ENBoost, Parallax & Performance

[Nearox]

I recently got back to using ENBoost, mainly for the parallax effects. However, no matter what I try there is always a 5-12FPS loss (with STEP:Core install). Some effects are still enabled. I read a post by Boris in response to the author of project parllax remastered, and he indeed confirmed that is not possible to single out FixParallaxBugs. Some kind of shaders will still run. 

 

https://enbseries.enbdev.com/forum/viewtopic.php?f=17&t=2215 

 

Does anyone happen to be using a different version than the latest 0.233 that does not have such a huge performance hit? Or some tips? See my specs at the signature.

 

I have literally put every single setting in enbseries.ini to FALSE (except for Useeffect=true, which must be enabled for Parallax to work), deferredrendering is off, vram settings etc. were done following that online tool, aliasing is off from both videocard and enb settings etc....everything is off except FixParallaxBugs...

[Aiyen]

Odd I only noticed a performance drop if I have water parallax enabled in the latest version. And that only if there is water nearby.

[Nearox]

Hmm I downloaed 0.119 and also the parallax remastered enb config files. Now there's hardly any performance drop anywhere because all the effects are properly disabled. 

 

His config is made for enb 0.116 and upwards and hasn't been changed since. I suspect that newer ENB's have more functions which are enabled by default and which can only be disableb by tweaking the numbers instead of true/false parameters. 

 

Either way. Fixed now :P Thanks :)

[Aiyen]

Cannot reproduce the issue you describe... I have no FPS hits that depend on the parallax alone. I get more performance if I disable most of the stuff, up until the point where there is only the overhead of having ENB enabled, which on my machine is hardly noticeable.

[Nearox]

Yeah such overhead also used to be largely unnoticable as well for me, but on older enb versions. I didn't use ENB for a month so I got a bit out of the running. But I have fiddled quite a bit with them before and I'm sure that I did disable everything. Parallax itself doesn't have any FPS loss, or at least it is less than 1 FPS on my system. 

 

Although I would like to use a newer enb due to the new fixes , using 0.119 has its advantages: There is full hardware AA support. The other disadvantage is that Boris started building in ENBoost functionality from 0.221 onwards, so I unfortnately don't have that. However in the last 3 days, using Safety Load, I have not crashed a single time - which is a personal record :P 

 

Would stil likle to use ENBoost on top of it, but since I'm testing a pack with parallax stuff and it is stable atm I guess I can leave it as it is for now. It's just so strange that I can't for the life of me figure out how to get it to work with the latest enb. Been like 2 hours of trial and error now, and the fps loss is still the same. I don't mind a 5 fps loss but 12 (or sometimes even 15) is unacceptable. 

[Nearox]

I get more performance if I disable most of the stuff' date=' up until the point where there is only the overhead of having ENB enabled, which on my machine is hardly noticeable.[/quote']

 

Over the last week, I've tried a lot of stuff... I'm still using 0.119, which has no performance hit, but anything above that and I get significant fps drops. The fps drops get worse as the version increases. I really don't get it, I must be missing a single setting somewhere. Would love to use a more recent enb version with enboost and some other game fixes.

 

Aiyen, could you share your enblocal.ini and and enbseries.ini with me for your config "where there is only the overhead of having ENB enabled'? 

[Aiyen]

Okay did an extensive set of tests of versions.

119

168

233

234

 

Setting useeffect to false has my FPS stuck at 45,3 near riverwood as it should be. (I play at 1440p)

 

This is the only setting you need to enable or disable to get the overhead, which is minimal.

 

One thing you need to remember when testing v119 is that AF is not enabled by default and is only set to 8x not 16x.

This difference is about 5-10 on my machine at my resolution.

[Nearox]

Thank you for testing.

 

However, the issue is about the parallax effect... Uuseffect must be set to true otherwise parallax effects do no work. This is what I was referring to all the time. Sorry if it wasn't clear.

 

When useffect=true but I disable all other effects in enbseries.ini, the FPS drop is still significant... That is what I am wondering about.

With version 119, the FPS drop is virtually nonexistent. With version 234, I get about 15 FPS drop. 

 

Could you perhaps test again with useffect=true, fixparallaxbugs=true and all other things disabled in enbseries & enblocal? 

 

Thanks in advance!

[Aiyen]

Nah I did understand what you meant, I just wanted to make sure that the overhead of just having ENB in was not different from version to version. Which it is not.

 

There is a slight decrease using the latest version however. This with only effect on, and everything else disabled, and only a single lite enbeffect.fx file as additional shader.

The difference is roughly 4 FPS for me on average. (I do not count large drops from moving around etc. These values are just from standing still and turning the camera around.)

So instead of 43 FPS i get 39.. and instead of 50 I get 46. It is true that this difference is smaller back in v.119 however.. there it is between 1-2 FPS on my system.

 

The ENB AF takes up roughly 1-2 FPS when set at 16... I cba to test if the driver based one can do better that then.

 

This is still with ENBoost activated.

If I disable all graphics so its only ENBoost, then there is no drop in performance, as expected.

 

So conclusion is that there is a drop when using the latest versions with just parallax... but I can only record in the order of 5 FPS at the most during stand still testing.... I cba to do a full running around checking for stuttering etc. since that is vastly more complex.

[Nearox]

I see! So there is something definitly wrong with my setting somewhere...

 

Sorry to ask somethign again, could you perhaps copy/paste the contents of your enblocal.ini and enbseries.ini in a spoiler/code tag? 

[hellanios]

What settings do I have to turn on/off in order to get just the parallax effect?

[Aiyen]

Everything under EFFECTS in the enbseries.ini. Then you should not loose any significant amount of FPS, and still have parallax.

 

fixparallaxbugs should also be set to true under FIX in the enblocal.ini. It is set to this by default.

 

Hope that answered it! :)

[hellanios]

Everything under EFFECTS in the enbseries.ini. Then you should not loose any significant amount of FPS, and still have parallax.

 

fixparallaxbugs should also be set to true under FIX in the enblocal.ini. It is set to this by default.

 

Hope that answered it! :)

Can I use the .221v that Osmodius has updoaded with the latest binary(239)? Will it do the same thing?

[Aiyen]

The steps I mentioned will also work for the latest binary version yes.

[Spock]

So conclusion is that there is a drop when using the latest versions with just parallax... but I can only record in the order of 5 FPS at the most during stand still testing.... I cba to do a full running around checking for stuttering etc. since that is vastly more complex.

On my system UseEffect=true, all individual effects disabled costs me ~8-9 FPS in some areas. Especially in the forest near Riften.

[keithinhanoi]

With UsePatchSpeedhackWithoutGraphics=true in enblocal.ini you should also see Boris' shadow fix, even if UseEffect=true and all individual effects are disabled in enbseries.ini. Not the Detailed Shadows effect, but the general improvement to Skyrim's "blocky" shadows.
 
If performance is the goal, in my testing I've found that by making sure all shadow resolution settings are at 512 in skyrim.ini / skyrimpref.ini, I still see the enormous FPS / performance gain outdoors (exterior cells) that I would get with no ENB active, on my fairly underpowered GPU. However, ENB takes away (or hides?) enough of the blocky edges at 512 resolution that I can accept the tradeoff between shadow quality and performance.

 

Here are my exact "performance friendly" shadow settings, for reference:

 

 

In skyrim.ini:

[Display]
iShadowMapResolutionPrimary=512

In skyrimprefs.ini:

[Display]
bShadowMaskZPrepass=0
bDrawShadows=1
bDeferredShadows=1
; (Note: the following 4 settings can be disabled for performance gains)
bShadowsOnGrass=1
bDrawLandShadows=1
bTreesReceiveShadows=1
bEquippedTorchesCastShadows=1

iShadowMapResolution=512
iShadowMapResolutionPrimary=512
iShadowMapResolutionSecondary=512
iShadowMode=3
iShadowFilter=3
iShadowSplitCount=2
iShadowMaskQuarter=4
fShadowBiasScale=0.4000
iBlurDeferredShadowMask=4
fShadowDistance=7000.0000
fInteriorShadowDistance=8000.0000
fShadowLODStartFade=200.0000

In addition, I have not yet tried this test, but in my research on the enblocal.ini settings, I've found posts claiming that setting UseDefferedRendering=false can bring slight performance gains even if all individual effects are disabled in enbseries.ini.

UseDefferedRendering toggles ENB's deferred shading rendering, which is required for a good number of ENB lighting effects, including ambient occlusion (SSAO), image based lighting, reflections, particle lighting and skylighting. If you're not using any of those, then it's perfectly safe to disable UseDefferedRendering - just be aware that you will see a red warning message in the main menu when you start Skyrim telling you that Deffered Rendering is disabled (similar to the ENBoost only message.) So, this is something else to try in any effort for a performance gain.

Personally I would much prefer to use binary 0.251 due to all the ENBoost-related optimizations and fixes that Boris has incorporated since that feature was introduced.
 
I am in progress with testing, so if anyone here is interested, I can report my performance-related findings when using binary 0.251 back here.

Edited March 25, 2014 by keithinhanoi