Dimajix Developer Blog

It is Winter again in Frankfurt! During the last 5 years that I have been living in Frankfurt, I never experienced that much snow in the city as today. Although people from other regions might say that 4cm are ridiculous, but for Frankfurt, that is a lot. SO I guess I will go for some bicycling later, together with my camera and take some shots of the forest in white.

This week, there haven't been many updates in this blog, my apologies. I have been thinking about some WLAN problems, I have been working on frustum culling (which is almost working perfectly, I only have to refine the frustum-box-intersection test) and I have encountered some difficulties, which seem to be related to my last Nvidia driver update on my notebook.

This last topic is really strange: Suddenly one of my test programs stopped to work on my notebook with Nvidia graphics, although it is still working fine on my PC with ATI graphics. I tried to debug the application, and I found something really wired: The program allocates a bunch of index buffers, each getting an even id on the Nvidia graphics card. Then the optimiser tries to merge all index buffers by allocating a new one - which for some odd reason gets an odd id from the driver. After releaseing the first bunch of index buffers (which aren't needed any more, as their content has been copied into the last one with the odd id), the big optimised index buffer seems to be deallocated, too by the driver.

On the one hand this seems to be a driver bug, but on the other hand I cannot believe that such a simple OpenGL command sequence will evocate such problems. Maybe I try to tackle it down, by writing a much simpler application that tries to reproduce the odd behaviour of the driver. Or I simply reinstall an older driver and see if the problem persists.

Kaya