June 2015 archive

dtk platforms

Today, several software engineering projects are being developed using dtk architecture (see former post) leading to specific platforms dedicated to very different scientific fields: num3sis, axel, medinria, carbonquant, enas, in@lgae, pib, windpos, sup, fsd3d++. Here follows a brief description of scientific goals for each of them.

Continue reading »

Trick of the day: monitoring the Linux filesystem changes in “real time”

Ever wanted to monitor the filesystem changes ? Since Linux 2.6.25, the kernel provides us with a signal driven I/O notification API and the inotify-tools RPM provides a set of tools to interact with the kernel. inotify in action: a simple way to test this API is to run the following command in a terminal …

Continue reading »

Atelier dtk-distributed

Les devexp 2015 ont été l’occasion pour nous de faire un atelier sur dtkDistributed, dans lignée du DreamTech: dtkDistributed. La présentation est maintenant accessible en ligne . Le tutorial proposant de développer l’algorithme PageRank se trouve sur github.

Trick of the day: prefetching a callback into the QApplication event loop

Qt provide a standard mechanism (aboutToQuit() signal) to insure that a callback is executed at the end of the QApplication event loop. This is typically used to add some cleanup code to your application. Infortunately, Qt does not provide a “aboutToRun()” if you want to execute an initializing code just before the event loop. Nevertheless …

Continue reading »

The LD_PRELOAD trick

If you ever wanted to replace some symbol in a library, in order to inspect what happens in a binary you don’t have the source code or even to inject some code, this can be done using the LD_PRELOAD trick. To use it, invoke your binary as such: $ LD_PRELOAD=somelib.so executable.

Continue reading »

FusionForge 6.0 Upgrade

The Inria Forge is fully upgraded (OS and FusionForge) today, Tuesday June 2nd. The service will be unavailable for a few hours during that day. Hopefully, the service should be restored for 12AM. We advise you to anticipate this forge upgrade, especially if you are close to a deadline. Changes in the upcoming 6.0 version: …

Continue reading »

The Bulk Synchronous Parallel model for graph processing

The BSP Model The Bulk Synchronous Parallel (BSP) is a model for designing parallel algorithms. The BSP model was developed by Leslie Valiant of Harvard University during the 1980s and published in 1990. Computation proceeds as a sequence of iterations, called supersteps. Each superstep involves a phase of parallel computations, that emits messages. Messages are …

Continue reading »