The Reg HPC: Mellanox gives InfiniBand a 5 BILLION PACKET/sec cloud dose [ Thu Jun 26 2014]

Dear etechnews today,

Join the conversation at The Reg HPC Community Forum

http://forums.theregister.co.uk/section/forums/hardware/hpc/

Your HPC newsletter from theregister.co.uk
for the week ending 26th June 2014


*** HPC News ***

Supercomputing speed growth hits 'historical low' in new TOP500 list
Chinese complete top-spot trifecta with Tianhe-2
http://www.theregister.co.uk/2014/06/25/growth_in_supercomputing_speeds_hits_historic_low_with_new_top500_list/

There be merriment in the Middle Kingdom: the Tianhe-2 supercomputer at
China's National University of Defense Technology is the most powerful
datacruncher on the planet (that we know about) for the third time in a
row.

The TOP500 list, published every six months, noted that the Chinese
system hit 33.86 petaflops on the Linpack benchmark. Tianhe-2 is over
ten times faster than the number 10 entrant on the list – a 3.14
petaflop Cray XC30 that's in use at an undisclosed US installation.

----

GAME ON: NVIDIA brings GPUs to 64-bit ARM servers
HPC users targeted with parallel processing code
http://www.theregister.co.uk/2014/06/24/nvidia_offers_cuda_parallel_processing_code_for_64bit_arm/

ARM's march into data centres and the world's most demanding
applications has taken a major stride forward with GPU king NVIDIA
throwing its weight behind ARM server-makers by releasing a version of
its CUDA parallel programming platform that works with the
architecture.

HPC types have, in recent years, increasingly turned to GPUs to do the
heavy lifting in some data-crunching applications. The reason for doing
so is that GPUs possess many cores, an architecture HPC types can take
advantage of by handing off small jobs to a dedicated core. With lots
of little jobs all going in parallel, more work gets done than would be
possible if a central CPU were asked to juggle as many jobs.

----

MIT boffins build 36 core processor with data-traffic smarts
Network-on-chip design uses internet-inspired scheme to solve bussing
problems
http://www.theregister.co.uk/2014/06/23/mit_boffins_build_36_core_processor_with_builtin_network_connectors/

Researchers at MIT say they have successfully built a 36-core processor
that uses an internal networking system to get maximum data throughput
from all the processing cores.

MIT's new multicore, multi-bus chip

The design, unveiled at the International Symposium on Computer
Architecture, gets around some of the problems with multicore
processors, namely bus sharing between cores, and maintaining cache
coherence.

----

Intel teams with Micron on next-gen many-core Xeon Phi with 3D DRAM
Introduces new 'fundamental building block of HPC systems' with Intel
Omni Scale Fabric
http://www.theregister.co.uk/2014/06/23/intel_knights_landing_xeon_phi/

Intel has released more details about its future "Knights Landing" Xeon
Phi many-core processor, including a new high-speed interconnect tech
called Intel Omni Scale Fabric, as well as on-package Micron Gen2
Hybrid Memory Cube (HMC) DRAM of up to 16GB.

Intel's Silvermont Atom processor microarchitecture makes a bid for HPC
(click to enlarge)

----

Mellanox gives InfiniBand a 5 BILLION PACKET/sec cloud dose
100 Gbps switch, HPC software, drinks less juice too
http://www.theregister.co.uk/2014/06/23/mellanox_spreads_infiniband_into_clouds/

Networking tech firm Mellanox is broadening its 100Gbps InfiniBand
portfolio – and is hoping that the new products and capabilities will
help it expand its footprint outside its supercomputer home territory.

Following on from its demo of 100 Gbps cables last March, the company
is using ISC 2014 in Germany to launch the Switch-IB, which has 36 100
Gbps ports, a claimed aggregate throughput of 5.4 billion packets per
second, and 130ns latency. The company's marketing VP Gilad Shainer
told The Register Mellanox will also be pushing a low-power theme for
the switch at the conference.

----

D-Wave disputes benchmark study showing sluggish quantum computer
Faulty benchmarks and sampling methods all wrong, claim Canadian
quantumoids
http://www.theregister.co.uk/2014/06/20/dwave_disputes_benchmarking_study_showing_sluggish_quantum_computer/

Quantum computing device manufacturer D-Wave is disputing a recently
published study that claims the Canadian firm's systems aren't reliably
faster than more-conventional computing systems.

On Thursday an international team of computer scientists published a
widely previewed paper in Science detailing a series of benchmark tests
pitting a 503-qubit D-Wave Two device against custom computer code
running on standard GPUs. The team reports that while the D-Wave system
was faster in some instances, it was considerably slower in others.

----

High-end storage tanked, but 'HDPA' storage about to soar says IDC
That'd be 'high performance data analysis' kit for HPC-inspired
Hadoopery
http://www.theregister.co.uk/2014/06/19/first_idc_high_performance_data_analysis_predictions/

Box-counter IDC recently decreed that 2014's first quarter was a
shocker for high-end storage sales, but also predicts things are
looking up in storage used for what it now calls "high performance data
analysis" (HPDA).

IDC's veep for HPC and data analysis Steve Conway puts HDPA in a little
perspective with a canned quote explaining that it is the kind of kit
one would buy for a high performance computer, but applied to
analytics. "Leading commercial companies in a variety of market
segments are turning to HPC technologies for challenging big data
analytics workloads that enterprise IT technology alone cannot tackle
effectively," he writes. "HPC systems can handle more complex queries,
more variables, and faster turnaround requirements."

----

India throws switch on new 282.6 Tflop super
Chemistry department cruncher boasts 15,360 Xeon E5-2760v2 cores
http://www.theregister.co.uk/2014/06/18/india_throws_switch_on_new_super/

Earlier this month, Indian university IIT Kanpur threw the switch on a
new HP-built supercomputer that puts the facility at position 130 on
the TOP500 supercomputer list.

The 282.6 Tflop, 98,304 GB of memory, 15,360-core HP machine went live
this week after being ordered in November. It's run by the university's
chemistry department. It's based on Intel Xeon E5-2760v2 cores with
Infiniband FDR interconnect.

----

Panasas: Avoid lengthy RAID re-builds - use our dodgy-file tart-up tech
Triple parity and erasure coding
http://www.theregister.co.uk/2014/06/12/lengthy_raid_rebuilds_attacked_by_panasas/

Storage biz Panasas has unveiled boosted hardware and software – and
attacked the RAID rebuild problem by rebuilding damaged files instead
of complete disks.

The company is an HPC storage system supplier which is moving into
technical computing/big data use cases in enterprises. It has more than
450 customers in 50+ countries round the globe.

----


*** Whitepaper ***

Integrated tiered storage for Big Data and HPC
Michael Feldman of Intersect360 Research writes about the market requirements for tiered storage and archiving in the context of big data and HPC.
http://whitepapers.theregister.co.uk/d/d0d/9e7f3/75a/d3c26532?td=week_sec_e



------------------------------------------------------------------------

This email was sent to garn14.tech@blogger.com

To change your email or your email subscriptions

http://account.theregister.co.uk/login/

To unsubscribe from all The Register newsletters

http://account.theregister.co.uk/unsubscribe/649203/acc978a1

The Register and its contents are Copyright © 2014 Situation Publishing.
All rights reserved.