team uniforms – the Statement Edition, inspired by teams’ desire to make a bold statement the moment they step on the court. Nike’s designers worked with the teams and the brand’s roster of athletes to create new and updated uniforms for this third edition.

The Statement Edition uniform follows the earlier unveiling of the Association and Icon Edition uniforms. As previously announced, the home teams will determine which uniforms to wear at their games. The fourth team edition will be released later this year.

In addition to the new uniforms, Nike and the NBA debuted the Nike Therma Flex Showtime Warm-up Jacket,the first on-court warm-up jacket with a performance hood. Celebrating the style of the game, Nike designers refreshed a beloved apparel staple into a new performance product for athletes on and off the court.

The jacket, created with new Nike Therma Flex fabric, helps athletes maintain the right body temperature during pregame warm-ups and downtime during a game – a critical element for peak performance. Developed with input from top NBA athletes, the reimagined hood allows for optimal vision, hearing and warmth, delivering a greater intersection between performance and style – a key request from NBA athletes.

The Nike NBA Connected Jersey is available exclusively online starting Sept. 29, and the Nike NBA uniforms and Nike Therma Flex Showtime Warm-up Jacket will debut on-court on Sept. 30, the start of the NBA preseason. All Statement Edition jerseys will be available at retail on Nov. 20, and will begin being worn on-court on Nov. 25.

[“Source-nba”]

DuPont launches landmark low-emission alternative to Delrin 100 for gear applications

Image result for DuPont launches landmark low-emission alternative to Delrin 100 for gear applications

The launch of DuPont™ Delrin® 100CPE NC010 marks a landmark moment in DuPont’s offering to the automotive sector as well as non-auto gear applications. For sixty years Delrin®100 was the resin of choice for major OEMs and Tier 1’s in high performance gears and particularly in automotive actuators such as window lifters, steering support, safety restraint systems, wipers, and many more. Now, with the launch of this new grade, DuPont can provide its customers with superior environmental performance. In collaboration with a major industry partner, DuPont developed and is implementing the new grade in large scale commercial production.

DuPont is constantly working closely with leading automotive actuator companies on developing innovative solutions. Delrin® 100CPE NC010 is the result of one of these fruitful and longstanding collaborations, out of which was created a new, low-emission resin that outperforms high viscosity polyacetal resins. The thermal properties and high-temperature tensile creep performance of Delrin® 100CPE NC010 are comparable with the state of the art product, and when it comes to low-emission- and molding productivity performance, Delrin® 100CPE NC010 exceeds all expectations.

In this regard, one of the key benefits of Delrin® 100CPE NC010 is its level of

formaldehyde emissions, which is below 2 ppm according to the VDA 275 testing method. This allows compliance with current and future toughest emissions industry norms.

Delrin® 100CPE NC010 is the only high-performance resin with formaldehyde emissions below 2 ppm and strong molding productivity performance, enabling continuous molding for several thousand shots. Using Delrin® 100CPE NC010 leaves a very clean surface with a step change improvement in mold deposit performance versus incumbent solutions resulting in a cost reduction for manufacturers. The smell of formaldehyde during molding is also significantly reduced with the use of Delrin® 100CPE NC010. Its excellent molding performance combined with low emissions renders Delrin® 100CPE NC010 an attractive product for high load and high productivity gears in appliances, furniture and consumer segment.

“Mutual trust is at the heart of sustaining rewarding relationships with customers. They know that we can help them formulate new resins that combine the highest performance requirements for most demanding gear applications. Customers know that they can also rely on our global production capability”, says Lukas Bartek – Global Gears Segment Leader, DuPont Performance Materials.

DuPont is committed to developing innovative solutions for their customers’ main challenges.  By directly working together with customers throughout the entire value chain, and by sharing expertise and pooling resources, DuPont continues to turn out of the box ideas into concrete offerings.

To demonstrate the commitment to its customers and partners, DuPont is an active contributor to the 7th International Conference on Gears 2017, held in Garching near Munich at the Gear Research Centre (FZG) of the Technical University of Munich from September 13th to 15th, 2017.

[“Source-automotiveworld”]

Reebok introduces next wave of UFC gear, including ‘UFC Legacy’ series

The UFC and Reebok announced their new Fight Night Collection on Tuesday, which includes updates to their standard gear, as well as a Legacy Series dedicated to fighters competing in the main events of pay-per-views or in championship bouts.

According to Reebok General Manager of Training Corinna Werkle, the changes are meant to further accentuate the individuality of the UFC’s athletes.

“With the new UFC Fight Night Collection, we wanted to give athletes a product that meets the quality and performance demands that they depend on in the Octagon, but also offer them an opportunity to let their unique confidence and personalities shine through,” said Werkle in a press release.

“Our design, product and research teams have spent countless hours talking to and working with the UFC athletes and MMA community to do just that. We are thrilled to debut our latest iteration of the UFC athlete fight night apparel with the new UFC Fight Night Collection and are committed to constantly reimagining and evolving the Reebok Combat collection to adapt to athlete requirements and appeal to UFC fans.”

Reebok’s exclusive apparel deal with the UFC has drawn ire from both fighters and fans alike due to issues with the distribution of sponsorship money, criticisms over the design of the gear itself, and numerous errors.

It remains to be seen if this is a step in the right direction to appeasing the athletes who have to wear the clothing and the viewers who are expected to purchase it.

The Legacy Series will make its debut at UFC 215 on Saturday in Edmonton, Alberta. Titleholders Demetrious Johnson and Amanda Nunes will sport customized walkout gear, as will respective challengers Ray Borg and Valentina Shevchenko.

Check out the gallery below to see what this weekend’s headliners will be wearing, plus additional shots featuring several UFC stars including champions Joanna Jedrzejczyk, Max Holloway, and Stipe Miocic.

[“Source-mmafighting”]