Jump to content

Strangely Smooth Framerate (Nvidia)


ZeroAndOne

Recommended Posts

43046384875_7cd0f2c65b_o.jpg


This came out of one of PC gamer's online articles, I couldn't find it (forgot the name too). But basically it centered around attaining a smoother framerate for Witcher 3 I think. It never really worked that well for FO4, although nothing really worked all that well for FO4 except for buying a very expensive video card or turning all the settings/config way down. :sad:


The article used Riva Tunner Statistics Server from Guru3D.com then mainly Witcher 3 for the test, but it hypothesized that framerate could be controlled in other similar games. I've run FO4 in various configurations, but it's never run all that well. So I kept messing with it. Ha hah! This last configuration was super smooth compared to the many other ways.


This test theory config was a 4 core Sandy Bridge intel i5, that's the really old now, i5 2500, it's not even Ivy Bridge, it's embarrassing actually, it's 4.017 - 4.223 GHz, you know crashing the O/S & applications as much as possible. But locked core, so it's a bus overclock, though this makes it pretty cheap to play FO4, should be tons of Sandy Bridge i5's around.


It was run with Subtle ENB & FO4's TAA +FXAA & locked to 4096 VRAM to match the video card in order to keep the game out of much slower virtual ram. This was a 960 GTX video card. I used a 1 (high) settings for textures & 4096 Shadow texture map, 16384 Shadow Draw, 8192 Shadow DIR, and 2048 Shadow LOD. Volumetric/GodRay was disabled, to use the shadows.


Vanilla texture streaming distances & vanilla textures (non - HD DLC) a few hundred mods running, bunch of quests, around 40+ hours into the game. I wanted to stress test it which is where the insanely huge shadow upgrade came in. Normally I don't run anything all that fancy, in fact I do fine with a 2048 S-map @ 2048 S-Draw, 2048 S-DIR, and 128 S-LOD.


Using a high setting for textures, low shadow, and Vanilla streaming distance all help the game to run larger chaotic battle scenes, as a trade off. I couldn't tell the difference with these turned down or off. Volumetric lighting (Crepuscular) & GodRay disabled, because there isn't much difference, but it all serves to lower the performance requirement.


Riva Tunner Statistics Server v7.1.0.15378 was used & set a framelimit of 30 FPS. To do that iPresentInterval = 0 to turn Vsync off inside FO4 & Vsync turned off in the video card & frame limit inside ENB local disabled, then triple buffering turned off (from the article) to get the correct buffer amount. Then this time we also set iFPSClamp = 30 (no idea if it still works).


I don't even remember what iFPSClamp did back in the FO3/FNV/SKY era. But I remember it centered around the physics & it was before we had much control at all how Vsync was or was not applied. But who cares, cause it worked. The industry standard is 30, so I started there, but it could work up to 40-50 FPS as a setting depending on how it performs.


The stress test started in Sanctuary & had the character sprint (thanks to 500+ action points) all the way down past Diamond City & the other actors were fighting the whole time nearly especially downtown, which is when we dropped 700 clutter items to test the physics (to see if physics applied smoothly in large groups, or if the iFPSClamp got them to flash forward).


The insane shadows dropped 5 frames per second once I got downtown. But until that point there was only a 1 frame drop when the game was loading in more textures/poly. More importantly, the stutter & micro stutter were gone. The animations were smooth while using a non-vanilla framerate controller (which is excellent). It's just a theory, for PS4/XBO speed PC gaming rigs.

Link to comment
Share on other sites

  • Recently Browsing   0 members

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