Linux Mint, Houdini and ATI

As part of my “3D journey” I came to a point where I feel the need to start investigating Linux. That particular point was exactly when I started making some heavy FLIP simulation. My “workstation” is actually an enthusiastic gamer machine: CPU i7 4770K overclocked and liquid cooled, 32 GB of RAM, AMD 7980 with 3GB and so on.

Now, for heavy FLIP simulation you need a huge RAM amount and even if 32GB seems a lot, actually they’re not. So, where to go at this point? Simple: add more ram! Wrong! The i7 47** family can handle 32GB of RAM as maximum so it’s useless to go that way, and if I want more RAM I have to change Motherboard, CPU, and obviously add some RAM and that’s a lot of money, at least for me!

In the meanwhile I’ve read good news about how Linux can handle the RAM. Let’s say you have a big FLIP simulation that uses 32GB: that same sim in Linux will use 21/22GB. For me that’s a good way to optimize my investment. So I started installing Linux distros in my pc and playing with that. Install Houdini, Ati drivers (mesa, radeonsi, proprietary drivers) see that isn’t going the way I wanted and entering a loop that ended only when all the combo works great!

The result was: Linux Mint 17 KDE with AMD catalyst 12.9 and H13, last production build (waiting for H14!)

Linux Mint is based on Debian and Ubuntu so the community is huge, and as I am totally new to this, it’s a big point! As the times passes and new kernels are out, I ended up upgrading my kernel to version 3.18 but then my catalyst drivers wont work, and there’s the solution.

UPDATE August 2015: I just bought a new SSD, a Samsung 850 EVO 1TB and it’s the perfect moment for a fresh install. This time I chose Linux Mint 17.2 “Xfce” because of his light graphic environment. It comes with the 3.16 kernel and the proprietary AMD video drivers installs flawlessly. After the boot the OS uses just 650MB of RAM and at the moment I’m writing this update it’s running 7 houdini batch converting particles to vdb to geo … with no lag or slow down of any kind (consider any of those batch renders will use 100% of the CPU if leaved alone xD).

I hope this can be useful to someone! I think I’ll use this post as my Linux diary 🙂