Posts Tagged ‘NVlink’

IBM Shouldn’t Forget Its Server Platforms

April 5, 2018

The word coming out of IBM brings a steady patter about cognitive, Watson, and quantum computing, for which IBM predicted quantum would be going mainstream within five years. Most DancingDinosaur readers aren’t worrying about what’s coming in 2023 although maybe they should. They have data centers to run now and are wondering where they are going to get the system horsepower they will need to deliver IoT or Blockchain or any number of business initiatives clamoring for system resources today or tomorrow and all they’ve got are the z14 and the latest LinuxONE. As powerful as they were when first announced, do you think that will be enough tomorrow?

IBM’s latest server, the Z

Timothy Prickett Morgan, analyst at The Next Platform, apparently isn’t so sure. He writes in a recent piece how Google and the other hyperscalers need to add serious power to today’s server options. The solution involves “putting systems based on IBM’s Power9 processor into production.” This shouldn’t take anybody by surprise; almost as soon as IBM set up the Open Power consortium Rackspace, Google, and a handful of others started making noises about using Open POWER for a new type of data center server. The most recent announcements around Power9, covered here back in Feb., promise some new options with even more coming.

Writes Morgan: “Google now has seven applications that have more than 1 billion users – adding Android, Maps, Chrome, and Play to the mix – and as the company told us years ago, it is looking for any compute, storage, and networking edge that will allow it to beat Moore’s Law.” Notice that this isn’t about using POWER9 to drive down Intel’s server prices; Google faces a more important nemesis, the constraints of Moore’s Law.

Google has not been secretive about this, at least not recently. To its credit Google is making its frustrations known at appropriate industry events:  “With a technology trend slowdown and growing demand and changing demand, we have a pretty challenging situation, what we call a supply-demand gap, which means the supply on the technology side is not keeping up with this phenomenal demand growth,” explained Maire Mahony, systems hardware engineer at Google and its key representative at the OpenPower Foundation that is steering the Power ecosystem. “That makes it hard to for us to balance that curve we call performance per TCO dollar. This problem is not unique to Google. This is an industry-wide problem.” True, but the majority of data centers, even the biggest ones, don’t face looming multi-billion user performance and scalability demands.

Morgan continued: “Google has absolutely no choice but to look for every edge. The benefits of homogeneity, which have been paramount for the first decade of hyperscaling, no longer outweigh the need to have hardware that better supports the software companies like Google use in production.”

This isn’t Intel’s problem alone although it introduced a new generation of systems, dubbed Skylake, to address some of these concerns. As Morgan noted recently, “various ARM chips –especially ThunderX2 from Cavium and Centriq 2400 from Qualcomm –can boost non-X86 numbers.” So can AMD’s Epyc X86 processors. Similarly, the Open Power consortium offers an alternative in POWER9.

Morgan went on: IBM differentiated the hardware with its NVLink versions and, depending on the workload and the competition, with its most aggressive pricing and a leaner and cheaper microcode and hypervisor stack reserved for the Linux workloads that the company is chasing. IBM very much wants to sell its Power-Linux combo against Intel’s Xeon-Linux and also keep AMD’s Epyc-Linux at bay. Still, it is not apparent to Morgan how POWER9 will compete.

Success may come down to a battle of vendor ecosystems. As Morgan points out: aside from the POWER9 system that Google co-engineered with Rackspace Hosting, the most important contributions that Google has made to the OpenPower effort is to work with IBM to create the OPAL firmware, the OpenKVM hypervisor, and the OpenBMC baseboard management controller, which are all crafted to support little endian Linux, as is common on x86.

Guess this is the time wade into the endian morass. Endian refers to the byte ordering that is used, and IBM chips and a few others do them in reverse of the x86 and Arm architectures. The Power8 chip and its POWER9 follow-on support either mode, big or little endian. By making all of these changes, IBM has made the Power platform more palatable to the hyperscalers, which is why Google, Tencent, Alibaba, Uber, and PayPal all talk about how they make use of Power machinery, particularly to accelerate machine learning and generic back-end workloads. But as quickly as IBM jumped on the problem recently after letting it linger for years, it remains one more complication that must be considered. Keep that in mind when a hyperscaler like Google talks about performance per TCO dollar.

Where is all this going? Your guess is as good as any. The hyperscalers and the consortia eventually should resolve this and DancingDinosaur will keep watching. Stay tuned.

DancingDinosaur is Alan Radding, a veteran information technology analyst, writer, and ghost-writer. Follow DancingDinosaur on Twitter, @mainframeblog. See more of his work at technologywriter.com and here.

Open POWER-Open Compute-POWER9 at Open Compute Summit

March 16, 2017

Bryan Talik, President, OpenPOWER Foundation provides a detailed rundown on the action at the Open Compute  Summit held last week in Santa Clara. After weeks of writing about Cognitive, Machine Learning, Blockchain, and even quantum computing, it is a nice shift to conventional computing platforms that should still be viewed as strategic initiatives.

The OpenPOWER, Open Compute gospel was filling the air in Santa Clara.  As reported, Andy Walsh, Xilinx Director of Strategic Market Development and OpenPOWER Foundation Board member explained, “We very much support open standards and the broad innovation they foster. Open Compute and OpenPOWER are catalysts in enabling new data center capabilities in computing, storage, and networking.”

Added Adam Smith, CEO of Alpha Data:  “Open standards and communities lead to rapid innovation…We are proud to support the latest advances of OpenPOWER accelerator technology featuring Xilinx FPGAs.”

John Zannos, Canonical OpenPOWER Board Chair chimed in: For 2017, the OpenPOWER Board approved four areas of focus that include machine learning/AI, database and analytics, cloud applications and containers. The strategy for 2017 also includes plans to extend OpenPOWER’s reach worldwide and promote technical innovations at various academic labs and in industry. Finally, the group plans to open additional application-oriented workgroups to further technical solutions that benefits specific application areas.

Not surprisingly, some members even see collaboration as the key to satisfying the performance demands that the computing market craves. “The computing industry is at an inflection point between conventional processing and specialized processing,” according to Aaron Sullivan, distinguished engineer at Rackspace. “

To satisfy this shift, Rackspace and Google announced an OCP-OpenPOWER server platform last year, codenamed Zaius and Barreleye G2.  It is based on POWER9. At the OCP Summit, both companies put on a public display of the two products.

This server platform promises to improve the performance, bandwidth, and power consumption demands for emerging applications that leverage machine learning, cognitive systems, real-time analytics and big data platforms. The OCP players plan to continue their work alongside Google, OpenPOWER, OpenCAPI, and other Zaius project members.

Andy Walsh, Xilinx Director of Strategic Market Development and OpenPOWER Foundation Board member explains: “We very much support open standards and the broad innovation they foster. Open Compute and OpenPOWER are catalysts in enabling new data center capabilities in computing, storage, and networking.”

This Zaius and Barreleye G@ server platforms promise to advance the performance, bandwidth and power consumption demands for emerging applications that leverage the latest advanced technologies. These latest technologies are none other than the strategic imperatives–cognitive, machine learning, real-time analytics–IBM has been repeating like a mantra for months.

Open Compute Projects also were displayed at the Summit. Specifically, as reported: Google and Rackspace, published the Zaius specification to Open Compute in October 2016, and had engineers to explain the specification process and to give attendees a starting point for their own server design.

Other Open Compute members, reportedly, also were there. Inventec showed a POWER9 OpenPOWER server based on the Zaius server specification. Mellanox showcased ConnectX-5, its next generation networking adaptor that features 100Gb/s Infiniband and Ethernet. This adaptor supports PCIe Gen4 and CAPI2.0, providing a higher performance and a coherent connection to the POWER9 processor vs. PCIe Gen3.

Others, reported by Talik, included Wistron and E4 Computing, which showcased their newly announced OCP-form factor POWER8 server. Featuring two POWER8 processors, four NVIDIA Tesla P100 GPUs with the NVLink interconnect, and liquid cooling, the new platform represents an ideal OCP-compliant HPC system.

Talik also reported IBM, Xilinx, and Alpha Data showed their line ups of several FPGA adaptors designed for both POWER8 and POWER9. Featuring PCIe Gen3, CAPI1.0 for POWER8 and PCIe Gen4, CAPI2.0 and 25G/s CAPI3.0 for POWER9 these new FPGAs bring acceleration to a whole new level. OpenPOWER member engineers were on-hand to provide information regarding the CAPI SNAP developer and programming framework as well as OpenCAPI.

Not to be left out, Talik reported that IBM showcased products it previously tested and demonstrated: POWER8-based OCP and OpenPOWER Barreleye servers running IBM’s Spectrum Scale software, a full-featured global parallel file system with roots in HPC and now widely adopted in commercial enterprises across all industries for data management at petabyte scale.  Guess compute platform isn’t quite the dirty phrase IBM has been implying for months.

DancingDinosaur is Alan Radding, a veteran information technology analyst, writer, and ghost-writer. Please follow DancingDinosaur on Twitter, @mainframeblog. See more of his IT writing at technologywriter.com and here.

 

z System-Power-Storage Still Live at IBM

January 5, 2017

A mid-December briefing by Tom Rosamilia, SVP, IBM Systems, reassured some that IBM wasn’t putting its systems and platforms on the backburner after racking up financial quarterly losses for years. Expect new IBM systems in 2017. A few days later IBM announced that Japan-based APLUS Co., Ltd., which operates credit card and settlement service businesses, selected IBM LinuxONE as its mission-critical system for credit card payment processing. Hooray!

linuxone-emperor-2

LinuxONE’s security and industry-leading performance will ensure APLUS achieves its operational objectives as online commerce heats up and companies rely on cloud applications to draw and retain customers. Especially in Japan, where online and mobile shopping has become increasingly popular, the use of credit cards has grown, with more than 66 percent of consumers choosing that method for conducting online transactions. And with 80 percent enterprise hybrid cloud adoption predicted by 2017, APLUS is well positioned to connect cloud transactions leveraging LinuxONE. Throw in IBM’s expansion of blockchain capabilities and the APLUS move looks even smarter.

With the growth of international visitors spending money, IBM notes, and the emergence of FinTech firms in Japan have led to a diversification of payment methods the local financial industry struggles to respond. APLUS, which issues well-known credit cards such as T Card Plus, plans to offer leading-edge financial services by merging groups to achieve lean operations and improved productivity and efficiency. Choosing to update its credit card payment system with LinuxONE infrastructure, APLUS will benefit from an advanced IT environment to support its business growth by helping provide near-constant uptime. In addition to updating its server architecture, APLUS has deployed IBM storage to manage mission-critical data, the IBM DS8880 mainframe-attached storage that delivers integration with IBM z Systems and LinuxONE environments.

LinuxONE, however, was one part of the IBM Systems story Rosamilia set out to tell.  There also is the z13s, for encrypted hybrid clouds and the z/OS platform for Apache Spark data analytics and even more secure cloud services via blockchain on LinuxONE, by way of Bluemix or on premises.

z/OS will get attention in 2017 too. “z/OS is the best damn OLTP system in the world,” declared Rosamilia. He went on to imply that enhancements and upgrades to key z systems were coming in 2017, especially CICS, IMS, and a new release of DB2. Watch for new announcements coming soon as IBM tries to push z platform performance and capacity for z/OS and OLTP.

Rosamilia also talked up the POWER story. Specifically, Google and Rackspace have been developing OpenPOWER systems for the Open Compute Project.  New POWER LC servers running POWER8 and the NVIDIA NVLink accelerator, more innovations through the OpenCAPI Consortium, and the team of IBM and Nvidia to deliver PowerAI, part of IBM’s cognitive efforts.

As much as Rosamilia may have wanted to talk about platforms and systems IBM continues to avoid using terms like systems and platforms. So Rosamilia’s real intent was to discuss z and Power in conjunction with IBM’s strategic initiatives.  Remember these: cloud, big data, mobile, analytics. Lately, it seems, those initiatives have been culled down to cloud, hybrid cloud, and cognitive systems.

IBM’s current message is that IT innovation no longer comes from just the processor. Instead, it comes through scaling performance by workload and sustaining leadership through ecosystem partnerships.  We’ve already seen some of the fruits of that innovation through the Power community. Would be nice to see some of that coming to the z too, maybe through the open mainframe project. But that isn’t about z/0S. Any boost in CICS, DB2, and IMS will have to come from the core z team. The open mainframe project is about Linux on z.

The first glimpse we had of this came last spring in a system dubbed Minsky, which was described back then by commentator Timothy Prickett Morgan. With the Minsky machine, IBM is using NVLink ports on the updated Power8 CPU, which was shown in April at the OpenPower Summit and is making its debut in systems actually manufactured by ODM Wistron and rebadged, sold, and supported by IBM. The NVLink ports are bundled up in a quad to deliver 80 GB/sec bandwidth between a pair of GPUs and between each GPU and the updated Power8 CPU.

The IBM version, Morgan describes, aims to create a very brawny node with very tight coupling of GPUs and CPUs so they can better share memory, have fewer overall GPUs, and more bandwidth between the compute elements. IBM is aiming Minsky at HPC workloads, according to Morgan, but there is no reason it cannot be used for deep learning or even accelerated databases.

Is this where today’s z data center managers want to go?  No one is likely to spurn more performance, especially if it is accompanied with a price/performance improvement.  Whether rank-and-file z data centers are queueing up for AI or cognitive workloads will have to be seen. The sheer volume and scale of expected activity, however, will require some form of automated intelligent assist.

DancingDinosaur is Alan Radding, a veteran information technology analyst, writer, and ghost-writer. Please follow DancingDinosaur on Twitter, @mainframeblog. See more of his IT writing at technologywriter.com and here

IBM Fires a Shot at Intel with its Latest POWER Roadmap

June 17, 2016

In case you worry that IBM will abandon hardware in the pursuit of its strategic initiatives focusing on cloud, mobile, analytics and more; well, stop worrying. With the announcement of its POWER Roadmap at the OpenPOWER Summit earlier this spring, it appears POWER will be around for years to come. But IBM is not abandoning the strategic initiatives either; the new Roadmap promises to support new types of workloads, such as real time analytics, Linux, hyperscale data centers, and more along with support for the current POWER workloads.

power9b

Pictured above: POWER9 Architecture, courtesy of IBM

Specifically, IBM is offering a denser roadmap, not tied to technology and not even tied solely to IBM. It draws on innovations from a handful of the members of the Open POWER Foundation as well as support from Google. The new roadmap also signals IBM’s intention to make a serious run at Intel’s near monopoly on enterprise server processors by offering comparable or better price, performance, and features.

Google, for example, reports porting many of its popular web services to run on Power systems; its toolchain has been updated to output code for x86, ARM, or Power architectures with the flip of a configuration flag. Google, which strives to be everything to everybody, now has a highly viable alternative to Intel in terms of performance and price with POWER. At the OpenPOWER Summit early in the spring, Google made it clear it plans to build scale-out server solutions based on OpenPower.

Don’t even think, however, that Google is abandoning Intel. The majority of its systems are Intel-oriented. Still, POWER and the OpenPOWER community will provide a directly competitive processing alternative.  To underscore the situation Google and Rackspace announced they were working together on Power9 server blueprints for the Open Compute Project, designs that reportedly are compatible with the 48V Open Compute racks Google and Facebook, another hyperscale data center, already are working on.

Google represents another proof point that OpenPOWER is ready for hyperscale data centers. DancingDinosaur, however, really is interested most in what is coming from OpenPOWER that is new and sexy for enterprise data centers, since most DancingDinosaur readers are focused on the enterprise data center. Of course, they still need ever better performance and scalability too. In that regard OpenPOWER has much for them in the works.

For starters, POWER8 is currently delivered as a 12-core, 22nm processor. POWER9, expected in 2017, will be delivered as 14nm processor with 24 cores and CAPI and NVlink accelerators. That is sure to deliver more performance with greater energy efficiency.  By 2018, the IBM roadmap shows POWER8/9 as a 10nm, maybe even 7nm, processor, based on the existing micro-architecture.

The real POWER future, arriving around 2020, will feature a new micro-architecture, sport new features and functions, and bring new technology. Expect much, if not almost all, of the new functions to come from various OpenPOWER Foundation partners,

POWER9, only a year or so out, promises a wealth of improvements in speeds and feeds. Although intended to serve the traditional Power Server market, it also is expanding its analytics capabilities and bringing new deployment models for hyperscale, cloud, and technical computing through scale out deployment. This will include deployment in both clustered or multiple formats. It will feature a shorter pipeline, improved branch execution, and low latency on the die cache as well as PCI gen 4.

Expect a 3x bandwidth improvement with POWER9 over POWER8 and a 33% speed increase. POWER9 also will continue to speed hardware acceleration and support next gen NVlink, improved coherency, enhance CAPI, and introduce a 25 GPS high speed link. Although the 2-socket chip will remain, IBM suggests larger socket counts are coming. It will need that to compete with Intel.

As a data center manager, will a POWER9 machine change your data center dynamics?  Maybe, you decide: a dual-socket Power9 server with 32 DDR4 memory slots, two NVlink slots, three PCIe gen-4 x16 slots, and a total 44 core count. That’s a lot of computing power in one rack.

Now IBM just has to crank out similar advances for the next z System (a z14 maybe?) through the Open Mainframe Project.

DancingDinosaur is Alan Radding, a veteran information technology analyst and writer. Please follow DancingDinosaur on Twitter, @mainframeblog. See more of his IT writing at technologywriter.com and here.

 


%d bloggers like this: