For more info, or to track the video release, read their full post here.

Delta-V systems are offered in four models, housing 4 (1U), 12 (2U), 16 (3U) or 24 (4U) hot-swappable hard disks. These systems provide up to 6, 18, 24, or 36 Terabytes of raw storage capacity in RAID6 configurations, respectively. RAID10 configurations can provide reduced usable storage in order to achieve greater performance for database operations. Delta-V presents storage simultaneously as file-based and iSCSI block, with NFS and CIFS/SMB. Delta-V units may also be aggregated into storage clusters providing Petabytes (PB) of capacity when utilizing cluster file systems and RAIN designs. The Delta-V demonstrated sustained data transfer rates of 500+ MB/s with greater than 1800 IOPs in a RAID10 configuration at SC08.

Delta-V expands SI’s storage product offering by adding a lower cost complement to their JackRabbit offering

“With our JackRabbit servers already providing reasonably-priced, best-in-class performance and reliability, we felt we needed to add a focus on the area where many organizations have the least leeway — cost,” stated Dr. Joseph Landman, CEO and Founder of Scalable Informatics.

“By optimizing our design to minimize cost, and tuning the software stack for reliability and performance, we are able to provide storage appliances that scale from 2 to 31 terabytes of usable space, for well under $1,000 per usable terabyte.”

In case you aren’t familiar with Scalable Informatics, that’s the outfit run by Joe Landman, a name you will no doubt recognize from reading this blog. I point to his insightful analysis over at scalability.org with some frequency. If you aren’t a regular reader of Joe’s, I recommend including his feed as part of your daily HPC nutrition program.

Dr. Ulrich Widmann, Head of Vehicle Safety Development at Audi, said after the acquisition: “The new supercomputer accelerates simulation workflows several times over. With our constantly growing model range it’s essential for every model to be put through its paces during development in every conceivable accident scenario.”

According to Audi, the developers can now run up to 5,000 simulation scenarios per week.  These may include individual finite element analysis of components or full-scale crash scenarios.

For more info on the new digital digs at Audi, read the full release here.

USC’s HPCC has achieved its current status among the world’s top supercomputer sites through moderate local investments and without national funding,” said Priya Vashishta, faculty executive director of High-Performance Computing and Communications. “USC’s approach to granting access to its supercomputing resources is one of the best in the nation: HPCC allows USC faculty and graduate students unfettered access to a world-class facility.”

Interestingly enough, the machine has a 10 gigabit interconnect backbone.  No mention on whether this is Myrinet or Ethernet.

For more info, read the full release here.

These days, Patterson is still at Berkeley, and he runs the Parallel Computing Laboratory that is funded largely by Intel and Microsoft - Par Lab for short. As the name suggests, the lab is trying to tackle the parallel computing problem in new ways. Both corporate and consumer computing today is wrestling with this parallelism problem, right there in the data center and on the desktop, a problem that has plagued supercomputing for decades.

…Solving this particular variant of the parallelism problem is going to take some mind shifting across the IT industry. First, we must stop asking the question about who will need a 100-core processor to run Microsoft Word. Patterson said that while he is a Word user - and he likes Word - the question irks him.

“Questions like that make me think we have failed as educators,” Patterson said. The real issue has nothing to do with supporting legacy applications. “I am pretty sure that the best software has not been written yet,” Patterson said. And he gave a few examples of neat projects that could eat up a lot of parallel processing capacity in a single system.

When Mott took over as HP’s CIO, the company had 85 data centers, and over the past three years, it has consolidated all of its operations down to a “six pack” of three redundant, highly virtualized data centers. The goal when Mott started was to move from 25,000 servers sprawled across those 85 data centers to around 14,000 machines in those three mirrored data centers. HP also chucked two-thirds of its 6,000-strong application portfolio, a legacy of the many companies it has acquired and disparate divisions.

…HP did not provide a final server count as the IT project finished, but did say today that it has cut the number of servers in use at the company by 40 per cent while boosting the processing power in its iron by 250 per cent (that means by a factor of 3.5). A 40 per cent reduction in the servers would mean it has around 15,000 boxes today. The company also said that it has cut its networking costs by 50 per cent over the three years while tripling network bandwidth. (Keeping CPU capacity and network bandwidth more or less in balance makes sense.)

Why bother? Two reasons: money now, and money later

But here’s the bottom line - in fact, two of them. The changes that HP has done have removed $1 billion in costs out of the company, and that is real money. In fact, internal IT spending will drop from about 4 per cent of fiscal 2005’s $86.7bn revenue to a projected under 2 percent of fiscal 2009’s revenue. HP is projecting revenues for fiscal 2009 to be in the range of $127.5bn to $130bn, thanks in large part to its acquisition this year of outsourcer Electronic Data Systems. Cutting IT spending was obviously low-hanging fruit in terms of cost cutting at the company, particularly considering the low-cost of experts, hardware, and software.

…But the IT effort is undoubtedly being used as a training mechanism for future HP engagements with its customer base to transform their inefficient IT operations. And that is where the six-pack data centers and massive application and server consolidation will pay off on the bottom line a second time for HP.

More and more people have come to the conclusion that locking is the wrong approach to solving the consistency issue. This is especially true for programmers who are not intimately familiar with all the problems of parallel programming (which means almost everybody).

…The problem of consistency is nothing new in the world of computers. In fact, it has been central to the entire solution in one particular area: databases….The solution in the database world is transactions.

The concept of the transaction is something that falls out of most programming tasks quite naturally. If all changes that are made as part of a transaction are made available atomically all at once, the order in which the changes are added to the transaction does not matter. The lack of a requirement to perform the operations in a particular order helps tremendously. All that is needed is to remember to modify the data sets always as part of a transaction and not in a quick-and-dirty, direct way.

…Fortunately, this is not needed. The concept of TM (transactional memory) has been defined without this restriction. Maurice Herlihy and J. Eliot B. Moss in their 1993 paper1 describe a hardware implementation that can be implemented on top of existing cache coherency protocols reasonably easily.

The con article mostly focuses on the performance impacts of software transactional memory, an intermediate step between where we are today and hardware supported TM. The title gives its focus away: “Software Transactional Memory: why is it only a research toy?”

TM (transactional memory)1 is a concurrency control paradigm that provides atomic and isolated execution for regions of code. TM is considered by many researchers to be one of the most promising solutions to address the problem of programming multicore processors. Its most appealing feature is that most programmers only need to reason locally about shared data accesses, mark the code region to be executed transactionally, and let the underlying system ensure the correct concurrent execution. This model promises to provide the scalability of fine-grain locking, while avoiding common pitfalls of lock composition such as deadlock. In this article we explore the performance of a highly optimized STM and observe that the overall performance of TM is significantly worse at low levels of parallelism, which is likely to limit the adoption of this programming paradigm.

But the closing section of the full paper indicates that the authors are…skeptical…that even a hardware implementation will help

Based on our results, we believe that the road ahead for STM is quite challenging. Lowering the overhead of STM to a point where it is generally appealing is a difficult task, and significantly better results have to be demonstrated. If we could stress a single direction for further research, it is the elimination of dynamically unnecessary read and write barriers—possibly the single most powerful lever toward further reduction of STM overheads. Given the difficulty of similar problems explored by the research community such as alias analysis, escape analysis, and so on, this may be an uphill battle. Because the argument for TM hinges upon its simplicity and productivity benefits, we are deeply skeptical of any proposed solutions to performance problems that require extra work by the programmer.

We observed that the TM programming model itself, whether implemented in hardware or software, introduces complexities that limit the expected productivity gains, thus reducing the current incentive for migration to transactional programming and the justification at present for anything more than a small amount of hardware support.

I read both of them, though admittedly at more of a skim. I’ll be reading both of these papers carefully on the plane tomorrow.

Honestly, this video is so good that it’s hard to believe it was created, or even commissioned, by the government. The bottom line is that because we humans have very short memories and fundamental discoveries take a long time, science does not stand on its own. It needs advocacy, and advocates. This kind of video is science’s ideal advocate.

Watch it, and share the link.

From The Register article

[Sun] bought StorageTek (high-end and mid-range tape and disk storage products) for around $4bn a few years ago and MySQL (open source database software) for a billion earlier this year. So far neither acquisition has prevented Sun’s sales revenue slump. Nor has its open source system software strategy, aimed at developers with consequent hardware and service sales dragged along in their wake. It’s a classic product transition problem with revenue from new and lower-priced products not rising fast enough to offset declines in older products.

What to do? The Register speculates

We can say that the main parts of Sun’s business are StorageTek tape and disk storage, X86 servers and server-based storage, SPARC servers, software, and services. EMC is a storage and virtual server company and thinks Open Source is as attractive as giving your eyes an acid bath. It might be interested in the StorageTek tape business, giving it a stronger entry into mainframe shops. And it might be interested in the server/storage hybrid products such as the X4500 (Thumper). The servers though, both SPARC and X86? I can’t see it. The open source software? Don’t think so - selling it off to Red Hat or Novell might be possible. So an EMC takeover or merger would be followed by dismemberment of Sun.

And the speculation continues in the article. The Register closes with the point that Sun has to make a call, and decide whether its in or out, and then decide that if it wants to live what parts of it have to die. Joe makes the same points more articulately than I could

Sun is in this position. It has multiple albatross’ or anchors. It doesn’t want to admit it. Sun execs, if this article ever gets to you, don’t make the same mistakes as SGI made here, choosing comfort over need. That obviously did not work out well for them, and it won’t for you either. Part of the job is recognizing what battles are still winnable, and closing down the battles that aren’t.

Recommend you read both articles.

Cool Crunching: Understanding Green HPC

on Wednesday (December 3, 2008) at 11AM EST. More information and FREE registration click here.