Boinc (Berkeley Open Infrastructure for Network Computing)

Discussion created by mmciastro on Sep 13, 2007
Latest reply on Feb 9, 2010 by diddly

Lately, AMD has been taking it on the chin from Intel processors. There has always been an AMD vs Intel rivalry. Boinc (Berkeley Open Infrastructure for Network Computing) started in 2004. It's middleware for Projects(generally university run) which wish to access large amounts of computer horsepower for what are generally scientific endeavors. Some of these projects(of which there are about 50) include Rosetta@home, Einstein@home, Seti@home, etc. In short it's distributed computing. Anyway some of the applications run by these projects can be open source as is the case with Seti@home ( There are currently 249,633 active boinc users, with 425,462 computers attached (according to BoincStats ).

At Seti@home they have a message board filled with computer enthusiasts of various skill sets. A few have taken it upon themselves to "optimize" the Setiathome application to run faster. However, they are all using ICC and IPP. This greatly decreases the Intel users processing times(increased performance), and thanks to the ICC's CPUID checks, greatly limits any improvements for AMD users. (I.E Intel gets to use SSE3, SSSE3 and those processors are compared to results from the AMD users who can't use SSE, SSE2, SSE3, etc). Basically, it makes AMD look bad when compared to the offerings of Intel, and yes, there's even some Barcelona's being used now. Even the Barcelonas' performance looks bad (even bad when directly compared to my K8 processors). Many purchase decisions are made by the recommendations of higher level users who are firmly convinced that the Intel processor is a much better choice.

I'm of the opinion that it only looks that way, because noone has written an application specifically compiled to work to the advantages of AMD. I'd like to see AMD shine for what it really is, and show them what it can do. Perhaps even boost, rather than hinder AMD sales to the 100K+ Boinc participants.

This brings me to the reason for this post. I have no programming skills, but am so fed up with their bleating, that I'm on a quest for a solution. I've gone to the PGI and other compiler websites, done some research, and decided that I'm not the person to buy a compiler program, learn what it is, and how to use it. I'm not even sure I could get enough comrades around to pitch in on the purchase price to get one license. There is one optimizer(Simon) who's using the ICC and runs a website for the distribution of it's binaries at, who might be interested or helpful to anyone who could use a different compiler to attempt to give AMD users the edge(and improve AMD's standing).

I'm looking for solutions, any solutions to this inequality. I've thought of trying to find compilers who might be interested, Buyiing a compiler (which one I do not know) and taking a crash course on compiling, searching out someone who might do it for a small stipend, heck, anything anyone can think of would be helpful.

What skills does it take to compile? Could someone learn it within the "free trial period" to see if it's even worthwhile spending the cash for a licensed version that would allow distribution of a GPL licensed binary?

thanks for your time


P.S Another good reason for AMD to get involved is to help provide balance to the bleatings of an Intel Senior Performance Analyst which goes by the username "who?" and is bad mouthing AMD every time he posts. Supposedly his name is Francois Piednoel. NOTE: he posts disclaimers that he alone is responsible for his posts, and Intel is not encouraging/supporting his actions during his off times. I don't want to get him in trouble.

P.S.S the following thread shows the first instance of Barcelona, and the posts by Who? (Intel Senior Performance Analyst, Francois Piednoel) are typical. I hope someone can help for the sake of AMD.