After some tweaking and testing, I finally found out what bFloatPointRenderTarget actually does. I've heard other people saying that there is practically no difference, neither performance-wise nor visually. That is why I made this thread, to enlighten people what I just found out. If this is old news to you, then I'm sorry, but it's completely news to me

Firstly, I have to say that if you have ENB series, you MUST have bFloatPointRenderTarget=1. There is no choice then.

Now, to get the point, bFloatPointRenderTarget has a minimal difference in general, but there are some thing that change.
Default value is bFloatPointRenderTarget=0.
Changed value is bFloatPointRenderTarget=1.
It can be found in SkyrimPrefs.ini.

1) When bFloatPointRenderTarget=1, FPS is lowered (on my GTX 260 card) by 2-5 FPS depending on where I am. That's pretty much.
2) When bFloatPointRenderTarget=1, shinyness on certain things is changed. Shinyness on objects is minimally changed. The biggest change is shinyness on water.
Here's a comparison showing both FPS and shinyness difference:
Example 1:
bFloatPointRenderTarget=0: http://oi39.tinypic.com/3149gzp.jpg
bFloatPointRenderTarget=1: http://oi42.tinypic.com/2e35yeg.jpg

Example 2:
bFloatPointRenderTarget=0: http://oi40.tinypic.com/4ihbio.jpg
bFloatPointRenderTarget=1: http://oi40.tinypic.com/2i6kqz5.jpg


You can save a lot of FPS on this with just a slight visual difference. Definitely worth making sure bFloatPointRenderTarget=0 in my opinion, unless you have ENB; then you must have it on 1.

You have picked two rather uninteresting scenes. From what I understand is the option important to the shader quality and will therefore not change much visually when you only look at static objects like the landscape or the town houses.

Take a close look at your second example. There is a small fire just next to the horse's head. The option should change the quality of the fire animation and other similiar effects. Magic effects, water reflections and refractions and heat refractions ...

The difference between using integer and floating point numbers for such effects is the precision with which they get calculated. A higher precision will allow for more smoothness in an effect, because of a reduced error propagation.

I just compared magic effects and fire animations. No difference at all, so I think you're wrong about that.
Water reflections though, a noticable difference, which I showed in my comparison screen. If you think you can capture more "interesting" screens, feel free to actually contribute.

So having bFloatPointRenderTarget=0 is recommended. 2-5 more FPS, and a small difference in water reflections shinyness.

I've already mentioned this..

this effects the Floating Point Shaders from FP12/16 to FP24 or potentially higher..

it increases the range of it's dynamic level significantly.. and on some video cards depending on the resolution and of course other factors can either have very little impact on fps.. or result in a massive impact on performance.

Visually there "should" be an improvement.. however as with your screenshots.. it's REALLY hard to say for certain While the shininess of the water is different, it's debateable if that is better or not or if the sunlight/clouds may have an impact.

Actually i wouldn't be surprised at all if setting a higher dynamic range has no significant impact due to not being able to actually make good use of those additional values... and there is A LOT of them.

Most video games stick to FP12/FP16 because it's the best optimised balance of quality and performance.... fp24 just costs far to much to do it properly at the moment.... Some of this could potentially be blamed on NVidia's methods of floating point dynamic range. But majority is due to consolitis.

LOL, NVIDIA just about always cut corners on their GPU architecture, like they did with pixel shader 2.0 on FX cards where the performance was terrible.

I have a pretty old GPU (GTX 260), can someone with a more modern GPU try and see your FPS/visual difference? Would be interesting to see how much this is GPU related.

I was not arguing with you, I was explaining to you what it means since you are only looking at fixed scenes. It is about animated effects so you will not see much by comparing screen shots.

I've done tests with a 5770 and 6950 with no decernable difference by freezing time using the console, freezing animation via the console as well.. so that the game world doesn't change or move.. and allowing for free cam to occur... taking screenshots at various angles and also recording fraps to see if i could do a comparison.

There are a few pixel differences.... but it's VERY VERY Bloody hard to tell, mostly just very suttle brightness differences.

This is of course running at 3240x1920 so i can definitely do per pixel detailing..... Using adobe photoshop is one good way to compare 2 identically taken screenshots and compare the differences, adobe can isolate those pixels.. and then i can do hue/brightness comparisons to see the exact differences.

It's so extremely suttle (excluding the water) that no way anyone is going to really see a benefit.

The game just wasn't designed for the higher precision/rage level.

It's definitely not limited to animated effects.... one doesn't NEED to record video to see a difference.... everything has to be animated and drawn frame for frame, so taking screenshots will most definitely show off the suttle differences.

Only a trained eye will see it, I agree. Noobs can turn it off.

No. Flames shift and flow and you will have a hard time capturing a still image of the exact same flames to compare them on a pixel-by-pixel basis.

No one is saying you cannot turn it off...

I've done the flame animation and shift...

cast a spell... console lock the animation/effect and take screenshots... do this a few times throughout the animation of the flame effect so that multiple reference points are made to attempt to capture a good number of points to compare.

I did it with lightning too.. as most often the lightning skills or eletrical/spark skills have a good amount of floating point ranges.


I admit it... i'm a bit of a freak in terms of graphics settings and such... for the first month i played with settings and such more than i played the game.

Me and DHjudas are cut from the same crop because that's what I do as well.

The visual difference is very small yes. But what about the performance difference on your cards? Did you notice a pretty big performance difference like I did (2-5 FPS)?

For the small brightness difference, having 2-5 more FPS is super worth it for me on my computer...

considering the various resolutions i run at.... my 6950 experienced about a 10-15fps drop which is pretty significant... it's about the same at the same resolution on the GTX 580 here too.

running 1080p it's about 2-5fps.... so nothing to huge.... specially unnoticeable if i'm pegged at well above 60fps lol..

However on a HD4830 512mb with it enabled at 1080p, the frame rate took a hit so bad resulting in sub 10fps....

My poor 9600GT I've got here also took a beating by enabling it.

Correction.. it's "DHJudas and I"

I know what you are saying. It is when you get an arrow to the knee.