I really think you may be onto something with this one. Like some other people in this thread, I got my Mini to setup as an HTPC (with Plex, in my case).
I have the low-end, dual core i5.
I started out setting up my new Plex with changing out the stock RAM for two 8 GB DIMMs. I ran the system migration tool to move settings from my old Mini, did the Software Updates that were available (10.8.2, principal among them), then installed Plex Media Server and the OSX Plex client on it. All my media is stored separately on a ReadyNAS NV+. All the Mini does is serve up the content. Plex will sense if you have the media you are trying to catalogue on a network drive and prompts for credentials needed to access it so it will make sure the connections stay alive and mounted. Did that, and the drives are getting mounted and staying mounted after PMS launches. So far, so good.
However, once I set out to catalogue the various locations using Plex Media Manager, or play any of the videos it did manage to catalogue, I'd get a kernel panic, or something that looked like them within 30 to 60 minutes every time. The screen would get a bunch of colored boxes in various places over what was left of the image that was displaying, and the box would be completely inaccessible through VNC, ssh, ScreenSharing, pings, and anything I'd try. I'd also get the blank screens and snow. Unplugging and plugging the HDMI cables in again did nothing. It would actually produce the snow and screen blanking, post-kernel panic. Turning the LCD on and off again likewise did nothing.
I also tried looking through the Console for clues, but didn't see much that appeared to be related to this in any way. I had three instances of a Plex Media Server crash there, but nothing reflecting the volume of actual system failures I was experiencing. I also tried completely shutting off Spotlight indexing, in case that was interefing for some reason. Didn't help. Tried an SMC reset. No improvement.
I finally pulled the DIMM out of Bank 1 as your link suggested. I haven't had a kernel panic, HDMI dropout, snow/blank screen since, and it's now been nearly 18 hours where previously it was an hourly occurance.
So, BDAqua. I'm curious, can you explain maybe why having multiple RAM channels filled would maybe make all the difference? Is it some bad interaction between the dedicated VRAM, GPU, and paging out to system RAM when that gets maxed out during certain operations? I really don't understand the hardware and OS well enough to know down at that level and probably sound like an idiot.
I got on the phone with an Apple tech this moring and described all of this to him. We checked the health of the disk and did a permissions verify and all was right with the world there. He waited while I catalogued a large collection of TV shows on my NAS and it sailed right through. Before pulling the RAM in Bank 1, that would have gaurateed a crash a little less than half way through that operation.
Interestingly, entries in the Console have also calmed down in frequency quite a bit.
I'll allow that this may still be somehow be Plex related, but since pulling a DIMM seems to have completely resolved this issue, I'm leaning toward this being something more likely hardware- or OS-related.
Has anyone else out there tried removing the top-most stick of RAM (the one on top when you open the case) and met with success other than me?
I'm currently running a catalague on my NAS-based music library, which contains tens-of-thousands of small files and has been running for over two hours, and haven't noticed even the slightest hiccup. I'll report back here with the results. I can get all the way through that, and playback a bunch of video for, say, three or four days without issue, I think I'll be writing this up and posting my story to http://www.apple.com/feedback/
I'm also going cross-post this to the Plex forums and see what they have to say as well.
I'd be interested to hear from anyone else what they've tried and if it worked.
Thanks!