<div dir="ltr"><span style="font-size:12.8px">Hello all,</span><div style="font-size:12.8px"><br></div><div style="font-size:12.8px">With some self-written plugin I am generating rather large "unstructured grids". With a model with something more than 700'000 blocks I realized that on my Lenovo notebook I did not get a complete display of all the blocks, so I made tests with subsets of the data. It turns out that with less than 1000 blocks everything works fine, with 20'000 blocks there are already some "reductions" and finally with the full model even more: see the attached Powerpoint with screenshots (plus details about my system and the Paraview version).</div><div style="font-size:12.8px"><br></div><div style="font-size:12.8px">My question is now if there is possibly some kind of "graphic card overflow" happening? I could so far not test the software with the same model on another computer.</div><div style="font-size:12.8px"><br></div><div style="font-size:12.8px">And whatever the reason is: is there something that I can do about it?</div><div style="font-size:12.8px"><br></div><div style="font-size:12.8px">It is to say that the unstructured grid is a bit "unconventional" by the fact that often adjacent blocks do not have common edges and points, sharing thus only common partial faces. With this one possibility would be that Paraview cannot properly carry out certain optimizations that are relying on the fact that neighbor blocks "normally" (??) share a face, edges and points. However I do not know whether this is an issue or not!? In this case it might be an option to change the entire data set to a set of just cubes that happen to touch each other.</div><div style="font-size:12.8px"><br></div><div style="font-size:12.8px">Regards, Cornelis</div><div><br></div>-- <br><div class="gmail_signature">Cornelis Bockemühl<br>Basel, Schweiz<br></div>
</div>