As IDF has started, the first benchmarks of Nehalem will probably pop up. It is without a doubt an impressive architecture that gets a much better platform to run on, but this CPU is not about giving you better frames per second in your favorite game than the Penryn family. …Nehalem is about improving HPC, Database and virtualization performance, much less about gaming performance. Maybe this will change once games get some heavy physics threads, but not right away.

…But that doesn’t mean Intel made a bad choice. Intel made a superbly good choice by improving the performance where Core (Merom/Penryn) was mediocre to good. Penryn was already a magnificent gaming CPU, but it could not beat the AMD competitor in HPC benchmarks. And AMD gave good resistance in the database performance benchmarks. That is all going to change.

Share This

Sun Microsystems today extended its family of Intel Xeon-based servers including the fastest 1U server for HPC workloads and the most expandable 2U enterprise-class system based on Intel Xeon processors….

The Sun Fire X2250 server is a perfect 1U server for HPC workloads such as MCAE, EDA, energy and financial services, giving customers an inexpensive compute engine for highly dense, power-sensitive environments that delivers lightning-fast performance and can easily scale for parallel processing. When used in conjunction with the freely available Lustre(TM) file system, customers running these applications can scale to tens of thousands of nodes, petabytes of data, and billions of files. Full Story or find out more on the Sun HPC Radio Podcast.

Share This

AS6000 – Designed to meet the needs of commercial organizations heavily invested in research and product development (R&D) for design, modeling and visualization, the AS6000 includes 20 GB cache per storage shelf with an integrated 10GigE switch that doubles the throughput performance per storage shelf to over 600 MB/s.

AS4000 – Designed primarily for companies dependent on simulation and analysis, like those in the oil & gas, aerospace and automotive sectors, the AS4000 includes an integrated 10GigE switch for unsurpassed speed, plus improved reliability and data availability with the Panasas Tiered Parity architecture.

The 6000 updates the AS5000 previous generation product, and the big advancements are the 10 GbE and 20 GB/shelf cache. Likewise the 4000 replaces the earliers AS3000 generation, and also has the new 10 GbE and a doubled cache/shelf (though this time to 10 GB, not 20). The company has also introduced was it calls the industry’s first parallel second tier storage solution

AS200 – The industry’s first parallel second-tier storage solution, the AS200 configuration includes 104 TBs of available storage space, 5 Gigabit Ethernet ports for fast data transfer, and Panasas Tiered Parity data protection. The solution extends parallel storage capabilities to second tier applications, providing a fully unified parallel storage platform for both primary and secondary storage and delivering improved performance, scalability and manageability.

Pricing ranges from $5/GB for the 6000, down to $1.2/GB on the 200. You’ll see that difference in the performance at the low end: where the 200 delivers 104 TB over 5 shelves with an aggregate bandwidth of 350 MB/s, the higher end products offer 600 MB/s per shelf.

HPCwire will be running a feature [UPDATE: LINK ADDED] on this later today/tomorrow (that I wrote), so you’ll be able to read more there.

Share This

Not that most people knew it was in the clustered storage business but Rackable has decided to sell its RapidScale clustered storage business. Taken with EMC, HP and IBM’s slowness in promoting their extreme scalability products it looks as if such products are very hard to get right, leaving Isilon in pole position.

Rackable also announced the news on its web site on Thursday. To whom will the sale be made? That, evidently, isn’t clear

And Rackable? Which storage systems supplier needs to buy a highly scalable clustered storage hardware and software development team and assets? A supplier with no product presumably and that would identify just three possible candidates: HDS; Sun; and possibly Dell, not a good selling environment for Rackable. Making extremely scalable storage work seems to be extremely hard work.

Rackable itself says it’s looking for a suitor and has subscribed to a dating service

As discussed on the company’s recent Q2 earnings call, the decision was made to seek strategic alternatives for the RapidScale product and thus Strategic Advisory Services International, LLC has been retained as the financial advisor.

Share This

One of the unconventional aspects of the system’s design is that it will incorporate both Power6- and Intel Nehalem-based clusters using IBM’s new iDataPlex platform. According to company sources, 300 teraflops will be Nehalem-based nodes and 60 teraflops will be Power6-based. This is yet another example of IBM’s increasing comfort with hybrid computing platforms.

Michael rightly points out that the sticky wicket in hybrid systems is software, from the OS to the app

Of course, the difficult part of hybrid computing is the software. Initially, individual applications will execute on one architecture or the other. Cluster Resources’ Moab cluster scheduler will be used to map jobs and will also include specific enhancements for a stateless, diskless multi-architecture cluster. The entire cluster is hooked to a storage system and uses GPFS as the backend file system. Both processor architectures will share these resources.

The rest of the post is interesting too, with a look back at the (considerable number of) Power6 systems announced this year: 600 TFLOPS worth. Worth a read.

Share This

The New York State Office of the Chief Information Officer and Office for Technology (CIO/OFT) and the New York State Foundation for Science, Technology and Innovation (NYSTAR) today announced that public and private researchers and businesses in New York state have nearly 150 million CPU hours on one of the world’s most powerful supercomputers. Over the next three years businesses, universities and state agencies can apply for time on the supercomputer located at Rensselaer Polytechnic Institute’s Computational Center for Nanotechnology Innovations (CCNI).

Where do the FLOPS come from?

New York State invested in a $100 million partnership with Rensselaer and IBM to create one of the world’s most powerful university-based supercomputer centers. As part of New York’s investment in the CCNI, the state was allocated 20 percent usage of the supercomputer. The state has adopted a usage policy which gives preference to economic development but also includes use for state agencies to conduct research.

Share This

Ashlee Vance caught a glimpse under Microsoft’s skirts on the interwebs last week

A crafty web site called iStartedSomething caught Microsoft revealing its metal haul in a promotional video. The question and answer session with Chief Environmental Strategist Rob Bernard included some visual snaps detailing the power consumption, jobs and server count of a number of Microsoft data centers. When totaled all up, you find 15 data centers with 148,357 servers, 17,406 racks and a total power consumption of 72,500KW as of Jan.

Vance does some Googling and finds that, with recent, publicly announced additions, that should bring MS to about 220,000 servers.

Moving forward, however, Redmond will need to try a bit harder if it’s to meet Bill Gates’ goal of owning millions of servers.

Indeed.

Share This

Many years ago, I started a “good new/bad news” joke about Linux HPC clusters. The good news is Linux HPC clusters give you the freedom to design and build your HPC machine to your needs. The bad news is Linux HPC clusters give you the freedom to design and build your HPC machine to your needs. One needs to be care [sic] with that freedom thing. In particular, there are two areas where this “freedom feature” can hurt the market place. Tools like xCAT go a long way in helping manage the “freedom” we’re all looking for.

The article goes on to outline value to ISVs, end users, and the market in general. Worth a read.

Share This

Share This

Online communities are a great information source, and can help you stay connected with latest industry trends. HPCwire offers free membership to their HPC Horizons Online Community. It’s open to qualified HPC professionals who are registered users on the HPCwire.com website. As before, vendor participation is limited to sponsors only. Join your 200+ colleagues who are already members of this peer-to-peer network. Stop by www.HPCwire.com, click on Sign In at the top of the page, and register to request your (free) membership.

When you do, you’ll have a chance to laugh publicly at the top 5 things I’d do if I were king of HPC for a day, and add your own five.

Share This