Savtira streams media and apps from the cloud with beefy PowerEdge C combo

April 18, 2011

Savtira Corporation, who provides outsourced Cloud Commerce solutions, has chosen Dell DCS’s PowerEdge C line of servers and solutions to deliver streamed media and apps from the cloud.  Dell’s gear will help power the Savtira Cloud Commerce platform and Entertainment Distribution Network (EDN).

With a little help from PowerEdge C, businesses will now be able to use EDN to stream all digital media (business apps, games, music, movies audio/ebooks) from the cloud to any device.  One of the particularly cool features is, since the state and configuration are cloud-based, consumers can switch between devices and pick up exactly where they pushed pause on the last device.

Talk about supercharging

To power Savtira’s EDN data center, the company picked PowerEdge C410xs packed with NVidia Tesla M2070 GPUs and driven by PowerEdge C6145s.  If you think GPUs are just for rendering first-person shooters, think again.  GPUs can also cost-effectively supercharge your compute-intensive solution by offloading a lot of the processing from the main CPUs.  According to NVidia, for 1/10 the cost and with only 1/20 of the power consumption, GPUs deliver the same performance as CPUs.

To  help you get an idea of the muscle behind this solution, the PowerEdge C410x PCIe expansion chassis holds up to 16 of the Tesla M2070s GPUs, each of which exceeds over 400 cores.  Two fully populated C410xs are in turn powered by one PowerEdge C6145 for a combined total of 33 Teraflops in just 7U.

Talk about a lot of power in a little space 🙂

Extra-credit reading

  • PowerEdge C6145 — Dell DCS unveils its 4th HPC offering in 12 months, and its a beefy one
  • PowerEdge C410x — Say hello to my little friend — packing up to 16 GPGPUs
  • NVIDIA: from gaming graphics to High Performance Computing

Pau for now…

1 Comment | Uncategorized | Tagged: , , , , , , , , , | Permalink
Posted by Barton George

Dell DCS unveils its 4th HPC offering in 12 months, and its a beefy one

February 14, 2011

Today Dell Data Center Solutions (DCS) is announcing the PowerEdge C6145, number four in our line of offerings targeted specifically at High Performance Computing.  This AMD-based system, which contains two four-socket servers for a total of 96 cores, ranked as the highest performing x86 2U shared infrastructure server on the market based on SPECfp_rate2006 results. In addition, the PowerEdge C6145 can deliver up to a 534% better price performance at 1/5 the cost and 1/4 of the rack space when compared to HP’s ProLiant DL980 G71.

The HPC beat goes on

When we in DCS launched our PowerEdge C line almost a year ago, our first HPC-focused machine was the Intel-based C6100.   We followed it three months later with our C410x expansion chassis to supercharge it and then, three months after that, we came out with the AMD version of the C6100, the PowerEdge C6105.   Now three months after that system debuted we are unveiling the C6145.  All three servers come in the same 2U package but with differing chips and architectures targeted at different HPC application types.

Check out the video below and let the C6145 architect, John Stuewe take you on a quick tour of this new muscle machine.

Hairy problem solver

The PowerEdge C6145 with its 755FLOPS and up to 1T of memory is specifically geared to solving big problems involving huge and complex data sets in mapping, visualization, simulations and rendering, and solving them faster.  With regards to efficiency, the shared infrastructure design of the system can reduce the number of individual fans by 1/4 compared to traditional 2U systems with less power needed to cool and resulting in higher performance per watt, per dollar.

Super charge it

As if 96 cores packed into 2U wasn’t powerful enough, you can take your workloads “to 11” with the help of the PowerEdge C410x.  The C410x PCIe expansion chassis allows you to double the server to graphics processing unit (GPU) ratio to 1:8 to create a number-crunching uber powerhouse.

Dell DCS has been listening to their HPC customers and rolling out systems to meet their needs, today we’ve announced the latest in our line up, the PowerEdge C6145.

Extra-Credit reading:

Pau for now…

1 Based on testing by Dell Labs. Dell PowerEdge C6145: SPECfp_rate2006 of 1310 in 2U as compared to HP ProLiant DL980 G7: SPECfp_rate2006 of 1080 in 8U.  SPEC® and the benchmark name SPECfp® are registered trademarks of the Standard Performance Evaluation Corporation.  Competitive benchmarks stated above reflect results published or submitted to www.spec.org as of Feb 14, 2011.  The comparison presented above is based on the best performing 8-chip x86 servers.  For the latest SPECfp_rate2006 benchmark results, visit http://www.spec.org/cpu2006. Actual performance will vary based on configuration, usage and manufacturing variability.

3 Comments | Uncategorized | Tagged: , , , , , , , | Permalink
Posted by Barton George

NVIDIA: from gaming graphics to High Performance Computing

September 22, 2010

A few weeks ago a group from NVIDIA was out visiting Dell.   Their Tesla series of GPU cards are the primary cards that are used in our newly announced C410X expansion chassis.  Filling up the C410X with NVIDIA cards and attaching it to a server can bring about ginormous increases in compute performance, helping to make HPC and scaled-out deployments wicked fast.

So how did NVIDIA get from rendering graphics for first person shooters to creating GPUs that accelerate modeling, simulation, imaging, signal processing,  etc?  Listen to the interview below with Geoff Ballew of NVIDIA’s Tesla unit and learn. 🙂

Some of the ground Geoff covers:

  • NVIDIA’s not just for gaming any more
  • A few years ago found that their graphic chips were getting a lot of raw math horsepower, so they added a few extra features into the chips and built a suite of software so that the graphic cards could be used for general computation.
  • How hard was it to convince HPC customers to take NVIDIA seriously in the compute arena?
  • What kind of performance gains are they seeing?
  • The accompanying software development tools and ecosystem of partners
  • The shift in NVIDIA’s workforce and culture as they’ve gotten into general compute processing – united by their love for GPUs 🙂

Extra-credit reading:

Pau for now…

1 Comment | Uncategorized | Tagged: , , , , , , | Permalink
Posted by Barton George

Schlepping a 410x across Austin – A documentary

September 9, 2010

Last month we introduced the PowerEdge C410X expansion chassis which, when populated with GPGPUs and attached to a server, brings about ginormous increases in performance in a very cost effective manner.

A couple weeks after the system debuted NextIO, who creates and sells virtualized IO capabilities, was looking to qualify the machine in their lab located here in Austin.   Wanting to add that personal touch, Franklin Flint and Corbin Moore from our OEM solutions group decided to pack the system in the back of Franklin’s truck and hand deliver it to Bob Shaw at NextIO.

What you have below is a no-expenses-spared documentary of their journey.  Enjoy! 🙂

Extra credit reading:

Pau for now…

2 Comments | Uncategorized | Tagged: , , , , , , , | Permalink
Posted by Barton George

PowerEdge C410x — Whiteboard topology

August 5, 2010

In the last of my GPGPU/PowerEdge C410x trilogy I offer up a whiteboard session with the system’s architect, Joe Sekel.

Some of the topics Joe walks through:

  • How does having remote GPGPUs connected via cable back to a server compare in performance to having the GPGPUs embedded in the server?
  • The topology of the PCI express  x16 (16 lanes per link) plumbing: from the chipset in the host sever through to the GPGPU.
  • The data transfer bandwidth that x16 Gen 2 gives you. 

Extra-credit reading:

Pau for now…

2 Comments | Uncategorized | Tagged: , , , , , , , | Permalink
Posted by Barton George

Deep dive tour(s) of the PowerEdge C410x

August 5, 2010

In my last entry I talked about the wild and wacky world of GPGPUs and provided an overview of the PowerEdge C410x expansion chassis that we announced today. For those of you who want to go deeper and see how to set up and install this 3U wonder you’ll want to take a look at the three videos below.

  1. Card installation: How to install/replace a NVIDIA Tesla M1060 GPU card in the PowerEdge C410x taco.
  2. Setting up the system: How to set up the PowerEdge C410x PCIe expansion chassis in a rack, power it up and pull out cards.  Also addresses port numbering.
  3. BMC card mapping: How to map the PCIe cards in the PowerEdge C410x via the BMC web interface.  Also covered are how to monitor power usage, fans and more.

Happy viewing!  (BTW, the C410x’s code name was “titanium” so when you hear Chris refer to it as that don’t be thrown)

Extra-credit reading:

Pau for now…

23 Comments | Uncategorized | Tagged: , , , , , , , , | Permalink
Posted by Barton George

Say hello to my little friend — packing up to 16 GPGPUs

August 5, 2010

While the name GPGPU, which stands for General-purpose computing on graphics processing units, doesn’t flow lyrically off the tongue, it’s an extremely powerful concept.

What’s the big idea?

The idea behind this sexy five letter acronym is to take a graphical process unit (GPU) and expand its use beyond graphics.  Through the “simple” addition of programmable stages and higher precision arithmetic to the rendering pipelines, the GPU is able to tackle general computing and off load it from the CPU.

So what does this mean and/or why should you care?  Well the connection of GPGPUs to servers bring about ginormous increases in performance helping to make HPC and scaled-out deployments wicked fast.  This works particularly well when you’re talking about modeling, simulation, imaging, signal processing, gaming etc.  Not only can the addition of GPGPUs boost these processes by one or two orders of magnitude but it does so much more cost effectively than by simply adding servers.

What is Dell’s DCS group offering up?

The Data Center Solutions (DCS) team have an Oil & Gas customer that is always looking to push the envelope when it comes to getting the most out of GPGPU’s in order to deliver seismic mapping results faster.  One of the best ways to do this is by increasing the GPU to server ratio.  In the market today, there are a variety of servers that have 1-2 internal GPUs and there is a PCIe expansion chassis that has 4 GPUs.

What we announced today is the PowerEdge C410x PCIe expansion chassis, the first PCIe expansion chassis to connect 1-8 servers to 1-16 GPUs.  This chassis enables massive parallel calculations separate from the server, adding up to 16.48 teraflops of computational power to a datacenter.

But enough of my typing, see for yourself in the overview/walk-thru below starring DCS’s very own Joe Sekel, the architect behind the C410x.

Extra-credit reading

Pau for now…

9 Comments | Uncategorized | Tagged: , , , , , , | Permalink
Posted by Barton George