The Market Meltdown & The Question of Trust (2)
share
digg
by
Om Malik (11)
on
GigaOM (28)
6 days, 18 hours
ago
permalink
Over the weekend, I ended up on San Francisco’s ritzy Fillmore Street. I stopped by to say hello to my dry cleaner and then to Ed Nahigian, a gentleman who has been taking care of my footwear since I moved to San Francisco. As with everything else, the conversation turned to the market meltdown and the pending auto-industry bailout. From our conversation emerged some common sense advice that’s applicable to business of all shapes and ...
Project Molecule ก้าวต่อไปของ Atom ที่ไม่ได้มาจากอินเทล (2)
share
digg
by
lew (0)
on
Blognone (0)
1 week, 2 days
ago
permalink
ซีพียูที่ร้อนแรงที่สุดในปีนี้คงต้องยกให้ Intel Atom ที่เปิดตลาด netbook อย่างรุนแรง แต่ถ้าคนอยู่ในวงการวิจัยหรือกระทั่งคนที่ต้องดูแลเซิร์ฟเวอร์จำนวนมากๆ แล้วคงแอบหวังที่จะเอา Atom ไปใช้ในเซิร์ฟเวอร์กันอยู่ เพราะการปล่อยความร้อนที่ตำเอาเสียเหลือเกิน และเจ้าแรกที่เปิดตัวเซิร์ฟเวอร์รุ่นต่อไปด้วยแนวคิดนี้จริงๆ ก็คือ SGI ที่เปิดตัวเซิร์ฟเวอร์ขนาด 3U ที่อัดซีพียู Atom ไว้ถึง 180 ตัว ชื่อรหัสของโครงการนี้คือ Project Molecule (เอาอะตอมมาเรียงกันต้องเป็นโมเลกุลสิ) โดยทำเมนบอร์ดแบบพิเศษที่มีขนาดเล็กมาก แล้วใส่กล่องเป็นบล็อกเรียงกันไป ชิปที่ใช้คือ Atom N330 ที่มีสองคอร์ในตัว ทาง SGI ระบุว่าแต่ละบอร์ดจะรองรับหน่วยความจำได้ 2 กิกะไบต์ ทำให้หนึ่งเครื่องจะมีคอร์ x86 อยู่ถึง 360 ชุดและหน่วยความจำสูงสุด 360 กิกะไบต์ โดยที่กินไฟน้อยกว่า 2 กิโลวัตต์ทั้งตู้ ก้าวต่อไปสำหรับโครงการนี้คือ การขายตู้ทั้งตู้ที่มี Atom อัดแน่นมากกว่า 10,000 คอร์ในตู้เดียว โดยจะมีระบบหล่อเย็นแบบของเหลวเข้ามาช่วยด้วย ปรกติแล้วระบบกริดขนาดเล็กๆ เช่นนี้ก็มักจะใช้ชิปขนาดเล็กๆ เพื่อความคุ้มค่าในแง่ของพลังงานกันอยู่แล้ว แต่ Atom นั้นจะเป็นครั้งแรกที่เราได้เห็นชุดคำสั่ง x86 บนเครื่องแบบนี้ ที่มา - Tech-On
-
Pruet said:
SGI กลับมาแล้ว!!!
Random Notes from the IDC HPC Breakfast Briefing (1)
share
digg
by
Josh Simons (0)
on
The Navel of Narcissus (0)
1 week, 3 days
ago
permalink
I went to the IDC HPC Breakfast briefing yesterday morning because they are usually pretty interesting. This one felt mostly like a rehash of earlier material and was somewhat disappointing as a result. I did hear a few things I thought were worth passing on and here they are. I made the above graph based on a table that was flashed quickly on the screen during the briefing. If N was specified, I didn't catch ...
XBRLミドルウェアの最新版「Interstage XWand V10」を販売開始 (1)
share
digg
Integrating Your Adsense Ads with Your Site and Using ColorZilla (1)
share
digg
by
Anna (0)
on
Building from Nothing (0)
1 week, 4 days
ago
permalink
If you are running Adsense Pay-Per-Click ads on your site, it is usually adviseable to blend the colors in with the rest of your site so that your ads don’t distract, detract from your design, or scream “I’m an Ad!” A general rule which usually works for me is to keep the background color of my [...]
Top500 Madness (2)
share
digg
by
John Leidel (0)
on
insideHPC (0)
1 week, 6 days
ago
permalink
The latest Top500 list was unveiled on Friday. For those who haven’t read the release, whoa nelly! The list has always been a highly debated and contested flexing of one’s supercomputing might. This list was no dissapointment. The biggest story of this year was the backyard brawl between ORNL’s Cray XT5 system, Jagaur, and the reining champ, RoadRunner at LANL. Fighters, to your corners! It was widely publicized that Jaguar had a distinct chance to ...
Økt bruk av OpenOffice i offentlig sektor (1)
share
digg
by
Christer Gundersen (0)
on
Friprog (0)
2 weeks
ago
permalink
Friprogsenteret mottar i dag ytterligere 2 millioner kroner i prosjektstøtte fra Fornyings- og administrasjonsdepartementet. Disse midlene skal vi bruke for å få OpenOffice til å fungere ukomplisert i offentlig sektor. Friprogsenteret skal lede prosjekter med å utvikle løsninger som gjør det mulig for fagsystemer i offentlig sektor å fungere sammen med OpenOffice. Fornyings- og administrasjonsminister Heidi Grande Røys vil at støtten også skal bidra til utvikling av et fri programvarebasert alternativ for samhandling og dokumenthåndtering. ...
Platform revs HPC management suite (1)
share
digg
by
John West (0)
on
insideHPC (0)
2 weeks, 1 day
ago
permalink
From the Platform newsroom Platform Computing…has announced its next generation of HPC management products: Platform LSF 7 Update 4, Platform Analytics 7.1 and Platform RTM 1.5. These products help customers manage their workloads, resources and control costs within their HPC and grid computing environments. The new products improve users’ ability to dynamically meet demand for compute resources with available supply at the lowest cost. You are probably already familiar with batch queue manager LSF, but ...
10,000 Hours (109)
share
digg
by
John Gruber (14)
on
Daring Fireball (13)
2 weeks, 1 day
ago
permalink
Fascinating excerpt from Malcolm Gladwell’s soon-to-be-released Outliers: The Story of Success: This idea — that excellence at a complex task requires a critical, minimum level of practice — surfaces again and again in studies of expertise. In fact, researchers have settled on what they believe is a magic number for true expertise: 10,000 hours. (Thanks to Tom Davis.) ★
-
S Anand said:
"This idea - that excellence at a complex task requires a critical, minimum level of practice - surfaces again and again in studies of expertise. In fact, researchers have settled on what they believe is a magic number for true expertise: 10,000 hours."
-
Jon Winstanley said:
Awesome article about genius.
-
lishevita said:
Genius takes 10,000 hours, or about ten years of 3 hours a day, to blossom into world class success says this article. So, if you aren't a world class expert at anything today, what do you want to spend 10,000 hours refining? Or what are you already putting your hours into?
-
Sandra said:
Craig...this is an excerpt from the article the highlights a similar statistic to the one I like to use about soccer players.
-
Kenny Stoltz said:
Reshared from Mike.I like the idea of this... expertise in a job would take 4.8 years. That probably explains the 5-year-itch, etc. But it doesn't really explain the burnout: why would you spend 10,000 hours at something to be an "expert" and not leverage your expertise? I suppose the best explanation is that human beings are flawed (or at least irrational).
-
CHC said:
A great excerpt from a for sure best-seller... I will buy this book!
-
Kluzter said:
ZOMG.. me wants to be expertsss... the easy way
Fundamental Concepts of Parallel Programming (1)
share
digg
by
Sujana (0)
on
MulticoreInfo.com (0)
2 weeks, 2 days
ago
permalink
Shameem Akhter writes an article for Dr. Dobb’s journal about moving from a linear to a parallel programming model. He is a platform architect at Intel, and Jason Roberts, a senior software engineer at Intel, are the authors of “Multi-Core Programming: Increasing Performance through Software Multithreading” on which this article is based. “Developers who are unacquainted [...]
JFQ - Point - Counterpoint - SWJ Early Exclusive (1)
share
digg
by
SWJ Editors (0)
on
SWJ Blog (0)
2 weeks, 4 days
ago
permalink
Colonel David Gurney (USMC Ret.), Editor of Joint Force Quarterly and Director of National Defense University Press, has been closely following the debate between John Nagl and Gian Gentile and our guest commentators here on Small Wars Journal. For SWJ newcomers or the uninitiated - this debate has centered on the kinds of threats the U.S. will face in the period ahead and how U.S. ground forces should prepare for those threats. Colonel Gurney has ...
Terascala intros new Lustre appliances (3)
share
digg
by
John West (0)
on
insideHPC (0)
2 weeks, 4 days
ago
permalink
HPC storage vendor Terascala announced two Lustre-based storage appliances Tuesday. The MTS 1000 solution is, according the company the world’s first storage appliance with built-in redundant metadata designed to provide high availability, high throughput storage. Based on a parallel file system, the MTS 1000 features an integrated, active-passive redundant metadata server architecture and delivers extremely high throughput, high capacity and superior price/performance in a fully configured and tuned appliance. The MTS 1000 leverages Lustre, the ...
-
mndoci said:
Really need to look into Lustre
Mellanox, Dell and AMD Allow Access to Cluster (1)
share
digg
by
John Leidel (0)
on
insideHPC (0)
2 weeks, 4 days
ago
permalink
As a part of the HPC Advisory Council, Mellanox, Dell and AMD have come together to jointly develop an HPC custer destined for open access. The team will build a cluster composed of twenty-four Dell PowerEdge SC1435’s, AMD Opteron silicon and Mellanox Connect-X Infiniband adaptors. The new cluster, named Osiris, will be dedicated to developing new applications and workload practices. With the main goal of creating best practices within compute-intensive market segments, the HPC Advisory ...
Terascala demos pNFS storage solution (1)
share
digg
by
John West (0)
on
insideHPC (0)
2 weeks, 4 days
ago
permalink
Terascala announced on Tuesday that it and Open Grid Computing would be demonstrating pNFS in the Advanced Simulation and Computing (ASC) Program at SC08 next week. pNFS (also referred to as parallel NFS or NFS V4.1) is an extension to NFS v4 that provides support for direct parallel storage access. pNFS is designed to leverage the full performance and scalability aspects of a parallel file system while maintaining compatibility with the widely-used NFS file system. ...
Антарктида. (1)
share
digg
Purdue’s Rack-A-Node Supercomputer Game (1)
share
digg
by
John Leidel (0)
on
insideHPC (0)
2 weeks, 6 days
ago
permalink
Purdue University has built an online game, Rack-A-Node, to highlight its students participating in this year’s SC08 Cluster Challenge. Rack-A-Node is an online game allowing players their chance at designing and operating a simulated research supercomputer. Players build a cluster using various different types of machines to run scientific experiments. As the player becomes more successful at running jobs, he/she receives more funding to build a bigger machine. [I wish it were that easy in ...