If you get an error like

/usr/include/c++/4.4/ext/atomicity.h(46): error: identifier “__sync_fetch_and_add” is undefined

when trying to compile your CUDA project which is using FindCUDA.cmake, there is quite an easy solution for this. NVCC does not really like the latest versions of gcc or g++, so you have to try it with version 4.3 of both of them.

So the first step is installing gcc-4.3 and g++-4.3, which should be in the repositories of most Linux distributions, though this may have been simply upgraded to 4.4. On Ubuntu Karmic as well as Debian Squeeze it has been possible to additionaly install the older version in parallel via aptitude.

After having these installed, I first tried to modify my CMakeLists.cmake file by adding this:

set(CMAKE_C_COMPILER “gcc-4.3″)
set(CMAKE_CXX_COMPILER “g++-4.3″)

Hoping that this does the job, I was quite disappointed, as it didn’t. You have to explicitly set the flags which are passed to NVCC, which actually seems to always use the files “gcc” and “g++”. Fortunately, you can pass it a flag which tells where to search for those files. So you can create symlinks “gcc” and “g++” to the 4.3-versions of the tools in a directory of your choice (e.g. /usr/local/bin/gnuc4.3) and add the following line to your CMakeLists.cmake file:

set(CUDA_NVCC_FLAGS –compiler-bindir /usr/local/bin/gnuc4.3)

This should probably be set after find_package(CUDA) has been called. Works like a charm for me, finally :)

Gerade sitze ich an der Korrektur meiner kleinen OOWA-Zusammenfassung. OpenOffice leidet offensichtlich ein wenig unter religiöser Verwirrung, denn nahezu überall, wo eine Domäne auftritt, passiert sowas (klicken, dann groß!):

Weiteres Beispiel: Domänenlogik -> Dämonenlogik! :S

Hence, we get to my First Law of Distributed Object Design: Don’t distribute your objects!

Irgendwie erinnerte mich das an…:

The first rule of Fight Club is, you do not talk about Fight Club!

The good news is that I can say that you should use a Domain Model whenever the complexity of your domain logic is greater than 7.42.

Sehr witzig! :P

Altes Wissen verdrängt neues Wissen!

Im Anschluss folgte ein “‘Ääääääääh…WAS?!” :)

Nach laaaaaaaaaaaanger Zeit dachte ich vorhin nochmal dran, dass ich doch das gute Video zum guten Visual Drumset aus dem letzten Sommersemester noch hochladen wollte. Das lustige Ding hatten der Herr W. und ich zusammen gebastelt und es funktionierte..zumindest ein bisschen, wie man im Video sieht :)

Für weitere Infos: Video anklicken und die Youtube-Info lesen, da ist dann auch ein Link zum Projektbericht.

Man verzeihe meine “ääääh”-Affinität :(

…sollte man gelernt haben. Das teilte mir der Traum der letzten Nacht mit. Inhalt: Tag x war angebrochen. Zwei Prüfungen fanden an diesem Tag statt. Es war 11 Uhr. Ich hatte noch für keine der Prüfungen gelernt und bekam langsam (:S!) Panik.

Verwirrend.

Wer hat sich’s nicht schonmal gefragt: “Wie groß muss mein Programm eigentlich sein, um eine bestimmte Ausgabe erzeugen zu können?”

Wie…Kolmogorov? Ja? Wer noch? Was? Sonst niemand? Soso…

(Dieser Eintrag wurde Ihnen präsentiert von: morgiger (bzw. mittlerweile heutiger) Vortrag mit Hauptthema “On the Length of Programs for Computing Finite Binary Sequences”)

Yesterday I wanted to check the Netbeans tutorial on “Converting and Validating Data in Visual Web JSF Applications” to gain some basic knowledge on this topic (again – as you might already think to yourself – this is due to an assignment ;)). But just like last time, problems were lying ahead: Testing the most basic application one could think of, Firefox (3, that is) refused to render properly when I first built the project. A quick reload of the page fixed this, so I thought “Well, maybe that will NOT happen again!”. At least i hoped so. Playing around with that application, about ten to twenty percent of the page loads led to a totally fucked up (!) page rendering.

I was really willing to blame Netbeans or Woodstock for that, but then I tried another browser and it worked perfectly. Still, the solution to this problem is not a browser change, but just installing the latest Woodstock components (as described here) fixes the issue, though I don’t know the exact cause of it. Ok, to be honest, it was completely satisfactory to me that I got it to work properly :P

Hope this is helpful to someone :)

Da ich gerade meine ganzen Papers ein wenig organisiere (unter Verwendung von JabRef), gucke ich auch ab und zu auf CiteSeer. Dort habe ich zu “Triangle: Engineering a 2D Quality Mesh Generator and Delaunay Triangulator” von Jonathan R. Shewchuk folgenden BibTeX-Eintrag gefunden:

@MISC{Triangulator_triangle:engineering,
author = {Delaunay Triangulator and Jonathan Richard Shewchuk},
title = {Triangle: Engineering a 2D Quality Mesh Generator},
year = {}
}

Lol? :S

© 2010 P=NP! Suffusion WordPress theme by Sayontan Sinha