IBM and OpenPOWER Members Deliver Accelerated Technologies to Break Big Data Speed Barriers

Author
SySAdmin
Posted
November 16, 2015
Views
1770

Page All:

Page 1
IBM and OpenPOWER Members Deliver Accelerated Technologies to Break Big Data Speed Barriers

Surge of Co-Developed, Accelerator-Based Solutions Sets Direction for Change in Tech Industry

AUSTIN, Texas, Nov. 16, 2015 /PRNewswire/ -- IBM (NYSE: IBM) and several fellow OpenPOWER Foundation members today revealed new technologies, collaborations and developer resources to enable clients to analyze data more deeply and with incredible speed. The new offerings center on the tight integration of IBM's open and licensable POWER processors with accelerators, dedicated high performance processors that can be optimized for computationally intensive software code.

http://photos.prnewswire.com/prnvar/20090416/IBMLOGO

The accelerated POWER-based offerings come at a time when clients are seeking the ideal server platform for Internet of Things, machine learning and other cognitive computing applications. Through an open, collaborative model, IBM and more than 90 members participating in the OpenPOWER Foundation's Accelerators Working Group are developing and delivering a wide range of accelerator-based solutions.

"There is a need for systems that provide greater speed to insight -- for data and analytics workloads to help businesses and organization make sense of the data, to outthink competitors as we usher in a new era of Cognitive Computing," said Brad McCredie, IBM Fellow and OpenPOWER Foundation President. "IBM and our fellow OpenPOWER members are on the forefront driving the changes necessary for innovation at all levels of the technology stack, including the development of the industry's first open, high-speed interconnects between processors and accelerators."

OpenPOWER Member Commitments to Accelerated Infrastructures
New products, collaborations and further investments in accelerator-based solutions on top of the POWER processor architecture include:

    --  Accelerating Watson: First made famous for beating humans on the
        Jeopardy! game show, IBM Watson's ability to "think" is transforming
        industries worldwide by enabling a new relationship between people and
        computers. IBM revealed today that the incorporation of NVIDIA(®
        )Tesla(®) K80 GPUs, the flagship offering of the NVIDIA Tesla
        Accelerated Computing Platform - coupled with Watson's POWER-based
        architecture - accelerates Watson's Retrieve and Rank API capabilities
        to 1.7x of its normal speed. This speed-up can further improve the
        cost-performance of Watson's cloud-based services. For example, a call
        center agent responding to an individual's health and insurance query
        can leverage Watson's natural language processing technology to obtain
        an answer in real-time even faster than before, potentially increasing
        customer satisfaction and lowering service costs. In addition to
        bolstering response time, the GPU acceleration also increases Watson's
        processing power to 10x its prior performance.
    --  Strategic Collaboration with Xilinx: Newest OpenPOWER Platinum member
        Xilinx and IBM today announced a multi-year strategic collaboration to
        jointly develop data center and network function virtualization (NFV)
        solutions that bring together the systems, software, and management
        components around Xilinx FPGA accelerators. The solutions focus on
        emerging workloads including high performance computing, cognitive
        computing, machine learning, genomics and big data analytics.
    --  Accelerated Networking: Mellanox announced the world's first smart
        network switch, the Switch-IB 2, capable of delivering clients an
        estimated 10x system performance improvement. NEC announced availability
        of its ExpEther Technology that is also suited for POWER
        architecture-based systems, along with plans to leverage IBM's
        innovative Coherent Accelerator Processor Interface (CAPI) technology to
        deliver additional accelerated computing value to its clients in 2016.
    --  New OpenPOWER-Based Systems: Two OpenPOWER members, E4 Computer
        Engineering and Penguin Computing, revealed new systems based on the
        OpenPOWER design concept incorporating IBM POWER8 and NVIDIA Tesla GPU
        accelerators.
    --  IBM Key Applications Ported and in Beta: IBM has ported a series of key
        IBM Internet of Things, Spark, Big Data and Cognitive Era applications
        to take advantage of the POWER architecture with accelerators.
Expansion of Developer Resources
Responding to growing demand from developers for test servers to build, port and optimize applications that can take advantage of accelerators on POWER-based systems, IBM and fellow OpenPOWER members have built a global network of physical centers and cloud-based services for no-charge access to accelerated POWER-based infrastructure. New and expanded resources include:

    --  Expanded GPU Services on SuperVessel: NVIDIA and IBM worked together to
        accelerate SuperVessel, a global cloud-based OpenPOWER ecosystem
        resource launched in June. SuperVessel now provides GPU-accelerated
        computing as-a-service capabilities, giving users access to
        high-performance NVIDIA Tesla GPUs to enable Caffe, Torch and Theano
        deep-learning frameworks to instantaneously launch from the SuperVessel
        cloud.
    --  Expanded FPGA Services on SuperVessel: Xilinx and IBM have developed a
        new FPGA accelerator service on SuperVessel that makes coherent
        reconfigurable accelerators available to developers via the cloud. By
        enabling high-level language programming like C, C++ and OpenCL, Xilinx
        and IBM are dramatically expanding how users can leverage FPGAs in the
        cloud for innovation on applications like machine learning, big data
        analytics and HPC.
    --  New Cluster at University of Texas at Austin: IBM and new OpenPOWER
        Foundation member, the Texas Advanced Computing Center (TACC) at The
        University of Texas at Austin, announced a POWER8-based accelerated
        computing cluster to be made available to academic researchers and
        developers. The new cluster is currently running successfully in an
        early user mode, and will begin accepting requests for access this week.
    --  Oregon State University Expands OSUOSL: Oregon State University's Open
        Source Lab (OSUOSL) has expanded the footprint of POWER8-based systems
        in their existing OpenStack cluster with additional compute and memory
        capacity. The expansion can significantly increase the number of
        distinct users OSUOSL can support for research and development on
        OpenPOWER/OpenStack infrastructures.
Transforming Industries with Open Acceleration
Several IBM clients understand firsthand how the coupling of accelerator technologies with POWER processors can and are delivering breakthrough performance to enable industries to transform.

    --  According to Bryan O'Neil, Director of Data Architecture, Allegiant
        Travel, "A company's data represents an immense opportunity -- and the
        speed at which you can glean insight from it is what will separate the
        winners from the losers in today's highly competitive industries. IBM
        and OpenPOWER members are delivering the kinds of acceleration
        technology our industry needs to enable real time decision making.
        Allegiant Travel is impressed with what we are seeing and eager to
        leverage accelerated solutions in our data center as we continue down
        the path to becoming a cognitive business."
    --  Last month a team of noted researchers led by Erez Lieberman Aiden, a
        geneticist and computer scientist with appointments at Baylor College,
        Baylor College of Medicine and Rice University, was nationally
        recognized for breakthrough genomics research - a new procedure designed
        to modify how a human genome is arranged in three dimensions in the
        nucleus of a cell, with extraordinary precision. Today, the team
        revealed that the feat was achieved using Power Systems accelerated with
        NVIDIA Tesla GPUs and Mellanox network infrastructure to build a 3-D map
        of the human genome and model the reaction of the genome to this
        surgical procedure, without disturbing the surrounding DNA.
    --  Coming on the heels of announcing its new supercomputer Delta based on
        the OpenPOWER server platform, Louisiana State University has released a
        white paper today documenting how an IBM POWER8-based server using
        acceleration technology rendered 7.5x to 9x performance increases over
        an Intel Xeon based LSU HPC server without acceleration technology
        ((1)).
For further details about the OpenPOWER Foundation visit http://www.openpowerfoundation.org.

             (1)    Source: "IBM POWER8(R) HPC System
                     Accelerates Genomics Analysis with
                     SMT8 Multithreading" - November 16,
                     2015. Results are based on testing
                     by Louisiana State University-
                     Center for Computational
                     Technologies (CCT) of a 40 Nodes IBM
                     Power System S824L cluster, SMT8
                     enabled, IBM Spectrum Scale Elastic
                     Storage, 512TB compared to the
                     SuperMikeII 120 node with two 8-
                     core Intel Sandy Bridge Xeon
                     processors at 2.6GHz core speed with
                     hyper-threading disabled.  Each IBM
                     POWER8 node had 256GB memory (RAM)
                     as compared to 32GB RAM in the LSU
                     SuperMikeII system. The IBM node
                     provides 16 independent, high
                     bandwidth memory controllers,
                     providing 16GB per memory
                     controller.  Results will vary
                     depending on individual workloads,
                     configurations and conditions. The
                     accelerator-based technology, the
                     CAPI/Flash Controller, is an option
                     on all IBM POWER8-based servers for
                     an additional charge.
Media Contact:
Kristin Bryson

IBM Media Relations

203-241-9190
kabryson@us.ibm.com

Logo - http://photos.prnewswire.com/prnh/20090416/IBMLOGO

SOURCE  IBM

Photo:http://photos.prnewswire.com/prnh/20090416/IBMLOGO
http://photoarchive.ap.org/
Photo:http://photos.prnewswire.com/prnh/20090416/IBMLOGO
http://photoarchive.ap.org/
IBM

Web Site: http://www.ibm.com

Title

Medium Image View Large