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”]

Asus Launches New ROG Laptops With Nvidia GeForce GTX 10-Series GPUs in India

Asus Launches New ROG Laptops With Nvidia GeForce GTX 10-Series GPUs in India

HIGHLIGHTS

  • The Asus ROG Strix GL502VS is priced at Rs. 1,81,990
  • The Asus ROG G752VS is priced at Rs. 2,47,990
  • Both the laptops come with the new NVidia 10-series graphics card

After launching the Asus ZenBook Flip UX360CA, the company has now introduced two new Republic of Gamers (ROG) laptops in India. Expanding its lineup, the new ROG Strix GL502VS and ROG G752VS are powered by Nvidia GeForce GTX 10-series graphics card. The laptops will be available across all online and offline leading retail outfits at a starting price of Rs. 1,81,990 and Rs. 2,47,990 respectively.

As mentioned, the big highlight of the two laptops are the VR-ready Nvidia GeForce GTX 10-series graphics cards, and they includes Nvidia G-Sync technology as well.

The ROG Strix GL502VS comes with the GeForce GTX 1070 GPU, a 15-inch display with 178-degree viewing angle, and weighs approximately 2.3kg. It offers up to 32GB of DDR4 RAM, and is designed in a way to maximise cooling efficiency. In addition, there is also the company’s in-built GameFirst III technology to ensure lag-free gaming, and an interface that lets users choose between four preset package priority modes (Optimization, Game, Media Streaming, and File Sharing).

Asus also launched the comparatively cheaper ROG G752VS. This laptop is powered by a sixth-generation Intel Core i7 processor paired with Nvidia GeForce GTX 1070 graphics card and up to 64GB of DDR4 RAM. The unlocked processor comes along with an overclocking toolkit, allowing users to overclock the system to speeds of up to 4.0GHz. Additionally, it has a backlit anti-ghosting keyboard with 30-key rollover. The ROG G752VS comes in Armor Titanium, and Plasma Copper colour variants.

Asus launched its new lineup ROG gaming laptops in India in April this year. At that time, the company launched four laptops and two desktops to flag off the ROG series. In July, Asus even launched the ROG GX700 at a whopping price of Rs. 4,12,990. The premium gaming laptop comes with a detachable ROG-exclusive Hydro Overclocking System cooling module that has dual 92mm radiators the company says are capable of dissipating over 500W of heat. It is even bundled with ROG-themed suitcase to lug around.

Tags: Asus, Laptops, PCs, Asus Republic of Gamers, Asus ROG Strix GL502VS, Asus ROG G752VS

[“Source-Gadgets”]

Nvidia GeForce GTX 1060 Announced, Available from July 19 Priced at Rs. 22,999

Nvidia GeForce GTX 1060 Announced, Available from July 19 Priced at Rs. 22,999

Nvidia GeForce GTX 1060 Announced, Available from July 19 Priced at Rs. 22,999
HIGHLIGHTS
Nvidia promises performance matching that of the GeForce GTX 980
The GTX 1060 will compete head-to-head with AMD’s Radeon RX 480
GTX 1060 graphics cards are aimed at value-conscious gamers
Nvidia has announced its new GeForce GTX 1060 GPU, aimed at mainstream gamers. Graphics cards based on the new GTX 1060 will be available worldwide from partner companies including Asus, MSI, Gigabyte and Zotac on July 19. There will be only one configuration featuring 6GB of GDDR5 RAM. Pricing in the US is $249 (approximately Rs. 16,858) but the official price in India has not been announced yet. We can expect it to be considerably higher to account for import duties and distribution costs.

UPDATE, July 8 2016: Nvidia’s recommended pricing for India has been confirmed at Rs. 22,999.

The new GeForce GTX 1060 comes hot on the heels of Nvidia’s high-end GTX 1080 and GTX 1070 GPUs, and seems to be aimed at the same market segment as AMD’s new Radeon RX 480. The timing of the launch, and of earlier leaks, could be a direct shot at AMD. Interestingly, the Radeon RX 480 with 4GB of RAM is priced $50 lower in the USA at $199, but AMD’s higher overheads give Nvidia a decisive advantage here in India. However, the GTX 1060’s price could make it less popular than its predecessor, the GTX 960 which sells for under Rs. 18,000.

Gamers will benefit from the same Pascal architecture that the GTX 1080 and GTX 1070 are based on, although the actual GPU is a smaller unit codenamed GP106, rather than the GP104 which the other two are based on. The GTX 1060 has 1280 programmable graphics processing units and can run at up to 1.7GHz when under full load. The 6GB of GDDR5 memory has a 192-bit pathway to the GPU, resulting in memory bandwidth of up to 8Gbps. The maximum power draw is rated at 120W, and cards featuring Nvidia’s reference cooler will have a single 6-pin PCIe connector.
nvidia_geforce_gtx1060_specs_nvidia.jpg
Most notably, the GeForce GTX 1060 dispenses with SLI connectors altogether. Nvidia had de-emphasised SLI with the launch of the GTX 1080 and GTX 1070, first saying enthusiasts would need a special key to unlock it, and then limiting it to two cards down from four. Now, the company says its research shows that only a miniscule number of people ever used SLI on mid-range cards, and it believes that one GTX 1060 delivers enough performance on its own.

Nvidia will offer a Founders’ Edition version of the GTX 1060, but it has a simple plastic shroud and blower fan rather than the elaborate metal vapour-chamber cooler that gives the GTX 1080 Founders’ Edition its premium pricing. It will be available only in France, Germany and the UK, directly through Nvidia’s online store, priced at EUR 319/GBP 275. Partner companies will offer GTX 1060 cards with their own custom cooler and PCB designs, not Founders’ Editions with their own branding, as is the case with the GTX 1080 and GTX 1070.

GTX 1060 cards will have three DisplayPort 1.4 outputs as well as a single HDMI 2.0b port and one dual-link DVI-D port by default. Partners might change this arrangement around on their custom designs, with extra HDMI outputs most likely replacing DisplayPort to allow users to connect a VR headset and monitor at the same time.

GeForce_GTX_1060_Back_nvidia.jpg
Nvidia’s new Pascal architecture, which underpins the entire GeForce 10-series, is the company’s first on a new 16nm process, which results in greatly improved performance and lower power consumption than previous offerings. Pascal has also been designed to enable simultaneous multi-projection which allows for multiple perspectives of the same scene to be rendered in one pipeline pass, greatly improving performance for VR and multi-monitor experiences. SMP support is built into the Unreal 4 and Unity engines, and will be coming to over 30 titles including Obduction, Raw Data, Unreal Tournament, and Adr1ft.

The company also points out that the Ansel screen capture feature which debuted with the more expensive cards is also fully supported, and will debut in July with The Witcher 3: Wild Hunt and Mirror’s Edge: Catalyst.

Nvidia promises equal or greater performance from the GTX 1060 as from last year’s GeForce GTX 980, plus significant overclocking potential. Stay tuned for a full review once these cards are available in the market.

Share a screenshot and win Samsung smartphones worth Rs. 90,000 by participating in the #BrowseFaster contest.

Tags: GeForce, GeForce GTX 1060, Nvidia, Nvidia Ansel, Nvidia GeForce, Nvidia GeForce GTX 1060, Nvidia GeForce GTX 1060 availability, Nvidia GeForce GTX 1060 features, Nvidia GeForce GTX 1060 launch, Nvidia GeForce GTX 1060 launch in India, Nvidia GeForce GTX 1060 performance, Nvidia GeForce GTX 1060 price, Nvidia GeForce GTX 1060 price in India, Nvidia GeForce GTX 1060 specifications, VR, gaming, Nvidia Pascal, Pascal

[“Source-Gadgets”]

Nvidia GeForce GTX 1060 Announced, Available from July 19 Priced at Rs. 22,999

Nvidia GeForce GTX 1060 Announced, Available from July 19 Priced at Rs. 22,999

HIGHLIGHTS

  • Nvidia promises performance matching that of the GeForce GTX 980
  • The GTX 1060 will compete head-to-head with AMD’s Radeon RX 480
  • GTX 1060 graphics cards are aimed at value-conscious gamers

Nvidia has announced its new GeForce GTX 1060 GPU, aimed at mainstream gamers. Graphics cards based on the new GTX 1060 will be available worldwide from partner companies including Asus, MSI, Gigabyte and Zotac on July 19. There will be only one configuration featuring 6GB of GDDR5 RAM. Pricing in the US is $249 (approximately Rs. 16,858) but the official price in India has not been announced yet. We can expect it to be considerably higher to account for import duties and distribution costs.

UPDATE, July 8 2016: Nvidia’s recommended pricing for India has been confirmed at Rs. 22,999.

The new GeForce GTX 1060 comes hot on the heels of Nvidia’s high-end GTX 1080 and GTX 1070 GPUs, and seems to be aimed at the same market segment as AMD’s new Radeon RX 480. The timing of the launch, and of earlier leaks, could be a direct shot at AMD. Interestingly, the Radeon RX 480 with 4GB of RAM is priced $50 lower in the USA at $199, but AMD’s higher overheads give Nvidia a decisive advantage here in India. However, the GTX 1060’s price could make it less popular than its predecessor, the GTX 960 which sells for under Rs. 18,000.

Gamers will benefit from the same Pascal architecture that the GTX 1080 and GTX 1070 are based on, although the actual GPU is a smaller unit codenamed GP106, rather than the GP104 which the other two are based on. The GTX 1060 has 1280 programmable graphics processing units and can run at up to 1.7GHz when under full load. The 6GB of GDDR5 memory has a 192-bit pathway to the GPU, resulting in memory bandwidth of up to 8Gbps. The maximum power draw is rated at 120W, and cards featuring Nvidia’s reference cooler will have a single 6-pin PCIe connector.

nvidia_geforce_gtx1060_specs_nvidia.jpgMost notably, the GeForce GTX 1060 dispenses with SLI connectors altogether. Nvidia had de-emphasised SLI with the launch of the GTX 1080 and GTX 1070, first saying enthusiasts would need a special key to unlock it, and then limiting it to two cards down from four. Now, the company says its research shows that only a miniscule number of people ever used SLI on mid-range cards, and it believes that one GTX 1060 delivers enough performance on its own.

Nvidia will offer a Founders’ Edition version of the GTX 1060, but it has a simple plastic shroud and blower fan rather than the elaborate metal vapour-chamber cooler that gives the GTX 1080 Founders’ Edition its premium pricing. It will be available only in France, Germany and the UK, directly through Nvidia’s online store, priced at EUR 319/GBP 275. Partner companies will offer GTX 1060 cards with their own custom cooler and PCB designs, not Founders’ Editions with their own branding, as is the case with the GTX 1080 and GTX 1070.

GTX 1060 cards will have three DisplayPort 1.4 outputs as well as a single HDMI 2.0b port and one dual-link DVI-D port by default. Partners might change this arrangement around on their custom designs, with extra HDMI outputs most likely replacing DisplayPort to allow users to connect a VR headset and monitor at the same time.

GeForce_GTX_1060_Back_nvidia.jpgNvidia’s new Pascal architecture, which underpins the entire GeForce 10-series, is the company’s first on a new 16nm process, which results in greatly improved performance and lower power consumption than previous offerings. Pascal has also been designed to enable simultaneous multi-projection which allows for multiple perspectives of the same scene to be rendered in one pipeline pass, greatly improving performance for VR and multi-monitor experiences. SMP support is built into the Unreal 4 and Unity engines, and will be coming to over 30 titles including Obduction, Raw Data, Unreal Tournament, and Adr1ft.

The company also points out that the Ansel screen capture feature which debuted with the more expensive cards is also fully supported, and will debut in July with The Witcher 3: Wild Hunt and Mirror’s Edge: Catalyst.

Nvidia promises equal or greater performance from the GTX 1060 as from last year’s GeForce GTX 980, plus significant overclocking potential. Stay tuned for a full review once these cards are available in the market.

Tags: gaming, GeForce, GeForce GTX 1060, Nvidia, Nvidia Ansel, Nvidia GeForce, Nvidia GeForce GTX 1060,Nvidia GeForce GTX 1060 availability, Nvidia GeForce GTX 1060 features, Nvidia GeForce GTX 1060 launch,Nvidia GeForce GTX 1060 launch in India, Nvidia GeForce GTX 1060 performance, Nvidia GeForce GTX 1060 price,Nvidia GeForce GTX 1060 price in India, Nvidia GeForce GTX 1060 specifications, Nvidia Pascal, Pascal, VR
[“Source-Gadgets”]