Here's a collection of highlights, selected totally subjectively, from this week's HPC news stream as reported at insideHPC.com and HPCwire.
>>10 words and a link
Sun announces Java port to Itanium, still no servers planned;
CRA and NSF establish Computing Community Consortium, pursue long-term vision;
Intel science fair opens this week; more than 20 percent of entrants have at least one patent;
Amazon's CTO Werner Vogels to keynote LinuxWorld and Next Generation Data Center conferences;
Version 4.2 of GCC toolset released this week, inludes OpenMP;
>>INCITE program to award 250 million hours in 2008
The U.S. Department of Energy has just announced its 2008 INCITE call for proposals. DOE's Office of Science expects to award up to 250 million hours in 2008, nearly three times the total 2007 awards.
The allocations of HPC resources and technical support will be made under DOE's Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program for computationally-intensive, large-scale research projects. The five-year-old program encourages proposals from universities, other research institutions and industry.
In 2007, 45 projects were awarded a total of 95 million processor-hours of computing time. More, including links to the call for proposals and past awards, can be found at http://insidehpc.com/2007/05/17/2008-incite-call-for-proposals/.
>>HP reports Q2 gains
Some bright news for one of the top tier HPC vendors this week following last week's disappointing quarterly postings by Cray and SGI: HP reported substantial gains in its second fiscal quarter.
According to Reuters, HP reported that a 13 percent growth in revenue and continued cost cutting measures lead it to grow its quarterly profits by 27 percent. Much of the revenue growth came from increasing strength in its personal computer and printer businesses. Read more of the Reuters story at http://biz.yahoo.com/rb/070516/hewlettpackard_results.html?.v=6.
>>Reed comments on multicore software crisis
Dan Reed posted a thoughtful entry on his blog this week about the high end computing community's failure to follow through on parallel research started a decade ago, and the trouble that failure is creating for software developers today facing a multicore future. In his entry Reed says of efforts the HPC community started over a decade ago to address the parallel software problem: “These approaches offered options to express large-scale parallelism while hiding many of the low-level details of message passing. They were immature and incomplete, but promising. However, we abandoned these research directions when they did not quickly yield commercial quality solutions. … Simply put, we might now be confidently exploiting data parallelism, even for irregular problems, on today's consumer multicore designs if we had stayed the high-end research course.”
This is, I think, a failure of the U.S. research establishment starting in the mid-90s to focus on long term investments. Thankfully this is starting to change. Professor Reed's post is at http://www.renci.org/blog/pivot/entry.php?id=31.
>>AMD's quad-core announcement
AMD announced this week that it will sell its forthcoming quad-core desktop processors under the Phenom brand. AMD has plans for two quad-core chips to be available in the second half of this year. The Phenom FX and the Phenom X4 will carry the company's performance processing business, with the Athlon 64 X2 in the mainstream and Sempron bringing up the value segment of the market. You can read more at CNET News.com, http://news.com.com/AMD+goes+quad-core+with+Phenom/2100-1006_3-6183379.html.