Jump to content
  • 0

Losing up to 12fps in average after applying the STEP INI Guide - why?


Question

Posted (edited)

Just looking for a short explanation as this time (I gave up my bugged last setup) I benched every step during STEP installation.

 

Compared to the STEP Launcher settings I'm losing about 10fps in average after having done the INI Guide. Initially I thought this was due to the popping tree fix and the 200 instead of 20 rendered animated trees but looking at Brittleshin Pass I'm wondering where this loss comes from. I've followed the guide PRECISELY with exception of the HUD and map fixes (as I'm gonna install mods fixing those things anyway).

I also "gained" about 100 MB of VRAM usage except for Brittleshin Pass, where I've had even less than before. So at least the VRAM thing must be something that is present in the worldspace aswell as in Solitude but not in dungeons.

As for Solitude, I have no idea why I don't run into the GPU limit there, that's probably also the reason nothing changed compared to the last run.

 

Any ideas? Just asking as I know I will NEED those 10fps later when it comes to ENB and vegetation (grass!) mods. ;)

 

Specs are

i7 3930K @Stock

GTX 680 4GB OC @Stock

8GB DDR3 1333 @Stock

SSD

 

Tools are

FRAPS for the FPS

Windows Perfmon Logging for CPU, RAM, Pagefile and SSD Transfer Rate

Nvidia Inspector for VRAM, GPU Load, Temp and Power

 

Here's the results:

 

0 Intro - Having FRAPS run for 240 seconds after the vanilla game intro start

skyrim_benchmark_stepvdskk.jpg

 

1 To Riverwood - Walking from the exit of Helgen Keep to Riverwood and stop benching after I've reached the town's exit

skyrim_benchmark_step2zj71.jpg

 

2 To Riften - Walking from Heartwood Mill to the entrance of Riften

skyrim_benchmark_stepg6j9g.jpg

 

3 Brittleshin Pass - Starting inside at the north entrance, doing a short run through the dungeon to the south entrance

skyrim_benchmark_stepshk44.jpg

 

4 Solitude - Starting at the entrance, walking on the right side to the palace, turning around, going through the Imperial castle and back to the entrance

skyrim_benchmark_step1yjg7.jpg

Edited by elwaps

Recommended Posts

  • 0
Posted

Well, the STEP Recommended Changes and the Other Useful Changes are two different animals. I suggest you start merely with the Recommended Changes before going onto the Other Useful Changes. Think of the later as tweaks that are in a "testing" phase.

 

Probably the one that would tank your system the most, which I thought I made clear as causing a rather severe performance drop, is the Reduce decal fade tweak.

 

Also, I see that in every case you seem to have a minimum near 60. That's optimal. You shouldn't be trying to get more than 60fps out of Skyrim.

  • 0
Posted (edited)

Thanks for your time.

I know that framerates way above 60 make the physics go wild, I've experienced this when benching the Vanilla Skyrim at "low" and "medium" Launcher settings. But be informed that I want to add an ENB, Verdant Grass, Skyrim Bigger Trees (Lite), JK's Major Cities and ETaC. The ENB alone will eat up fps like Homer Simpson eats donuts.

 

Luckily I haven't touched Reduce Decal Fade as its description made me a bit afraid of it. A potential factor of instability wasn't what I was looking for after giving up a messed up installation. So all I changed about decals was the number of decals shown.

 
In general I sticked with the Recommended Changes. To be exact, from the Other section I only took
 
bAllowScreenShot=0
 
fDefault1stPersonFOV=81

fDefaultFOV=81
fDefaultWorldFOV=81

 

fLightLODRange=50000

 

iNumHWThreads=12

 

fGrassFadeRange=10000

 

iMinGrassSize=40

 

fBookOpenTime=200.0

 

fMeshLODLevel1FadeTreeDistance=6144

fMeshLODLevel2FadeTreeDistance=4096
fTreesMidLODSwitchDist=8192

 

iWaterReflectHeight=1024

iWaterReflectWidth=1024

 

All benchmark scenes are at daytime so it is improbable that the light LOD did any harm, HWThreads should if anything improve performance (tho there's almost no serious CPU load anyway), grass fading and density shouldn't apply to Solitude, same goes for the trees and there's no water in the dungeon. That's why I was heavily confused.

 

But nevermind, in the meantime I've applied STEP 2C (ENBoost, SKSE, Papyrus, StableuGrids) and it ate another ~5% performance. BUT reduced the stress on the SSD (~2 MB/s instead of ~4). If this goes on like that I'm gonna end up with a negative framerate once the build is done. :D

 

By the way, is it possible that I had frequent freezes with my last build because I used the XP32 Extended skeleton while STEP says "do NOT use it"?

Edited by elwaps
  • 0
Posted

You probably need to play with the ENBoost memory parameters a bit to gain back performance there. 

 

The mods in STEP aren't as taxing as you might think, and there are a lot of performance-boosting mods included. I think at one point I concluded that running STEP Baseline was about the same as running Vanilla + HRDLC. I run it on an APU with relatively little performance drop.

  • 0
Posted (edited)

Another idea why my old setup was instable as hell. I've found out that I've used the ENB Memory Checker for DX11 and got a value of almost 30000. This time I've run the DX9 one and got closely over 4000. Any hints on how to gain performance in general via ENB? Or should I just finish my setup and try afterwards?

 

I've just finished installing and benchmarking 2D (Fixes). Lost a little bit but should be okay. Except for the Intro and Solitude.

 

/edit

Realized that Media Player Classic Home Cinema with a paused SD video was still running in the background. This should explain parts of the framerate loss from 2D. 2E looks way better.

 

 

0 Intro - Having FRAPS run for 240 seconds after the vanilla game intro start

0hnu5s.jpg

 

1 To Riverwood - Walking from the exit of Helgen Keep to Riverwood and stop benching after I've reached the town's exit

1aluw7.jpg

 

2 To Riften - Walking from Heartwood Mill to the entrance of Riften

2qou77.jpg

 

3 Brittleshin Pass - Starting inside at the north entrance, doing a short run through the dungeon to the south entrance

353u83.jpg

 

4 Solitude - Starting at the entrance, walking on the right side to the palace, turning around, going through the Imperial castle and back to the entrance

4r3u4g.jpg

Edited by elwaps
  • 0
Posted

Not sure if there'd be a difference when you decided to tweak the ENBoost settings. That thread is huge. Basically you got to test it like you're doing and figure it out, as it seems to be extremely system specific.

  • 0
Posted

Allright, lets see how many fps I got left in the end then. ;)

 

Update, made it to 2G. The last two sections basically just added lots of VRAM usage.

 

0 Intro - Having FRAPS run for 240 seconds after the vanilla game intro start

0vrjag.jpg

 

 

1 To Riverwood - Walking from the exit of Helgen Keep to Riverwood and stop benching after I've reached the town's exit

13dkc3.jpg

 

 

2 To Riften - Walking from Heartwood Mill to the entrance of Riften

2ozjh2.jpg

 

 

3 Brittleshin Pass - Starting inside at the north entrance, doing a short run through the dungeon to the south entrance

3skjhn.jpg

 

 

4 Solitude - Starting at the entrance, walking on the right side to the palace, turning around, going through the Imperial castle and back to the entrance

4kzjb0.jpg

 

ALL BENCHMARKS COMBINED

5orjhv.jpg

  • 0
Posted (edited)

Indeed. Still tons of VRAM left to fill with visual mods.

 

I'm almost done with STEP Extended by now, I've only left out Dragons Shout with Voice, regardless of lore I think it sounds way less frightening.

 

However, now as DynDOLOD Worlds is running I got a question before benching the last time.

It only works by deactivating and reactivating BirdsAndFlocksHF. Before I deactivate it, the esp is priority #40. After reactivating it got the highest priority in the load order. Should I put it back to 40 or leave it at the bottom?

 

/edit

Put it back to 40, seemed to make sense to me. Now here we got the performance changes during the STEP installation packed into some fancy charts ^^

 

stepperformance0combid4jcy.pngstepperformance1introxekdo.pngstepperformance2riverl7kse.pngstepperformance3riftyxkra.pngstepperformance4brittkkjsr.pngstepperformance5solitt9st4.png

 

While beginning the installation of REGS I've also already found out what I'll be needing all that performance for. JK's Cities alone eats up 18% performance and increases VRAM usage by 18% in Solitude!

Edited by elwaps
  • +1 1
  • 0
Posted

Thanks, I'll try that. I would guess you wouldn't notice it that much anyway as in JK's cities and ETaC's towns there's so much detail to look at and as in the wilds there's grass everywhere. That's because I've added Verdant in the meanwhile, which killed what was left of my performance. ;)

  • 0
Posted (edited)

Looking through the tweaks installation list I saw that they decreased frame rate by ~42%. Is there a recommended frame rate that the game should be run? I thought that higher was better.

 

I've been experiencing random CTDs. One time I was sitting still on my horse on a small back trail about half way up a mountain enjoying the view.  I thought that I would try some of the ENB efficiency mods/patches ( ENBoost ? ) to see if that would help.

 

I have an AMD Phenom II 3.2 Ghz cpu, 8gig DDR3 ram and a Gforce GTX 550 Ti video card ( 1gig ram ). I should receive a GTX 750 Ti ( 2gig ram and Haswell chip ) tomorrow or the next day. The reviews say that it has twice the performance of the 550 Ti. I have the game installed on a 120gig SSD and the overwrite, mods and webcache folders installed on a 160gig HDD.

 

I only have ~80 mods installed and never have them all activated, depending on the character type I'm playing. Usually about 75. I've been playing Skyrim since it first came out but have been playing it modded for about a year. I'm kinda behind the curve.

 

I'm going to install performance monitors and do some testing on my current video card and then install the new one to see for myself what the difference is. My other question is: do you have a standardized testing procedure that I can use so my testing scheme can be consistent with yours? I would be more than happy to submit my results if you are interested in reviewing them.

Edited by BuddaBelly

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

By using this site, you agree to our Guidelines, Privacy Policy, and Terms of Use.