DRL - Is this normal or did I break something

#1 I know from our testing that DRL has a relatively large impact on fps, and it seems worse on nvidia cards, but the ratio shouldnt be that high... It's probably driver dependant too.

Try testing it without HDR and see what the difference is.

addikt

#2 It was our discussion in another thread that sparked my experimentation. I know it has an effect on performance... but wow.

Drivers are up to date. It could be the card. Anything that isn't in the x800's for nVidia is slightly gimped. Unfortunately after installing my 6800 is now (to my surprise) an RMA number for BFG so I won't be able to test it for a week or so.

#3 I assume it also depends on how many things are been affected by the lighting, you might find that it multiplies the fps drop based on the number of model surfaces that need to be calculated...

I tested with just atlas terrain and 1 waterblock and my fps hit from DRL is only 150 down to 135 (give or take) on a 7800GTX... We are still testing out different things before we populate our world, as the change in fps can be very dramatic on different systems... so we kinda wanna establish what we can and cant do.

I replaced all our additional waterblocks with LODwater for now, still gotta figure out how to swap the dts flat plane with a waterblock ;) but it allows us a pretty solid fps base to test from..

I found this old post interesting: www.garagegames.com/mg/forums/result.thread.php?qt=54159 Not exactly relevant to us, but the fact that the nvidia cards got worse by a greater amount with the new engine is weird.

addikt

#4 In the first image you have HDR enabled - depending on the hardware's floating point buffer performance, the hit can be rather high.

Fortunately the HDR/DRL systems are designed to be identical - DRL can handle 99.9% of the levels created, HDR is only necessary in extreme low light conditions, where increasing the exposure can create artifacts.

What frame rate do you see when running only DRL?

#5 It does vary some, but not what I think you'd be expecting... 1-5FPS maybe.

I've made some pretty extensive modifications to TGEA, but nothing that would seem to even bump up against DRL functionality.

Just to verify I compiled a fresh HEAD of TGEA 1.0, tested it with differing combinations of DRL/HDR, and the results were the same. HDR had little to no impact on FPS, but with only a skybox, fog settings and DRL, performance goes from 160-180FPS down to around 60FPS.

#6 Thats good to know John.

Anyway to relate my story. I have never seen an FPS drop like that with a card. I have a 7900GT, and the difference between HDR and DRL isn't enough to even be picked up by the fps counter in most situations. The DRL will cut off 5-20 fps, very similar to addiktive's numbers. Addiktive/Cogburn: What drivers are you using? I'm using 93.71, it could be some problem with the latest nvidia drivers, I could believe that :)

#7 Im using 93.71 drivers too. I find they are the most stable and seem to work best for most games at this stage...

addikt

#8 I reverted to 84.63
The fps impact with "higher end effects" on 90 series is inacceptable on my 7600Go (7300GT)
Beside the fact that the 93 series, while beeing stable, even went that far to disable Shader 3 support (found out by accident when trying out some shader effects in shader studio and NVs shader creation suite)
The most current drivers (97.XX) are that slow that I am unable to play SupremeCommander even in minimal settings (with 84.63 I get around 30 FPS, compared to ~10 with 97.XX drivers)

PS: Did you enable tone mapping in the prefs? Because that one costs a fair amount of FPS.

#9 Tone mapping, oddly enough, has an impact akin to that of HDR... virtually nothing.

I'm wondering if it isn't my driver version... I'm running 91.47, which is not the best.

I'm going to goof around with installing different driver versions tonight. I'll let you know what I find.