Talk:Blue Gene

From Wikipedia, the free encyclopedia

 This article is within the scope of Computing WikiProject, an attempt to build a comprehensive and detailed guide to computers and computing. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and/or contribute to the discussion.
Start This article has been rated as start-Class on the quality scale
High This article has been rated as high-importance on the importance scale

The note about MDGRAPE-3 that was recently added at the beginning of the intro seems inappropriate. The LLNL Blue Gene/L machine's position at the top of top500 hasn't been mentioned at that point in the article. Also, the statement is a bit misleading, as MDGRAPE-3 is so specialized that it can't run the benchmarks normally used to compare the performance of supercomputers -- it's raw ability to perform floating point operations is the greatest in the world, but it isn't really a general purpose computer. The actual Wikipedia article on MDGRAPE-3 does a better job of making that clear. →pjsteiner

I'm guessing by the name that this supercomputer is intended for DNA/Genome processing, is there some info on this that can be added? Indosauros 16:35, Jul 15, 2004 (UTC)

It looks like the computer will try to model protein folding, but I'm sure it will be used in lots of different applications —User:Mulad (talk) 17:09, Sep 29, 2004 (UTC)

There is a wrong information. It states that peak performance is only 360 FLOPS (that means that my laptop is MUCH faster)??? Also, there is no mention about what the final purpose of this computer will be. It is build to figure degradation of US nuclear weapons stockpile for the 21st century (if you knew about the project little more you would know that!) so that US armed forces will not have to do real nuclear testing in the future. There is NO mention of that in article??? It's final version has NOTHING (and read again - NOTHING!) to do with protein folding or computations in molecular biology. That is just BS that is being sold on official bluegene webpage (to make it look more "friendly"). Maby some protein folding computatiojns were conducted for the purpose of testing the machine but the final use of the computer at LLNL for US nuclear weapons testing is widely known and even publicly aknoweledged in many articles by both IBM and US goverment (again no mention of that here on Wikipedia??). The deciedion to build this huge, for any other hpc manufacturer unmached machine was made after 2000 elections for sole purpose of improoving us nuclear weapons. So, thet's it... ...about on how bluegene will be used for fighting cancer and other useful things???

First, it's important to make the distinction that there are *5* blue gene projects - at least one of which (/L) has not even started yet. They are run by totally seperate groups of people (I'm working on /C -- specifically, the testing and verification). I cannot speak for the other projects, but BlueGene/C will be multi-purpose - more or less everything is exposed to the programmer, meaning that a good programmer can write more or less any program to run on it. I'm told that they will be purchasable off the shelf (rader telemetry analysis was the area I was told they would be particularly useful for), so saying that they will be available for all those other purposes is not wrong. →Raul654 06:10, Jan 12, 2005 (UTC)

- Kinda outdated now -

Blue Gene/L is now present not future [1] --202.164.193.221 23:20, 27 Mar 2005 (UTC)

Have you noticed that the operations per second are reaching the order of magnitude thought necessary to simulate the human brain in realtime? Obviously there would be a software issue here

[2]

As a former programmer/user of a Connection Machine CM-2 and CM-5, I'm struck by how similar the architecture is between those machines (in particular the CM-2) and this machine. Pretty much the same idea...just a lot faster CPU's! Even the current number of nodes (65K) was the same as the maximum configurable on a CM-2. I bet Thinking Machines is turning in its grave. ScottRShannon 10:04, 26 February 2006 (UTC)

Contents

[edit] Columbia/Earth Simulator

Hello, just wondering about this line: "This is almost ten times as fast as the Earth Simulator, the fastest supercomputer in the world before Blue Gene."

Wasn't the fastest supercomputer before Blue Gene NASA's Columbia?

NASA press release

Blues Genes did take back it's title a few days after, but shouldn't this be metioned? --69.0.48.86 20:55, 21 July 2005 (UTC)

[edit] Purpose and suitability

In addition to discussing the purpose of each of the projects, the article needs to briefly discuss why this architecture is suitable for those projects. For those of us more familiar with a single-processor computer (which means, I believe, 99% of the readers), it would be interesting to touch on how having thousands of processors is the best current solution for physics simulation. Tempshill 18:36, 31 October 2005 (UTC)

The reasoning for these architectures is the same as for any other supercomputers. As such, such a discussion is probably more suited to the supercomputer article. That reason is that scientific applications tend to be compute-node bound -- e.g, it's the lack of computing power that is holding them back. Fortunately, most scientific tasks posses inherent thread-level parallelism, which is why throwing tons and tons of compute nodes at the problem makes a solution come faster. →Raul654 19:16, 31 October 2005 (UTC)


[edit] Blue Gene/P

It says that Blue Gene/P is supposed to be released in 2006. I don't think that's going to happen because I haven't heard any updates on it and it's a 1 petaflop system (5 times faster than Blue Gene/L). If anyone knows anything about this, we can change to the appropriate date.

IBM was quiet with their milestones before, too. They met their timeline with /L. I think your assessment is premature. There are seven months left in 2006, so just wait. —Joseph/N328KF (Talk) 16:15, 25 May 2006 (UTC)
It will probably never be built. Instead, IBM will build the IBM Roadrunner.--Arado 14:51, 9 September 2006 (UTC)

[edit] Designer genes?

These can already be made and don't really require much in the way of computing power. See Molecular biology. TimVickers 19:42, 1 June 2007 (UTC)

[edit] Blue Gene/P info

[3] details on how many nodes/racks will be needed to hit one petaflop. -Ravedave 17:41, 26 June 2007 (UTC)

Line 76/

It makes no sense to talk about performance per Ohm, performance per watt does.

Learn how to spot vandalism, and correct it on sight. -- Henriok 23:46, 15 July 2007 (UTC)

[edit] Update on speed of Blue Gene/L

Blue Gene/L seems to have been upgraded over summer, and is now clocking in at 478 TFLOPS. References http://news.bbc.co.uk/1/hi/technology/7092339.stm http://www.hpcwire.com/hpc/1889245.html http://www.hemscott.com/news/latest-news/item.do?newsId=53878217307964

Should maybe change the article to reflect this.Malbolge 12:09, 13 November 2007 (UTC)