Intel slips details of Poulson-based Itanium 9500 in advance, teases a big boost to 64-bit servers

Intel slips out Poulsonbased Itanium 9500 details in advance, tease a big boost to 64bit servers

If you think Intel took awhile to roll out the Xeon E5, imagine the mindset of Itanium server operators — they haven’t had any kind of update to the IA-64 chip design since February 2010, and they’re still waiting. Much to their relief, Intel just dropped a big hint that the next-generation, Poulson-based Itanium is getting close. Both a reference manual and a Product Change Notification have signaled that the new, 32-nanometer part will get the Itanium 9500 name as well as a heap of extra improvements that haven’t been detailed until now. We knew of the eight processing cores, but the inadvertent revelation also confirms about a 50 percent hike in the interconnect speed and a matching increase in the cache size to 32MB. Clock speeds also start where current Tukwila-running Itaniums stop, with four processors between 1.73GHz and 2.53GHz giving the line a much-needed shot of adrenaline. Few of us end users will ever directly benefit when Poulson ships to company server farms later this year; after these increases, though, don’t be shocked when the database at work is suddenly much quicker on its toes.

Filed under:

Intel slips details of Poulson-based Itanium 9500 in advance, teases a big boost to 64-bit servers originally appeared on Engadget on Thu, 19 Jul 2012 01:33:00 EDT. Please see our terms for use of feeds.

Permalink The Register  |  sourceIntel (ZIP), CPU World  | Email this | Comments

Quad GeForce GTX 690 server scoffs at your parallel processing needs

NVIDIA’s GeForce GTX 690 has already won the hearts and minds of many gamers, with its potent combination of twin Kepler cores, but how about using it for a compact GPU compute rig? That’s just what custom PC system maker AVADirect decided to try, opting for not just one GTX 690 but a four card rig squeezed into a standard 2U server.

That’s a total of eight Kepler cores all running in parallel. AVADirect hasn’t shared the rest of the specifications of the server itself, nor benchmarks – which is, we must admit, what we’re particularly keen to see.

Why would you want four high-power graphics card in a server? Well, while gaming probably isn’t high on the agenda, turning NVIDIA’s CUDA cores into a parallel processing workstation could have some significant benefits for anyone doing graphics or 3D rendering, or crunching huge quantities of mathmatical data.

NVIDIA normally pushes its Quadro or Tesla cards for dedicated parallel processing tasks, but there’s no reason the eminently capable GTX 690 – which has 3,072 CUDA cores apiece – shouldn’t turn its hand to something more serious than Crysis. No word on overall system pricing, but with each of those EVGA GeForce GTX 690 4GB cards coming in at nearly $1,100 you’re looking at almost $4,400 for CUDA cores alone.


Quad GeForce GTX 690 server scoffs at your parallel processing needs is written by Chris Davies & originally posted on SlashGear.
© 2005 – 2012, SlashGear. All right reserved.


Windows Server 2012 set for just four iterations

This time around, Microsoft is not about to get messy with the massive amount of version of their IT-aimed Windows Server software they’ve had in the past. Where back in 2008, Window Server had twelve, count them, twelve versions of Server 2008 R2, this version focusing on just four different groups which its editions will be ideal for. These versions include Datacenter, Standard, Essentials, and Foundation, and will be spattered across businesses across the IT landscape before the year is done and over with – will you be jumping aboard?

This set of versions starts with Foundation, that being an economical general purpose server aimed bit of software with pricing only revealed to manufacturers of the devices it’ll be pushed to. Next you’ve got an Essentials version of Windows Server, this edition made for small business environments and costing $425. Pricing on each of these editions, mind you, represents Open No Level (NL) ERP.

The Standard edition of Windows Server 2012 is aimed at low density or non-virtualized environments and is one of two editions to be made for Processor + CAL licensing models, here priced at $882 USD. The other of these two editions is the Datacenter edition which is aimed at highly virtualized private and hybrid cloud environments. The Datacenter edition has unlimited virtual instances and will cost you $4,809.

You can also try the software out before you drop hundreds or thousands of dollars for your business right over at the Windows Server 2012 Release Candidate site right this second!

Check out the timeline below to dive deep into more Windows Server action and let us know if your business will be upgrading this year!


Windows Server 2012 set for just four iterations is written by Chris Burns & originally posted on SlashGear.
© 2005 – 2012, SlashGear. All right reserved.


Samsung Samples Industry’s First 16-Gigabyte Server Modules Based on DDR4 Memory technology

Samsung today announced that it has begun sampling the industry’s first 16-gigabyte (GB) double data rate-4 (DDR4), registered dual inline memory modules (RDIMMs), designed for use in enterprise server systems.
Using 30nm-class process technology, Samsung sampled new 8GB and 16GB DDR4 modules in June, in addition to providing them to major CPU and controller makers. The modules will bring the highest density and performance levels to premium enterprise server systems. Samsung previously …

Users reporting BBM down, joins Instagram in being less social (update: RIM’s on it)

Users reporting BBM down, joins Instagram in being less social

First Netflix, then Instagram, now BBM? We’re getting reports along with plenty of Twitter noise that the popular BlackBerry service has headed south. This makes for one heck of an anti-social weekend if you’re a photo-sharing, BBM-er for sure. Let us know if you’ve been affected by the service, in the meantime why not read up on some tech news? We’ve plenty!

Update: We’ve just received word from RIM itself explaining the situation. Basically, you might experience some hiccups with BBM and web browsing, but everything else should be back to normal. Check out the response below:

Today, we experienced a service issue which may have affected some of our customers in certain parts of Europe, the Middle East and Africa. Some minor issues relating to browsing and BlackBerry Messenger may still persist, but all other services are operating as normal. We apologize to any customers in these regions who may have been inconvenienced.

Users reporting BBM down, joins Instagram in being less social (update: RIM’s on it) originally appeared on Engadget on Sat, 30 Jun 2012 09:54:00 EDT. Please see our terms for use of feeds.

Permalink   |  sourceTwitter  | Email this | Comments

Google Compute Engine challenges Amazon with thousands of cores

Google has launched Google Compute Engine, an Amazon EC2 rivaling infrastructure-as-a-service that “just works” for scalable virtual machines. App Engine powers over 1m active applications each day, Urs Holze, Google Senior VP of infrastructure, said during the Google IO keynote today, and now developers will be able to harness the power of the search giant’s server farm themselves.

App Engine currently sees 7.5bn hits per day, and two trillion DataStore operations per month. Google Compute Engine throws in the scale and experience Google has built up over the years, together with the connectivity of the company’s mobile backbone.

Google’s primary example of the power of Compute Engine was a DNA crunching Institute for Systems Biology program. The center had been using a 1,000 core cluster that produced a single result every ten minutes; with Google Compute Engine’s 10,000 cores, it punched out a result every few seconds. With 600,000 cores, there were multiple results every second.

As for pricing, Google says Compute Engine will offer up to 50-percent more processing, per dollar, than the company’s rivals. Access is in limited availability from today, with up to 10,000 cores for systems that have heavy I/O requirements, while those which have lighter I/O needs can take advantage of hundreds of thousands of cores – Google added 771,886 alone during the presentation.


Google Compute Engine challenges Amazon with thousands of cores is written by Chris Davies & originally posted on SlashGear.
© 2005 – 2012, SlashGear. All right reserved.


Google Compute Engine brings Linux virtual machines ‘at Google scale’

As anticipated, Google has just launched its cloud service for businesses at Google I/O 2012, called Google Compute Engine. Starting today Urs Holzle announced “anyone with large-scale computing needs” can access the infrastructure and efficiency of Google’s datacenters. The company is promising both performance and stability — Amazon EC2 they’re coming for you — claiming “this is how infrastructure as a service is supposed to work”. It’s also promising “50 percent more computes per dollar” than competitors. Beta testers will be on hand at later meetings to give impressions of the service, if you want to know how running your apps on 700,000 (and counting) cores feels. During the presentation we got a demo of a genome app and we’re sure if we understood what was going on, it would have been impressive. Hit the source links below for more details on “computing without limits” or to sign up for a test yourself.

Check the live blog for more details as they’re revealed.

Check out our full coverage of Google I/O 2012’s developer conference at our event hub!

Google Compute Engine brings Linux virtual machines ‘at Google scale’ originally appeared on Engadget on Thu, 28 Jun 2012 13:47:00 EDT. Please see our terms for use of feeds.

Permalink   |  sourceGoogle Developers Blog, Google Compute Engine  | Email this | Comments