Sun Microsystmes Releases Open Source Fortress
Link: Sun Microsystmes Releases Open Source Fortress
Sun Microsystems has released an open-source software prototype called Fortress “interpreter”.
The software is a programming tool to execute Fortress programs line by line.
Fortress is the planned replacement for Fortran, a programming language launched in the 1960s by IBM. Fortran is still used for high-performance computing tasks such as forecasting the weather.
Fortress was originally developed from a project funded by the US Defense Department, but it has mainstream computing applications such as extracting work from new processing engines appearing in multicore processors.
Multicore is expected to become more widely used in ordinary desktop systems meaning that programmers will have to use a language such as Fortress in order to gain most advantage from the hardware.
Mainstream x86 chips from Intel and Advanced Micro Devices currently have two or four processing cores, while Sun Niagara chips have eight cores and will move to sixteen soon. It is difficult to break down software into independent chunks that run in parallel across all the cores or across multiple processors.
Fortress is designed to allow programmers to write programs in a way that functions better on multicore.
Fortress also tries to store data intelligently, locating it near the processor that needs it. This is expected to help with large compute clusters that gang together independent servers on a high-speed network.
In addition to the interpreter, Sun hopes to develop a compiler, which translates the software in advance into a form a computer can understand from the code a person wrote. Compiled software is generally faster than interpreted software.
Sun is also planning to develop an optimizing compiler which will enhance performance by adjusting the compiled version of software as it runs.
Yahoo! and IBM offer free enterprise search
Link: Yahoo! and IBM offer free enterprise search
Yahoo! and IBM have teamed up to deliver a free enterprise search application for SME’s.
The IBM OmniFind Yahoo! Edition will apply IBM’s search technology but use an interface from Yahoo!.
The aim is to undercut products from rival companies such as Microsoft and Oracle, by delivering it for free to companies unwilling or unable to spend $30,000 on IBM’s high-end Omnifind software.
The new product may not be so powerful, but Yahoo is almost certainly hoping to leverage the software to help get more search and enterprise users into Yahoo!, especially as the product aims to link to Yahoo! online resources.
Just how effective the strategy may be remains to be seen, though.
IBM writes faster than Flash memory
Link: IBM writes faster than Flash memory
IBM have developed a memory chip that is 500 times faster than Flash memory.
It also uses only one half the power to write it as well.
The results were achieved using Phase Memory, which developers at IBM, Macronix and Qimonda, hope will be the successor to Flash.
Additional storage is going to be a prime requisite in multi-media rich mobile applications, and IBM think they’re onto a winner here.
The Phase Memory cell requires a complex semi-conductor alloy, which was developed using mathmatical models to track the predicted behaviour of phase memory cells.
However, much as though we all applaud them for the current advance, the big test comes not just from having working applications in consumer technology.
It also comes from manufacturers opting to use it in the first place.
Toshiba and SanDisk have already invested $2.6 billion in a join Flash memory development plant, and Intel and Samsung are also chasing successors.
May the best chip win.
AMD: Opteron chipsets on high
Link: AMD: Opteron chipsets on high
Advanced Micro Devices (AMD) continues to increase its market share of the server processor market, now accounting for over 20% of x86 server processors in Q1 of this year, according to Mercury Research.
The company looks so inviting that IBM, who has normally only distributed AMD chipsets in higher end solutions, is now looking at options to more widely distribute them in lower-end servers.
It follows recent successes that Hewlett-Packard and Sun have had with sales of AMD chipset servers.
Despite this week recalling 3,000 Opteron processors, after potential malfunctions working at high temperatures, AMD continues to be an ever strengthening brand.
This was highlighted by Intel, who yesterday highlighted that AMD in-roads into the computer chipset market was having a serious effect on earnings.
Overall, it shows a key shift in the overall chipset market. In an area that once saw Intel as the top dog and AMD the struggling budget provider, the tables are now turning as AMD continues to claim mainstream markets.
Discussion: Opteron Increases Market Share
IBM to enter continuous data protection
Link: IBM to enter continuous data protection
IBM, the US computer company is developing software that will continuously back up information on the PCs of small and medium-sized firms. The company is adapting its corporate Tivoli product as part of its strategy to secure more business from firms with less than 1,000 staff.
Smaller firms often have inadequate protection against computer crashes or virus attacks.
The IBM Tivoli Continuous Data Protection for Files software will also be able to back up work on laptops from staff at remote locations. It will copy data within seconds of it being entered or changed on a PC or a more powerful server computer. This is a significant improvement on rival crash recovery products which only back up information several times a day.
According to Reuters, the software will be unveiled on Friday and will go on sale on September 16. It will cost $35 (£19.41) per laptop or desktop PC and $995 per server processor.
Dianne Macadam, an industry analyst at US-based DataMobility Group said: “The enterprise market is not seeing the high growth rate that we are seeing in the small and medium-sized market”.
Ron Riffe, IBM’s director of storage software strategy said: “Small and medium-sized businesses are contending with a ‘data avalanche’, vulnerable to computer viruses or crashes”.
IBM tests SHARK for text input
Link: IBM tests SHARK for text input
Researchers at IBM expect tracing letters on a touchscreen to become as easy as using a keyboard to write emails. The company has developed software, known as Shark (Shorthand-Aided Rapid Keyboarding) that works by recognising the pattern of words.
Although handheld computing devices such as PDAs and Tablet PCs are becoming increasingly powerful, their usage is limited by the current methods of tapping out words on an on-screen virtual keyboard, using software to decipher hand-writing or learning a kind of shorthand. All of these methods are slow and error rates are high. Speech recognition software is another option, but this has problems with long, complete sentences.
IBM researchers Dr Shumin Zhai and Per-Ola Kristensson found that people remember patterns and not letters, via higher level learning that is encoded in human memory. The researchers used this idea to develop software that uses geometrical patterns to represent words. Instead of tapping out letters or writing a word on screen, the user traces each letter in a single stroke. The keyboard on screen shows the shape of the word.
According to Dr Zhai, 100 patterns cover approximately 40% of the words most people write.
In tests, some users achieved speeds of up to 70 words per minute, which is slower than touch-typing but much faster than tapping out words with a stylus.
An experimental version of the software has been available as a free download from IBM’s Alphaworks site for a year. It is now ready to go into commercial development and could be featured on PDAs in the near future.
BlueGene/L supercomputer still most powerful computer
Link: BlueGene/L supercomputer still most powerful computer
IBM’s BlueGene/L supercomputer, which is under construction at the Lawrence Livermore National Laboratory in the US, has retained its leading place in the list of the most powerful machines on Earth. It has a processing peak of 136.8 teraflops, which will be doubled when construction is complete.
The Top 500 list is drawn up every six months. IBM’s BlueGene/L was placed first for the second time because its processing capacity doubled in size since the last list was drawn up.
When it is completed the BlueGene/L will use 65,536 processors to look at problems such as molecular dynamics, metrial modelling and turbulence and instability in hydrodynamics.
A BlueGene system was also in second place. It was recently installed in IBM’s Thomas J Watson Research Center in New York and has a processing peak of 91.2 teraflops.
IBM produced 51.8% of the machines on the Top 500 list, alongside manufactures such as Silicon Graphics, NEC, Hewlett-Packard and Cray.
Intel was the leading supplier of hardware, which was used in 333 of the systems listed.
Nasa’s Columbia supercomputer at the Ames Research Center in California was in third place.
The basic power requirement needed to be placed in the Top 500 list is 1.166 teraflops, compared with the basic 850.6 gigaflops needed to be included in the November 2004 list.
The Top 500 list is compiled by Hans Meuer of the University of Mannheim in Germany, Erich Strohmaier and Horst Simon of NERSC/Lawrence Berkeley National Laboratory in the US and Jack Dongarra of the University of Tennessee in Knoxville.
Giants stand behind open source
Link: Giants stand behind open source
IBM and Novell are pushing on the development of SUSE linux applications to run on IBM’s eServer platforms.
The companies are seeking closer working ties, in terms of “consulting support and technical expertise” to help them develop applications for SUSE Linux on IBM platforms.
The move is part of a wider push by IBM to move Linux into the forefront of business technology, and has already spent years developing Linux for wider mainstream use.
Linux is an open source operating systemm developed from Unix, and comes in a number of “flavours”, with SUSE Linux being a particular flavour developed by Novell.
Also this week, Google announced the opening of Google Code, which seeks to use programming languages such as C++ visual basic, and Python, for development of internet applications that could benefit the company.
This follows the earlier unveiling by Yahoo! of the Yahoo! Developer Network, which is intended to focus on development of products that utlise Yahoo! Search technology.