HPC > More stories

NASA Nature Run screen grab

NASA launches new climate model at SC14

After 75 days running on the 43,000-core, 1.21 petaflop Discover supercomputer cluster at the Goddard Centre for Climate Simulation, producing 4 PB of data, NASA has put together a global model showing how atmospheric carbon dioxide – CO2 – gets distributed in air currents around the world. Youtube Video The model is based on …
Nvidia

Nvidia doubles Tesla grunt at SC14

Nvidia's SC14 eye-catcher is the next increment in its HPC GPU accelerator, the Tesla K80. The successor to the Tesla K40, the K80 is pitched as double-your-everything: twice the performance and twice the memory bandwidth. Unsurprisingly, the company reckons its target will be data analytics and scientific computing applications …

The optic NERVE of it: Intel declares WAR on InfiniBand

It's no surprise when Intel announces a new processor architecture – that's its main game, after all. However, along with the Knight's Bridge chips released at SC'14, the company seems to have declared war on the HPC mainstream, announcing an optical comms architecture designed to compete with InfiniBand. The Omni-Path …
USGI UV 2000

HPC bod SGI racks UV brains, reaches 30 MEEELLION IOPS

How does 30 million IOPS from a 32-socket server grab you? We've learnt SGI has taken its UV 300H server, used in its "UV for SAP HANA" product, and boosted its performance with 64 Intel DC P3700 SSDs, which use the NVMe interface. The team working on this includes SGI's acquired Starboard Storage engineers. Starboard was …
Chris Mellor, 17 Nov 2014
Cray Sonexion

Cray heaves out even mightier, Lustre-ous Sonexion 2000

Cray has updated its scale-out, parallel access, Lustre-using Sonexion array with what it calls the "2000" product - providing 50 per cent more capacity and performance than the previous 1600 product in the same footprint. The 2000 delivers in excess of 45GB/sec bandwidth to a Linux cluster using Cray Cluster Connect, and has …
Chris Mellor, 17 Nov 2014

SAVE ME, NASA system builder, from my DEAD WORKSTATION

HPC Blog As I walked past one of the booths on the GTC’14 trade show floor, I suddenly had the nagging feeling that I had missed something. I couldn't figure out why, but this particular booth was familiar. On closer examination, I still didn’t see anything to explain my feeling. I didn’t know either of the guys manning the booth, and …

Cry Havoc and let slip the dogs of Patent WAR! Samsung strikes back at Nvidia

Samsung Electronics has counter-sued graphics processing units (GPU) manufacturer Nvidia for alleged infringement of its chip-related patents. Nvidia filed a patent lawsuit against Sammy and semiconductor and wireless tech giant Qualcomm in September, alleging the companies used its patented GPU technology without proper …
Kat Hall, 12 Nov 2014
Image of HAL eye from 2001 movie with Chrome logo in eye

729 teraflops, 71,000-core Super cost just US$5,500 to build

Cycle Computing has helped hard drive giant Western Digital shove a month's worth of simulations into eight hours on Amazon cores. The simulation workload was non-trivial: to check out new hard drive head designs, the company runs a million simulations, each of which involved a sweep of 22 head design parameters on three types …
cubicle_farm_computers_channel

Converge THIS: Network titan Cisco on how to VDI in 10 days

HPC Blog Thinking about virtualising your desktops? Want to run all those sessions in your data centre and simply broadcast the screen and UI out to whatever device your users desire? There are a lot of vendors out there who are looking to ease your shift into VDI (virtualized desktop infrastructure), helping to figure out how much …
resellers_computers_desktops_workstations

HPC ace grills vendor over virtual desktop flavours: What about the POWER USERS?

HPC blog The idea that you can run a bunch of user desktop sessions on servers isn’t new. The benefits promised by desktop virtualisation are pretty profound, including large reductions in enterprise tech costs, better support for users, improved security and even increased user happiness. That last promise, the increased user happiness …

Pimp my HPC cluster: Luscious cores, GPU gravy, optimised apps...

HPC blog You’ve pimped out your fancy new HPC cluster with the latest technology, each node jammed to the gills with shiny new CPUs, loads of memory, and maybe even a double brace of GPUs or other accelerators. The installation work is done, and users are getting their first shot at the system. The performance boost on the new box is so …
ETERNUS_CD10000

Fujitsu waves 50PB+ MONSTER at hyperscale storage freaks

Fujitsu has got itself a 50PB-plus scale-out array for very big data, the CD10000. Properly named the ETERNUS CD10000 Hyperscale Storage System, the box uses Ceph, open source software that presents file, block and object storage from a distributed cluster of object storage nodes across which data is striped. Theoretically, Ceph …
Chris Mellor, 5 Nov 2014
ClusterStor 1500

Why Seagate spent $374m on Xyratex: It's part of a cloud system

Comment Seagate has recruited an EVP for its cloud business and shone a light on where that business is going. Former Cisco bod Sri Hosakote joins Jamie Lerner's Cloud Systems and Solutions (CSS) group as an EVP and reports to Lerner, who joined in February this year. Both men are ex-Cisco, so Lerner has recruited a known quantity. …
Chris Mellor, 3 Nov 2014
Kinetic Terascale

WHOMP, there it is: Seagate demos Kinetic disk drive

Seagate is demonstrating its Kinetic disk drive at the OpenStack summit in Paris, claiming to have broad support for its proprietary, single-source technology. Kinetic is a disk drive directly addressed over Ethernet using Get and Put-style object storage commands using an open source API. The idea is that applications can …
Chris Mellor, 3 Nov 2014

IBM, backing away from hardware? NEVER!

HPC Blog So, it looks like IBM has moved its chipworks over to Globalfoundries, as described by Reg hack Tim Worstall here. But what are the implications for IBM's systems business? I agree that the brutal and implacable economics of the chip business forced IBM’s hand here. Manufacturing microprocessors is a hugely capital intensive …
Bored girl

Cray-cray Met Office spaffs £97m on VERY AVERAGE HPC box

The UK's Met Office has settled on Cray as the vendor for its next supercomputer, with a 16 petaflop XC40 machine* to be shared between the Met Office and Exeter Science Park. The £97m HPC box will give the Met Office 13 times more supercomputing muscle than it is currently able to flex. The weathermen want to improve the …

It's Big, it's Blue... it's simply FABLESS! IBM's chip-free future

Worstall on Wednesday IBM has coughed up $1.5bn to get Global Foundries to take its chip manufacturing arm off its hands as reported here at El Reg. There are a number of economic implications to all of this, perhaps the most important of which is why the eco-lunatic fantasy that we could reverse globalisation is simply nonsense (of course, whether …
Tim Worstall, 22 Oct 2014

HPC Wales stretches its reach to Northern Ireland

Northern Ireland is to get what's touted as its first HPC service, courtesy of a deal between Fujitsu and HPC Wales. Extending a collaboration that's been in place since 2012, HPC Wales says the vendor will sling £1.1 million towards the service, and seek collaborations from Irish business and academic partners. HPC Wales has …

Vote for us: We're slinging cash at supercomputers!

The government of the Australian State of Victoria has slung AU$6.65 million in the direction of the Victorian Life Sciences Computation Initiative (VLCSI), deeming the investment a vote-winner in upcoming elections . The VLCSI, one of Australia's leading HPC facilities, is operated in collaboration with IBM and serves campuses …
Archimedes

'Urika': Cray unveils new 1,500-core big data crunching monster

Big data analytics people are constantly panning for nuggets of gold, and Cray has just the machine for them — its Urika-XA. Said to be a single-platform entity, consolidating a wide range of analytics workloads previously needing separate systems, its design has been optimised for compute- and memory-intensive and latency- …
Chris Mellor, 16 Oct 2014

IBM goes gunning for Intel with Nvidia GPU-charged Power8 servers

Now that it's officially out of the x86 server biz, IBM is hitting the gas on its Power processor strategy with new Power8-based iron that incorporates GPU acceleration tech from Nvidia for the first time. The IBM Power S824L is both the first Power server to include Nvidia Tesla GPUs and the first product to be born out of the …
DSolar's PV concentrator concept

Supercomputer water-cooling comes to solar power

Solar power outfit Airlight Energy has borrowed technology from the world of supercomputers to make its large-scale photovoltaic collectors more efficient, on the way heating water to get a second source of energy. As explained by the IEEE, the Swiss company has created a spin-off called Dsolar to develop its dish-style large- …
DataDirect Networks GS2K appliance

DataDirect Networks stuffs tricky GPFS file system into GS7K box

DataDirect Networks has bundled IBM's GPFS parallel file system, which uses GridScaler technology, into its GS7K appliance. The appliance can use its WOS object storage as a huge back-end vault, which DDN hopes will remove GPFS set-up complexity while preserving standardised NAS access. GPFS systems are known to be powerful but …
Chris Mellor, 1 Oct 2014

AMD, Oracle sweep their glasses off the table, ARM wrestle Hadoop for the data center

JavaOne AMD thinks the future of the data center lies in the ARM processor architecture, and it's betting that servers based on 64-bit ARMv8 will be humming in racks near you sooner than you expect. To prove it, AMD's Leendert van Doorn took to the stage at Oracle's JavaOne conference in San Francisco to demo two ARM Cortex A57–based …

Dude, you're getting a Dell! Uni of Queensland adds AU$275,000 cluster

The University of Queensland is the latest contributor to Australia's academic high-performance computing boom, with a AU$275,000 ($238k) cluster going live late in September. Installed at the University's Australian Institute for Bioengineering and Nanotechnology (AIBN), the new cluster has nine Dell PowerEdge R620s. Each runs …
Cray_XC40

Supercomputers: The Next Generation – Cray puts burst buffer tech, Intel Haswell inside

Cray has new XC40 and CS400 superduper computers using Haswell processors and DataWarp burst buffer tech to keep the Haswell cores crammed with data to process. The XC40 goes twice the speed of the existing XC30, courtesy of its Intel Xeon E5-2600 v3 ("Haswell") processor, scaling past a million cores. The architecture …
Chris Mellor, 30 Sep 2014
Larry Ellison

Ellison: Sparc M7 is Oracle's most important silicon EVER

OpenWorld During his OpenWorld keynote on Sunday, Oracle CEO CTO Larry Ellison took time out from talking up his company's cloud strategy to remind the audience that the database giant is in the hardware business, too – all the way down to the silicon. Many of Oracle's "engineered systems" are powered by Intel processors – and Intel …
Neil McAllister, 29 Sep 2014
HP

ARMs head Moonshot bodies: HP pops Applied Micro, TI chips into carts

HP taking another shot at its Project Moonshot by today announcing two new ARM-powered servers, one 64-bit and the other 32-bit. The US giant hopes its two new ProLiant models will live up to the hype: for years now the ARM world has claimed the architecture will provide high-density, power efficient systems for data centers. …
Pawsey's Magnus super

Pawsey powerhouse schedules first projects

Pawsey's petascale powerhouse has gone through its commissioning cycles, and its operators have announced that they've granted 90 million CPU-hours to the machine's first research users for the period September to December. Magnus, the eight-cabinet, 35,000-core Cray XC30, is based on Intel's Xeon E5-2600 v3 processors and has …
Appro LLNL Edge Cluster

Enough with the rack-grappling, kids. Try this MYSTERY app

HPC blog The organisers of the ISC Student Cluster Competitions (here's a quick guide on how these HPC battles work) are a wily bunch. Theirs was the first competition to throw a “mystery application” at the student teams, forcing them to optimise an unfamiliar workload on the fly. The teams only have a few hours to learn the mystery app …

Rookie southies S Korea and S Africa prep for cluster-wrestling match

HPC Blog Rounding out the field for the recent ISC14 Student Cluster Competition in Leipzig are two semi-experienced teams that both have “South” in their names. Even though Team South Africa and Team South Korea are separated by a large amount of land and water, these two teams do have some things in common. Both of these teams have …

Cherry Creek overflows with HPC goodness at Uni of Nevada

The University of Nevada, Las Vegas, has signed a deal giving it access to a year-old TOP500 and Green500 supercomputer, the Intel-built Cherry Creek which has arrived at the city's Switch Supernap. In a partnership with Switch, the university says it will be offering the machine for private sector research as well as its own …
IBM's System x3630 M3 server

Students playing with impressive racks? Yes, it's cluster comp time

HPC blog This article and the ones following it are the most comprehensive and in-depth analysis of a cluster competition that the world has ever seen. If you can find better coverage, then I'll eat a handful of spider webs. Let's dive into the results from the benchmark and HPC application portion of the ISC14 Student Cluster …
flag.Scotland

Scotland wins WORLD RECORD as voters head to referendum polls

HPC blog The LINPACK portion of the ISC’14 Student Cluster Competition (LINK) was supposed to be routine, according to the cluster competition wise guys. Sure, some student team might set a new record, but no one was expecting the new mark to break through the 10 TFLOP/s barrier. Almost everyone expected the LINPACK crown to go to one of …

A3Cube turns RAM up to 11 with FORTISSIMO kit

A3Cube has fleshed out another part of its plans for high performance interconnect, adding an I/O access system to the RONNIE data plane technology it launched in February 2014. As explained to El Reg when it hit, RONNIE provides direct, switchless connection between nodes using PCIe NICs, claiming around 900 nanosecond latency …

Sweden orders TWO PETAFLOP supercomputer

Sweden's KTH (Royal Institute of Technology) is continuing its Cray fandom, placing a US$13 million order for an XC super and taking the institute into the petascale era. Cray first got its foot in the door at KTH in 2010, with a 93 teraflop-rated XT6m midrange system. The XT6m was replaced with a 305 TF XE6, which went in …
Seagate hybrid drives

Seagate's BUMPER State of the Storage Nation announcement

Analysis Positioning itself as a data steward, Seagate has presented its storage state of the nation pitch, announcing: three new disk drives; LaCie Thunderbolt external storage; two server flash cards and a controller; an HPC array; a new backup appliance; and a bright and shiny cloud and enterprise business unit. Seagate is continuing …
Chris Mellor, 15 Sep 2014

eBay DROPS DEAD AGAIN - tat bazaar says sorry, scrambles to resurrect site

eBay went titsup earlier today, and the company is now attempting to bring its site back to life. The online tat bazaar coughed to an unexplained technical blunder preventing an unknown number of its subscribers from accessing the site, which many buyers and sellers of used goods enjoy using in their spare time on the weekends …
Kelly Fiveash, 14 Sep 2014
Intel Xeon E5-2600 package

Intel's DDR4-friendly Xeon workhorses bolt for workstations, servers

Intel's latest-generation Xeon E5 v3 processors first showed up in systems from the likes of Dell last month, but Chipzilla made them generally available on Monday – with all of 32 different parts heading for OEMs and the channel. The new Xeon E5-2600 v3 and E5-1600 v3 chips are all based on Intel's x86-64 Haswell …
"Image from the glassbrain project, neuroscapelab.com, UCSF

Nvidia builds CUDA GPU programming library for machine learning – so you don't have to

Nvidia has released a set of software routines for accelerating machine-learning algorithms on its parallelized graphics processors. Over the weekend, the GPU maker uploaded cuDNN – CUDA Deep Neural Networks – which is a library of primitives for building software that trains neural networks. The component is optimized for …
Gavin Clarke, 8 Sep 2014

HAMR time for Google's MapReduce, says not-so-startup

Like the idea of chewing on terabytes data using Google’s MapReduce but think it's too slow, too hardware-hungry and too complicated? A fledgling big-data analytics venture reckons it’s got the answer - a Hadoop programming framework built using Java it claims is 20 times faster than using ordinary Hadoop and that it claims uses …
Gavin Clarke, 8 Sep 2014
ASKAP from the air

Square Kilometre Array reveals its 1.6TB-a-day storage and network rigs

Australia's first pass at the Square Kilometre Array – the Boolardy Engineering Test Array – is about to get commissioned into a fully-live system. The test array, known naturally enough as BETA, is part of the science-before-the-science: a proving ground for some of the new technologies being used for the SKA project, in …