Tweaking Skyrim Image Quality @ [H]

burningrave101 said:
I just ran across this review here which seems to show a clear advantage to the 3960x in Skyrim:

http://www.techspot.com/review/467-skyrim-performance/page7.html

20% faster than the 2600k

Would be nice to see Kyle and them do some of their own testing as well to see what advantages SB-E offers in CPU intensive titles like Skyrim
Click to expand...

Comparing the 3960 to the 2500 really points out the weirdness for this game.

No more usable threads (Skyrim uses 2 threads, and that's basically it), the two processors are operating at the same clocks and have the same underlying architecture, yet the 3960 blows past the 2500 by more than 25%. That means the game is either desperately starving for:
- on-chip cache
- memory bandwidth

...as those are the only two things relevant to Skyrim different between the parts! I really wonder which it is...
 
Well, if it is cache, then the 2600K ought to be better than the 2500K, right? I don't think it is, but I also find it hard to believe that Skyrim is the only game in the world that is starving the SB chips for memory bandwidth. Maybe they screwed up the testing somehow?
 
Setting Shadowmap resolution to 8192 gives me about 1-3 FPS indoors. Not sure why but I'm definitely going to have to use 4096 instead unless there's a fix.
 
Forceman said:
Well, if it is cache, then the 2600K ought to be better than the 2500K, right? I don't think it is, but I also find it hard to believe that Skyrim is the only game in the world that is starving the SB chips for memory bandwidth. Maybe they screwed up the testing somehow?
Click to expand...

I doubt its due to screwing up the testing, the results line up with Hardware Heavens Skyrim benches using a 2700k vs 3960x and a faster GTX 590.

http://www.hardwareheaven.com/revie...ch-core-i7-3960x-processor-review-skyrim.html
 
Forceman said:
Well, if it is cache, then the 2600K ought to be better than the 2500K, right? I don't think it is, but I also find it hard to believe that Skyrim is the only game in the world that is starving the SB chips for memory bandwidth. Maybe they screwed up the testing somehow?
Click to expand...

The 2600K is actually considerably better than the 2500K. Granted, there is a clock difference at "stock", but not enough to account for the difference in performance - based on what gains we see in overclocking 2600s.

I'd certainly like to see 'same clock' comparison between the two, with HT disabled on the 2600K. I'm 99% sure that the 2600K would come out convincingly ahead, which would point to cache badly...and, by extension, memory bandwidth may be a big deal, too - the game just has a lot of stuff to constantly load (memory overclocks, then, may be interesting to look into, as well).
 
Could someone do me a favour and post their complete, stable .ini file here with the tweaks mentioned in this article (without uGridsToLoad)? I've been trying to tweak some of these things but they don't seem to make a lot of difference...
 
Nice work on the article. It's good to see some actual performance numbers assigned to all the tweaks floating around. I'm sure we all do some minimal before/after testing when we put in tweaks, but nothing as thorough (and laborous!) as the [H].

With regards to the low quality textures. That's the main reason I run the FXAA injector mod. Not using default settings, but some modified/tweaked ones myself. Just a minimal amount of sharpening does a world of difference to make the low detail textures bearable. A little bit (nothing near what the defaults are) of colour correction to make the game a little less grey, and it's quite nice and pretty. With all the tweaks it takes the game from looking like a Minor Oblivion Upgrade, to a reasonable middle of the pack 2011 title.

Sample Shot:
oIf6Yh.jpg



Small Collection of Nature Shots

Fxaa Injector Settings (As Requested) (Using FXAA Injector 1.3, use this as a ref for values, I edited the code for brevity on the forum, so it won't copy/paste verbatim. Use the default file and post in the values below. Note this uses the pre-sharpen shader, and so for best effect should happen before the FXAA pass. Ie. turn off in-game FXAA and use the injector)

Code: 
#define USE_ANTI_ALIASING
#define USE_PRE_SHARPEN
#define USE_BLOOM
#define USE_TECHNICOLOR
#define USE_TONEMAP

FXAA Shader
fxaaQualitySubpix = 0.50;
fxaaQualityEdgeThreshold = 0.025;
float fxaaQualityEdgeThresholdMin = 0.012;

PRE_SHARPEN
bool highQualitySharpen = 1;
#define Average 1.0 
#define CoefBlur 3.2 
#define SharpenEdge  0.5 
#define Sharpen_val0  1.2 

BLOOM


#define NUM_SAMPLES2 4
BloomPreset = 0;
BloomThreshold = 4;
BloomWidth = 5;
BloomPower = 6;

TECHNICOLOR
#define TechniAmount 0.10
#define TechniPower 4.5
#define redNegativeAmount 1.0
#define greenNegativeAmount 1.0 
#define blueNegativeAmount 1.0 

TONEMAP
#define Gamma 1.05
#define Exposure -0.10
#define Saturation 0.30 
#define BlueShift 0.1 
#define Bleach 0.00 
#define Defog 0.000 
#define FogColor float4(0.0, 0.0, 0.0, 0.0)
 
Last edited:
aggies11 said:
Nice work on the article. It's good to see some actual performance numbers assigned to all the tweaks floating around. I'm sure we all do some minimal before/after testing when we put in tweaks, but nothing as thorough (and laborous!) as the [H].

With regards to the low quality textures. That's the main reason I run the FXAA injector mod. Not using default settings, but some modified/tweaked ones myself. Just a minimal amount of sharpening does a world of difference to make the low detail textures bearable. A little bit (nothing near what the defaults are) of colour correction to make the game a little less grey, and it's quite nice and pretty. With all the tweaks it takes the game from looking like a Minor Oblivion Upgrade, to a reasonable middle of the pack 2011 title.

Sample Shot:
oIf6Yh.jpg



Small Collection of Nature Shots
Click to expand...

That stone wall on the left looks really good. The trees in the distance look a little off though - but much better than the other injector shots I've seen. Nice settings.
 
Forceman said:
That stone wall on the left looks really good. The trees in the distance look a little off though - but much better than the other injector shots I've seen. Nice settings.
Click to expand...

The image resize from the hosting ups the contrast a bit much on the small details, at native res they are pretty nice and crisp. Obviously no substitution for a hi-res pack, but toggling it on and off makes the original look incredibly "blurred" by comparison. Default injector looked too much like an oil painting for my tastes. I'm trying my best to stop tweaking and actually play the game for once ;)
 
aggies11 said:
The image resize from the hosting ups the contrast a bit much on the small details, at native res they are pretty nice and crisp. Obviously no substitution for a hi-res pack, but toggling it on and off makes the original look incredibly "blurred" by comparison. Default injector looked too much like an oil painting for my tastes. I'm trying my best to stop tweaking and actually play the game for once ;)
Click to expand...

Can you post your settings somewhere?
 
Interesting...

On my 2600K/5850 rig, that shadowmapresolution tweak cuts my framerate in a third - goes from 45 fps to 14 @ Ultra 4x/16x/FXAA. None of the other tweaks had anywhere near that kind of impact and are reasonably playable (at least insomuch as Ultra preset is playable on this card to begin with... it's definitely a close call)
 
Forceman said:
Can you post your settings somewhere?
Click to expand...

Indeed - could you copy+paste your whole config file in? No need to trim for brevity, the forum 'code' tags will make a field that automatically scrolls, so we can copy it out without it taking up a lot of space on the screen!
 
dderidex said:
Indeed - could you copy+paste your whole config file in? No need to trim for brevity, the forum 'code' tags will make a field that automatically scrolls, so we can copy it out without it taking up a lot of space on the screen!
Click to expand...


Fair Enough: FXAA Injector version 1.3 found here. Using a modified "base" file from here, although no longer available. I found a preset that looked slightly "realistic" in terms of how the lighting looks in game. Then modified that to my own tastes. I found that the pre-sharpen shader made the most impact, while still remaining subtle, and the post-sharpen was unecessary as I was aiming for the minimal amount of sharpening necessary to improve clarity. Since the pre-sharpen was designed to happen BEFORE the FXAA pass, I turn off in-game FXAA and included the AA pass in the injector itself. Sharpen settings were tweaked to provide more of a lighter outline, vs the darker one seen in the injector default config. (I find that one looks too much like an oil painting). The lighter outline seems to allow for a nicer impact on the lighting. Indoors it can make the specular lighting a tad extreme, but I made some attempt to tone that down so there is a decent balance vs outdoor/indoors. Color palette / tone was kept as close to vanilla as possible, with just a hint of more vibrance (to reduce the grey/dreary almost "fog" appears of vanilla), but nothing like the prevalent injector settings. Outdoors should look pretty, but not Disney vibrant ;)

If you follow the previous posts link to the image gallery, and then click to "larger" the images, and then view at full size, you get a better idea of what the final impact would be.

As I said, I spent probably way too much time playing around with mods, and this one seemed "good enough" for my tastes, to allow me to finally start to play and enjoy the game, without having to wait for all the re-texture mods. Toggling the mod off after using it for a while ("pause" button on keyboard ingame), leads to quite the startling difference (especially in wood and rock textures, and distant landscapes).

Entire File "InjFX_Settings.h"
Code: 
/*======================================================================================
                             "USER" ADJUSTABLE SETTINGS
======================================================================================*/

// TODO: Normalize values to be on a human range scale, whole numbers prefered, decimals usable for micro adjustments
// These values should have min/max limit checks included in their functions, so that the end user doesn't get crazy results

/*------------------------------------------------------------------------------
						FILTER SELECTION
------------------------------------------------------------------------------*/
// Comment to deactivate an effect. 
// Example: To disable the tonemap effect, use // in front of #define USE_TONEMAP
#define USE_ANTI_ALIASING
#define USE_PRE_SHARPEN
#define USE_BLOOM
#define USE_TECHNICOLOR
#define USE_TONEMAP
//#define USE_SEPIA
//#define USE_VIGNETTE
//#define USE_POST_SHARPEN
//#define USE_FINAL_LIMITER



/*------------------------------------------------------------------------------
						FXAA SHADER
------------------------------------------------------------------------------*/
// Set values to calculate the amount of Anti Aliasing applied
float fxaaQualitySubpix = 0.50;  // Default: 0.75 Raise to increase amount of blur
float fxaaQualityEdgeThreshold = 0.025; // Lower the value for more smoothing
float fxaaQualityEdgeThresholdMin = 0.012; // Lower the value for more smoothing



/*------------------------------------------------------------------------------
						PRE_SHARPEN
------------------------------------------------------------------------------*/
//For higher precision in the calculation of contour, requires slightly more processing power
bool highQualitySharpen = 1; //0 = Disable  | 1 = Enable

// Set values to calculate the amount of  AA produced blur to consider for the sharpening pass
#define Average 1.0 //Def-Realistic 1.0
#define CoefBlur 3.2 //Def-Realistic 3.2

// Set values of the sharpening amount
#define SharpenEdge  0.5 //Def-Realistic 0.6, amount of sharpening, maxes pretty early. Not sure if edge detection?
#define Sharpen_val0  1.2 //Def-Realistic 1.4, higher = greasy (enhance dark lines), lower = specular/frosted (enhance white lines)


/*------------------------------------------------------------------------------
						BLOOM
------------------------------------------------------------------------------*/
// Number of samples per pixel taken for the Bloom effect. Don't set it to high!  4 = 25spp, 8 = 81spp, 16 = 289spp
#define NUM_SAMPLES2 4 // Must be set with a value dividable by 2
float BloomPreset = 0; // Disabled = 0 (Valid Preset Values = 1 to 9) Preset value 1 to 9 takes control over the next 3 settings.
float BloomThreshold = 4; // The min. level at which the effect starts (Valid Values = 1 to 9, use decimals for finetuning)
float BloomWidth = 5; // Sets the width of the effect (Valid Values = 1 to 9, use decimals for finetuning)
float BloomPower = 6; // The power of the effect (Valid Values = 1 to 9, use decimals for finetuning)


/*------------------------------------------------------------------------------
						TECHNICOLOR
------------------------------------------------------------------------------*/
#define TechniAmount 0.10 // 1.00 = Max
#define TechniPower 4.5 // lower values = whitening

// lower values = stronger channel
#define redNegativeAmount 1.0 // 1.00 = Max
#define greenNegativeAmount 1.0 // 1.00 = Max
#define blueNegativeAmount 1.0 // 1.00 = Max


/*------------------------------------------------------------------------------
						TONEMAP
------------------------------------------------------------------------------*/
#define Gamma 1.05
#define Exposure -0.10
#define Saturation 0.30 // use negative values for less saturation. 
#define BlueShift 0.1 // Higher = more blue in image.
#define Bleach 0.00 // Bleach bypass, higher = stronger effect
#define Defog 0.000 // Strength of Lens Colors.
#define FogColor float4(0.0, 0.0, 0.0, 0.0) //Lens-style color filters for Blue, Red, Yellow, White.


/*------------------------------------------------------------------------------
						SEPIA
------------------------------------------------------------------------------*/
#define Earthyellow // Color Tone, available tones can be seen in ColorTones.PNG (Do not use spaces in the name!)
#define SepiaPower 0.5 // 1.00 = Max


/*------------------------------------------------------------------------------
						VIGNETTE
------------------------------------------------------------------------------*/
// Vignette effect, process by which there is loss in clarity towards the corners and sides of the image, like a picture frame
#define VignetteCenter float2(0.500, 0.500) // Center of screen for effect.
#define VignetteRadius 0.80 // lower values = stronger radial effect from center
#define VignetteAmount -0.15 // Strength of black. -2.00 = Max Black, 1.00 = Max White.


/*------------------------------------------------------------------------------
						POST_SHARPEN
------------------------------------------------------------------------------*/
// Controls additional sharpening applied after previous processing. Strength should be max 0.25!
float Sharpen = 0.028;


/*------------------------------------------------------------------------------
						FINAL_LIMITER
------------------------------------------------------------------------------*/
// Controls the strenght of the limiter. 1.000 for default setting
int LimiterStrenght = 1.000;
 
Great stuff.

I'm wondering if you all ran afoul of the game's obvious memory leak which can really show up the more "aggressive" you get with tweaks that put more demand on the game.

It'll just show up out of nowhere in the form of performance taking a swan dive. Have you all seen that? It's the second biggest reason behind stability why I've been very conservative with any .ini tweaks.

http://forums.bethsoft.com/index.ph...-comparisons-and-explanation-default-57911/cv

First post is apparently the magic formula for doing Ugrids properly in this game. For the short version it's
The thing that made my game more stable was to get rid of interior cell buffer, and calculate exterior cell buffer like this:

(uGrids + 1)^2

So ugridstoload at 7 would have exterior cell buffer at 64.
Click to expand...

Kouresh's tweak guide for this game is supposed to show up some time this week so it'll be interesting to compare his findings to your all's here.


http://www.skyrimnexus.com/downloads/file.php?id=1013

Essential for anyone with more than 2 gigs of RAM.



Feeling frisky? Try this: http://donotargue.com/cfg-makers/skyrim/ ;)
 
Thanks for review.

I ve tested your tweaks and it is ok when i am outisde, framerate but in caves it is unplayable, 4,5,6 10 fps. On gtx 580 + 2600k..

Problem is in shadow resolutin. When i use 4096 it runs fine in caves. 8192 is too much for gtx 580s vram in that situatons..
 
Last edited:
I would have to say anyone not FXAA injector is missing out with regards to textures. I also run a custom config.

Only the sharpness is used on my machine though, as oblivion keeps its "dreamy" look.
 
jzodda said:
Nvidia has put out a small tweak guide.

http://www./Optimize/Guides/five-fa...guaranteed-to-make-your-game-look-even-better
Click to expand...

I applied some of those I hadn't yet and my frame rate tanked. I think all I added was forcing Anisotropic Filtering and changing the exterior cell buffer (I already had the shadows, ambient occlusion, and ugrids to 7), so it must have been the cell buffer one. I didn't notice any graphical or stability difference between the default exterior cell buffer and the higer number, so I may change it back.
 
I just went through a three hour heart attack. I used the ugridstoload=7 tweak and had some bad performance issues. I deleted the setting. I went back in and my saved games wouldn't load anymore and claimed to be corrupted, even though I never saved while using that setting. Ended up I had to put the setting back in, load the game, alt-tab out and change the setting to 5, then go in game and refini. I saved the game, exited, then deleted all ini mods. I reloaded the game and my saved games load again.
 
bigdogchris said:
I just went through a three hour heart attack. I used the ugridstoload=7 tweak and had some bad performance issues. I deleted the setting. I went back in and my saved games wouldn't load anymore and claimed to be corrupted, even though I never saved while using that setting. Ended up I had to put the setting back in, load the game, alt-tab out and change the setting to 5, then go in game and refini. I saved the game, exited, then deleted all ini mods. I reloaded the game and my saved games load again.
Click to expand...

Yeah, I made a backup of my saves folder before messing with that setting. Didn't have any issues, but still... most of the guides out there mention the possibility of save file issues.
 
Sabrewulf165 said:
Yeah, I made a backup of my saves folder before messing with that setting. Didn't have any issues, but still... most of the guides out there mention the possibility of save file issues.
Click to expand...
Like I said though, I never saved when that setting was active. It seems to put some type of data into the saved game even without saving.
 
bigdogchris said:
I just went through a three hour heart attack. I used the ugridstoload=7 tweak and had some bad performance issues. I deleted the setting. I went back in and my saved games wouldn't load anymore and claimed to be corrupted, even though I never saved while using that setting. Ended up I had to put the setting back in, load the game, alt-tab out and change the setting to 5, then go in game and refini. I saved the game, exited, then deleted all ini mods. I reloaded the game and my saved games load again.
Click to expand...

You can do that or do what I do: Load a current save, make a new save out of it, and that new save becomes your dummy/experimental fodder that you can go crazy with and if things don't work you you simply put your settings back to normal and load that previous save. Nice and easy. :)
 
So does not adjusting the uExterior Cell Buffer accordingly to your ugridstoload have any impact? Say for those of us who have performance issues indoors with ugridstoload=7 enabled?
 
bigdogchris said:
So does not adjusting the uExterior Cell Buffer accordingly to your ugridstoload have any impact? Say for those of us who have performance issues indoors with ugridstoload=7 enabled?
Click to expand...

This is the magic formula supposedly: (Ugrids value you want here + 1) * 8 = your exterior buffer value here.

So if you want Ugrids 7 it should be (7+1)*8 = 64 for your exterior cell buffer value.
 
bigdogchris said:
So does not adjusting the uExterior Cell Buffer accordingly to your ugridstoload have any impact? Say for those of us who have performance issues indoors with ugridstoload=7 enabled?
Click to expand...

Q-BZ said:
This is the magic formula supposedly: (Ugrids value you want here + 1) ^ 2 = your exterior buffer value here.

So if you want Ugrids 7 it should be (7+1)*8 = 64 for your exterior cell buffer value.
Click to expand...

I tried it both ways last night and didn't see any difference either way. I was using uGrids at 7 with cell buffer at 36 for a week or so with no problems. Changed it to 64 after reading that Nvidia article, also no problems, then tried playing with it last night. As far as I could tell, it didn't make any difference.

I kind of wonder why Nvidia didn't get Bethesda to give them some info on .ini settings - that article sounds like they just fooled around with it themselves. If Nvidia can't get Bethesda to cough up some info, what chance do the rest of us have.

Edit: Q-BZ, I fixed your formula - it is the number squared, not just *8.
 
Forceman said:
I tried it both ways last night and didn't see any difference either way. I was using uGrids at 7 with cell buffer at 36 for a week or so with no problems. Changed it to 64 after reading that Nvidia article, also no problems, then tried playing with it last night. As far as I could tell, it didn't make any difference.

I kind of wonder why Nvidia didn't get Bethesda to give them some info on .ini settings - that article sounds like they just fooled around with it themselves. If Nvidia can't get Bethesda to cough up some info, what chance do the rest of us have.

Edit: Q-BZ, I fixed your formula - it is the number squared, not just *8.
Click to expand...
Even indoors you didn't have issues? How much GPU memory do you have? I have 1.2GB and the game is very laggy indoors with ugridstoload=7 Any known fixes?
dderidex said:
No more usable threads (Skyrim uses 2 threads, and that's basically it)
Click to expand...
I don't think this is true. The Skyrim launcher uses 2 threads but the game uses many more. Bring up task manager and you can enable column 'threads' to see for yourself.
 
bigdogchris said:
Even indoors you didn't have issues? How much GPU memory do you have? I have 1.2GB and the game is very laggy indoors with ugridstoload=7 Any known fixes?
Click to expand...

I have a 1.5GB GTX 580, and I was getting 60 FPS pretty solid indoors (buildings and caves). Outdoors, depending on the setting, I get something between 45 and 60 - I was getting a pretty solid 60 before, but I added some shadow changes and increased the grass draw distance and that seems to have knocked it down some.

Did you change the shadow map resolution by chance? Setting that to 8192 killed my indoors frame rate, so I went back to 4096.
 
Forceman said:
I have a 1.5GB GTX 580, and I was getting 60 FPS pretty solid indoors (buildings and caves). Outdoors, depending on the setting, I get something between 45 and 60 - I was getting a pretty solid 60 before, but I added some shadow changes and increased the grass draw distance and that seems to have knocked it down some.

Did you change the shadow map resolution by chance? Setting that to 8192 killed my indoors frame rate, so I went back to 4096.
Click to expand...
Well, I changed the shadow map to 4096 and ugrids to 7 (along with calculated exterior grids) and the performance inside is just fine now. I actually set the ugrids to 6 and don't notice a difference than 7, so figure I will keep it on 6. I did not make any other of the hard ocp changes. Thing is, in [H] they say to change blurDeferredShadowMask=32, but the nvidia site says that 0-7 are the values and the 0 is the best.

*edit* OK I'm having horrible luck. I thought I would compare FXAA with AA+SuperSample. Well I didn't like the performance loss of AA+SuperSample, so I disabled it. Well the FPS is staying the same as it was with it on! I completely wiped my ini's and reset them and it's still the same bad performance as it was when the AA was on. The NV CP settings were 'restored' and are now at default.

*edit* ended up wiping out all ini's and the game data, then removing and reinstalling the GPU drivers, then the game itself. Fixed the 'stuck in aa even when it's off' issue. What bothers me is that even at 40-50 FPS this game feels stuttery. I can't enjoy ugridstoload=6/7 because of this. Left at the default of 5 the game runs at 60FPS and feels perfectly smooth. Even a drop to low 50's in FPS makes it feel stuttery/jerky. Is there anything to do about that? I tried disabling VSYNC. It just is horrible with that turned off, so I turned it back on. It didn't fix the stutterying anyways.
 
Last edited:
Doesn't uGrids have to be an odd number? I think I remember that from Oblivion.
 
Forceman said:
Doesn't uGrids have to be an odd number? I think I remember that from Oblivion.
Click to expand...
Oh, maybe it does. That would explain why I saw no difference between 6 and 7. Either way, even with 6/7 I still saw a lot of pop in, which is what I wanted to prevent. If I'm still going to see pop in, at least with 5, I do it at a rock solid 60 FPS.
 
Last edited:
bigdogchris said:
Even indoors you didn't have issues? How much GPU memory do you have? I have 1.2GB and the game is very laggy indoors with ugridstoload=7 Any known fixes?I don't think this is true. The Skyrim launcher uses 2 threads but the game uses many more. Bring up task manager and you can enable column 'threads' to see for yourself.
Click to expand...

While this is strictly true (I've seen upwards of 20threads i the process at times), generally the workload is spread (and not evenly mind you) between 3 maybe 4 threads tops. Cumulatively those add up to about 2 full parallel streams of execution. Which is around 100% on dual core, 50% on quad, etc... While the engine has been retrofit, it's foundations are still showing it's age. It's from a time before our multi-core world and I don't think the folks at Bethesda are particularly skilled enough in that area of expertise to do much to change that.
 
I noticed increasing the uExterior Cell Buffer to 128 is a nice way to use up more ram if you have it. I wonder if going to 256 would be even better ;) Hopefully it results in even less pop in with ugrids @7.
 
I can't imagine it would help pop-in, unless you were streaming stuff off a slow hard drive. Pop-in is just the engine deciding you are now close enough for it to draw the item, having more stuff loaded into memory isn't going to change that. Even increasing uGrids doesn't eliminate pop-in, it just makes it happen farther away.
 
For those who care, one of the things that really bothered me was tree LOD pop in and small trees/brush still drawing in, even with grass startfade set to 10000. So I made these changes and there is no more tree/shadow popin & bush popin
Skyrimprefs.ini

[Display]
fMeshLODLevel2FadeTreeDistance=4096.0000
fMeshLODLevel1FadeTreeDistance=5688.0000
fTreesMidLODSwitchDist=15000.0000

[Grass]
fGrassMaxStartFadeDistance=10000.0000

[Trees]
uiMaxSkinnedTreesToRender=40
 
Anyone having this issue? I've been gaming with most of the mods listed except the shadow map. Everything was going fine for the past few weeks. Now every so often during the day time the entire screen goes a bright pink psychedelic color pattern. The game is still playable and does not crash. It does not happen at night though.

No new mods or card patches or over-clocks. Was working great for the past few weeks. Not sure what the cause is. If I save it will save the pink. I have to go back to a previous save to get the correct color back. Thanks for any help.
 
aggies11 said:
While this is strictly true (I've seen upwards of 20threads i the process at times), generally the workload is spread (and not evenly mind you) between 3 maybe 4 threads tops. Cumulatively those add up to about 2 full parallel streams of execution. Which is around 100% on dual core, 50% on quad, etc... While the engine has been retrofit, it's foundations are still showing it's age. It's from a time before our multi-core world and I don't think the folks at Bethesda are particularly skilled enough in that area of expertise to do much to change that.
Click to expand...

I don't think the problem is skill so much as it costs $$$ and time

They have their foundation in place, it works "good enough" and they sell lots of copies. Maybe come next console generation...
 
For some reason in the area around Riften, I'm having problems with ugridstoload=7. It's a big FPS loss. When I disable it I gain 20 FPS in the Riften area. In the rest of the world it stays capped at 60. Anyone else notice this issue around Riften?
 
bigdogchris said:
For some reason in the area around Riften, I'm having problems with ugridstoload=7. It's a big FPS loss. When I disable it I gain 20 FPS in the Riften area. In the rest of the world it stays capped at 60. Anyone else notice this issue around Riften?
Click to expand...

Probably loading more geometry since it's near the city? Toggle the wireframe (twm console command) see the diff between views that are normal, and framerate drops. I personally find that it's mostly a case of excess geometry becoming a chokepoint.
 
You must log in or register to reply here.
Back
Top