Crashing when generate model above 5mln triangles (SOLVED)
Hi,
I would like to share with you my experience, to save you a bit of time and money. I have to admit, as a person who worked in Support, I approach this problem like an idiot. 2nd reason to share it with you.
When I was trying to get better details and I changed the settings to let say 7.5mln triangles, my PC was crashing with an error "Not enough video memory". Of course the word "video", I read after I bought more ram, hence the word "idiot".
So. Core i7 6850K, 64GB of Ram, 1TB SSD PCIe, GTX 1060 6GB and ... "Not enough video memory".
By a share coincidence, I found on eBay NVidia Tesla M2090 6GB. For £35. It is generally GTX1080, but 8 years old and very power-hungry. As it comes from render farm, it doesn't have a bracket and cooling. When I was building the cooling for it and adjusting the bracket from broken GTX470, I reinstall my drivers for my GTX, I install the new CUDA 10.1 and I also installed MatLab runtime (for a different project). I am on Windows Insider. I read somewhere that Photogrammetry software, may not work on it. I was just to reinstall everything and put the Tesla in it when I decided to crash my PC just for fun.
Nope. It doesn't crash anymore. I am just making a model with 10mln triangles and boost up quality and it is doing it without the hiccup.
I am sort of new to Capturing Reality. I only would like to know if there is a log that I could check what error I was getting, to narrow it down further. So if I will get it again in the future, I will be able to fix it, not by accident. If there is, where is it?
I hope that this short story will help you resolve some of your problems and teach you 2 lessons. It did it to me. Don't give up. Don't spend the money, only because the computer said so.
EDIT: I just checked on which drivers I am on. I am on NVidia Game Driver 432.00. The GeForce Experience is telling me that there is a newer driver for Studio 431.86. I am set to Studio version as default, but I wanted to test something.
I installed CUDA 10.1 - cuda_10.1.243_426.00_win10.
One of these 2 things, solved the problem.
Please sign in to leave a comment.
Comments
0 comments