I've read that Operation Optimization isn't recommended to use. Why?

Something to do with the meshes being wonky somehow? AFAIK, the author basically did manually what PYFFI does automatically though, so you might as well use PYFFI instead. :shrug:

In some cases they removed more than they should have, making the mesh unstable. And yes, you're better off just using PyFFI instead. Operation Optimization doesn't even deal with that many meshes, so the performance improvement is going to be negligible to begin with, never mind the odd game crash.

Well there was a link in the PyFFI thread to someone's experience with PyFFI'ing meshes. This person said that it decreased performance. Got me interested. So I used my vanilla, non-PyFFI'd meshes. My FPS went up outdoors by 10 - 15 FPS. What's up with that?

You're all thinking of Operation PolyGone. That's the one with the numerous defective meshes.

Operation Optimization is an entirely different mod and doesn't have any known issues other than being redundant now due to PyFFI.

If you're talking about Rancen's testing, I wouldn't put much stock in tests done on a system wholly incapable of playing the game at proper performance levels.

Well I'm saying that I noticed an FPS increase without PyFFI. And I have a solid mid-range computer.
Q9450 @ 2.6GHz
2 GTX 480's in SLI

I'm sorry I have a very hard time believing you gained 10 - 15 FPS outside by swapping out meshes. I would like to know your procedure used for testing, and also what variables were involved.

I misspoke. I didn't test it too well before I posted that. I gain peaks of 10-15 FPS. Spikes. But the average FPS gained is 3 - 5. That's outdoors. Indoors I lose FPS by @ 5. Weird.

I find that hard to believe frankly. I gained 5-10 FPS overall when I PyFFI'd my outdoor meshes. That's a pretty substantial improvement. And you're right Arthmoor, I was thinking of Operation PolyGone. But I did try Operation Optimization as well and didn't really see any noticeable improvement for the most part. It deals with so few meshes that it's not going to have much of an impact.

The only annoyance I remember from Operation Optimization is that it makes the base of one of the pillars in house of mania (or dementia) vanish

I don't know. Just reporting the FPS.

Most of what this does is in SI as well. So few meshes on the Cyrodiil side are affected that I can't see it making a substantial difference. The one area it might make a difference is in the Palace District of the IC since it has modified versions of the tower meshes. The god statues are used in so few places that they won't even be on screen in most cases.

I'd also wager that most, if not all, of these files got included in later versions of the UOP anyway so it's probably been made redundant.

Unless someone can EXACTLY mimic all movement and do a video capture of each test to compare I do not think you will ever have a true measure. Only if the perception is big enough than you can say it improved the game or not.

I do not know for anyone else, but I find sometimes that my performance changes enough for me to notice when loading the game from the same save. Not always but sometimes. Possibly a hardware or OS function causing this. So for me this would even further complicate performance comparisons.

My philosophy now is, I have to perceive an improvement/decline to say that what I did made a difference.

You can't really measure a tiny FPS increase (if it exists at all) easily by just walking around somewhere. Too many variables. NPCs with high resolution armor who walk into view during the first test but not the other, general AI stuff, or a background process on your PC could go active for whatever reason (updates), eating away some resources while you run your test, lowering the measured framerate...

It's not worth it for me if there is no obvious improvement, especially if there's a chance it causes issues.

OO won't cause issues, but it's performance gains will be negligible in normal play. PyFFI + The UOP will probably cover things as good, if not better.

I installed it anyway. As long as it won't cause any issues.

You know what bugs me: I have 2 GTX480's in SLI, and I'm getting 15 FPS average outdoors. Less when there are NPC's around.

I often found my performance decreased when runnning in SLI. In fact I disabled it. SLI uses more CPU power and combine that with a CPU intensive game like Oblivion and I just didn't get the net results I expected. Others have reported performance improvements when running in Crossfire (ATI), but I found I got better when SLI was disabled...

Interesting. I'll test without SLI and see what happens. My 2 480's score lower than when I had 1 280 in 3dMark06. May be an SLI thing.

EDIT:

I'll be damned. My FPS is about 4 - 5 higher outdoors without SLI enabled. I paid for the card though, so I'm going to enable SLI and set the rendering mode to single card.