GTA V PC Performance Thread

I assume grass quality only affects your performance once you hit the countryside? I bumped it up to ultra earlier and didn't see a difference in fps (I'm only at the very beginning)

Yes, kiss goodbye to your perf one you go in the forests.

Is there any difference in high vs very high grass quality?
Cause i can't see it and i'd rather save those frames if there really isn't any noticeable difference.

Draw distance of grass and shadows on all grass on Ultra.

Any idea if Geforce Experience settings will change? The settings it recommends are pretty intense for my i5 3570k and 760.

We're working on them, but the VRAM, CPU and GPU requirements of various settings are making things difficult. Not to mention the number of settings, the number of possible setting configurations, and the amount of hardware configs we support... Recommend using the guide and tweaking GeForce Experience suggestions.
 
Is there any difference in high vs very high grass quality?
Cause i can't see it and i'd rather save those frames if there really isn't any noticeable difference.

The quality of the grass seems the same, but I'm guessing the distance at which that grass is drawn is farther. It's likely more noticeable on a hillside where grass is consistently visible and thick out to a far distance. The example in the above guide has a big highway in the middle where I think the difference might otherwise be more noticeable.
 
I've cleared the final hurdle, it's here!

Well done!

Qytb.gif
 
The open world actually allowed me to avoid all the story grind and repetitive replaying that I have in a linear game, and I've waited for this game for a damn long time so you can bet your ass I'm gonna play it properly soon :)
Play online man, it's a blast. Thanks for the guide it's a long read and crazy informative. How many hours you think it took to make if you factor everything in?
 
Shouldn't need it enabled in-game in 99.9% of games as NVCPL's function is an override.

Ah, had it enabled as a global setting. Setting it to on in the program settings section fixed it. Is that normal? Also for some reason the performance setting in the CP looks better than the high quality setting, which is weird (first screen is performance anisotropic filtering w/ sample optimization, second is high quality).

Rytb.jpg

Sytb.jpg
 
Play online man, it's a blast. Thanks for the guide it's a long read and crazy informative. How many hours you think it took to make if you factor everything in?

I don't want to know. Started early Tuesday, worked all week, did some more on the weekend (to the annoyance of my girlfriend), finished it yesterday, and used time while in review to tweak and copy edit.

Putting it another way: I haven't done any other work since Tuesday.
 
Draw distance of grass and shadows on all grass on Ultra..

Andy, when you tested the PostFX settings, did you also test it with grass set to both very high and ultra? I notice when I had both set to ultra it absolutely murdered my framerate at night, since grass (and everything else) was casting shadows from my headlights. Could that account for the bigger PostFX gap between ultra and very high some people have reported?
 
I don't want to know. Started early Tuesday, worked all week, did some more on the weekend (to the annoyance of my girlfriend), finished it yesterday, and used time while in review to tweak and copy edit.

Putting it another way: I haven't done any other work since Tuesday.

Even though I essentially scanned over your article to check and confirm what I thought some settings meant, I noticed how insanely detailed it is and how much work must have went into it.

Might I suggest a new comparison picture for either the long shadows or extended shadow distance options? If you set the game to around 12-2pm on a clear day, Mt. Chiliad actually casts a massive shadow over Paleto Bay which looks incredible.
 
Andy, when you tested the PostFX settings, did you also test it with grass set to both very high and ultra? I notice when I had both set to ultra it absolutely murdered my framerate at night, since grass (and everything else) was casting shadows from my headlights. Could that account for the bigger PostFX gap between ultra and high some people have reported?

Everything was at max and I tested foggy forests. No one can explain it.

Even though I essentially scanned over your article to check and confirm what I thought some settings meant, I noticed how insanely detailed it is and how much work must have went into it.

Might I suggest a new comparison picture for either the long shadows or extended shadow distance options? If you set the game to around 12-2pm on a clear day, Mt. Chiliad actually casts a massive shadow over Paleto Bay which looks incredible.

I'll check it out tomorrow, thanks for the suggestion.
 
Guess I'll change it from ultra to very high grass at least and see what the benchmark says. I mean I could try with ultra but you all are much more educated in these things than I am so I'm sure you are right
 
I've cleared the final hurdle, it's here!



Now, what's next? Project Cars? Witcher 3?! :)

P.S. If anyone's awesome at Photoshop and has the time to make a better feature graphic I'll happily use it.

Thanks Andy!!

A quick question....you mention MFAA in the article, but don't mention that it only works with MSAA. Am I correct in thinking it only works with MSAA?
 
Thanks Andy!!

A quick question....you mention MFAA in the article, but don't mention that it only works with MSAA. Am I correct in thinking it only works with MSAA?

Yea, only MSAA. Were you reading the bit at the end about MFAA? I'm copy-editing myself and that's never great, so if you think it can be clarified/better explained I'll look into.

Guess I'll change it from ultra to very high grass at least and see what the benchmark says. I mean I could try with ultra but you all are much more educated in these things than I am so I'm sure you are right

If you don't have a TITAN X or fast SLI system you probably want to use Very High.
 
[Asmodean];161042146 said:
Man, anyone who hasn't tried the Nvidia PCSS shadows yet, do yourself a favour and give it a try. PCSS + High Resolution Shadows + Shadow Distance maxed out, looks amazing. Really adds to the overall scene lighting.

I actually think they might be the best looking shadows/shading that I've seen in a game, thus far.

If anyone's looking for a HW to settings reference, I'm running everything maxed - except for Grass is on Very High(ultra's not worth the fps hit in scenic areas for the minor visual difference imo), and HW AA is disabled, at 1440p. Average frame rate is 60fps, it can dip to the 40/50's in nature-heavy areas, though.

You should try turning grass down from Very High to High. There is negligible visual difference (save for the loss of some shadows) but I guarantee you'll then be able to achieve a pretty solid 60fps in the same nature heavy areas you've been to before
 
So what is the benchmark actually testing? I'm running it right now but I'm getting like 5 fps while in game I get 40 or so. Is it testing max settings instead of the settings I've picked?
 
I've cleared the final hurdle, it's here!



Now, what's next? Project Cars? Witcher 3?! :)

P.S. If anyone's awesome at Photoshop and has the time to make a better feature graphic I'll happily use it.

Man this must felt like The twelve labours of Hercules.

Great job Andy, was really interested in both what the "Advanced" options did and their performance cost, was also interested to see the AO comparison, especially fps wise, shame its bugged right now.
 
Anyone having any issues where the game will just drop to 30 FPS and be locked there for a while? I can usually get a locked 59 (which is really annoying, it causes a micro stutter and I can't get to 60 at all for some reason) but sometimes the game just drops all the way to 30 and becomes near unplayable with all the mouse lag i get.

It's an issue with the game's vsync. You could turn it off, and try to force vsync within your GPU settings. If that doesn't work, I found out that you can temporary fix it by turning vsync off in the game's settings, then turning vsync back on. Now you should not have to worry about the frame rate cutting in half. That is, until you quit the game & start it back later. Then you would have to turn vsync off and on again.
 
Thanks Andy, will check out the guide asap.

Considering how long it took Rockstar to fix all the graphical and performance issues in the PS4 version, I am not expecting a perfect build anytime soon. :(
 
2 x 970s SLI, dunno if that is a "fast SLI system"

It is pretty fast, but depending on the resolution you are running at and all your other settings, you may or may not be able to pull it off without noticeable drops. Ultra grass is really the biggest performance hit you will see from a single setting change. It's so unoptimized that the game forces you to restart after every change.
 
Thanks for the guide Andy. Started tweaking it even more and seems like it has a rockhard 60 fps in Los Santos. Still need to check Paleto Bay but seems near perfect.
 
So what is the benchmark actually testing? I'm running it right now but I'm getting like 5 fps while in game I get 40 or so. Is it testing max settings instead of the settings I've picked?

No. It tests the worst case scenario for the settings you've picked. If that's what you get in the benchmarks, then you really need to scale it back if you want the rest of the game playable.

This is assuming your benchmarks are working as intended (haven't heard anyone complain thus far)

Left FXAA - Right 4x TXAA
Yytb.jpg

Ha. Nice catch. Think Andy got them swapped.
 
I've cleared the final hurdle, it's here!



Now, what's next? Project Cars? Witcher 3?! :)

P.S. If anyone's awesome at Photoshop and has the time to make a better feature graphic I'll happily use it.

Another question Andy.

In the guide, you give comparisons between 2xTXAA, 4xTXAA. However in game, TXAA only can be turn on or off with no multipliers when enabled with MSAA. How did you get TXAA to x2 and x4?

Or are you saying that comparing 4xTXAA to 4xMSAA, is really 4xMSAA with TXAA vs 4xMSAA?
 
Thanks for the guide, AndyBNV. That, along with the new patch and the latest nvidia driver update for GTA5 I forgot to download, and wow, getting some seriously nice performance now. I mean it was good before, but this is even better.
 
Just posting these for those curious, and for AndyBNV.

This screenshot has a vague outline of the mountain's shadow in it. It's harder to see from this angle but I wanted to show off the beautiful light shafts that come with it.

I was wrong about the time the shadow is at its peak, it's around 8am and not 2pm. Interestingly enough no setting seems to disable it entirely, but Extended Shadow Distance seems to make it more accurate and less faded.
100% Extended Shadow Distance:
0% Extended Shadow Distance:
 
With the latest patch, I noticed a performance decrease in grassed areas (grass set on 'High') when MSAA is enabled, I've now dropped down to FXAA (enabled in the NCP, it gives a MUCH better result than the in game option) and after reading Any's guide, I might experiment with DSR and see how I go.

Yesterday I upgraded my RAM to 16GB from 8GB, and noticed a definite performance increase on my i5 4590 with MSI GTX 970 4G. My settings, for a solid 60fps EXCEPT in the vaunted Vinewood Blvd stretch and some isolated pockets map wide (it drops to the mid 50's) are below:

--Graphics-

Fullscreen

1920x1080 60Hz

FXAA – Off (enabled in the Nvidia CP, MUCH better looking)

MSAA – Off

TXAA – Off

VSYNC – On

Population Density – Max

Population Variety – Max

Distance Scaling – Max

Texture/Shader/Shadow/Reflection Quality – all Very High, Ultra Reflections

Reflection MSAA – Off

Water/Particles/Grass Quality – Very High, however Grass on High

Soft Shadows – Nvidia PCSS

Post FX – Ultra

Motion Blur Strength – Minimum

Depth of Field – Off

Anisotropic Filtering – Off, (Forced on in the Nvidia Control Panel - seems to give a better coverage)

Ambient Occlusion – High

Tessellation – Very High



-Advanced Settings -

Long Shadows – On

High Resolution Shadows – Off

High Detail Streaming when Flying – On

Extended Distance Scaling – Minimum (I can put it at 50% with a few frames lost)

Extended Shadow Distance – Maximum
 
Hmm...driver level FXAA is serviceable. way better than the ingame implementation. seriously considering turning off 2x MFAA so i could use other configs.
GTX 970 here.
Anyone else had/is in a similar situation?
 
No. It tests the worst case scenario for the settings you've picked. If that's what you get in the benchmarks, then you really need to scale it back if you want the rest of the game playable.

This is assuming your benchmarks are working as intended (haven't heard anyone complain thus far)



Ha. Nice catch. Think Andy got them swapped.

no, they are correct.
 
Hmm...driver level FXAA is serviceable. way better than the ingame implementation. seriously considering turning off 2x MFAA so i could use other configs.
GTA 970 here.
Anyone else had/is in a similar situation?
I find downscaling from 1440p and using fxaa works really well, mostly locked 60fps depending on settings
 
I've cleared the final hurdle, it's here!

Now, what's next? Project Cars? Witcher 3?! :)

P.S. If anyone's awesome at Photoshop and has the time to make a better feature graphic I'll happily use it.

Hey Andy, for reflection quality, you might want to post an example of an actual mirror, your character looking into it. That's what made me decide to increase the setting to Ultra.

In your examples, the difference between Ultra and Very High seem nearly non-existant - but when looking in a mirror you can see the entire world is rendered at what looks like 1/2 resolution and it's really blatant and apparent.
 
extended distance scaling was the cause of my frame drops while driving. theres serious issues with it. lots of drops with gpu and cpu utilizationg far from max.
 
Whaa…? TXAA looks far worse than FXAA on that shot.

that shot just happens to expose some of the areas that miss msaa coverage. in motion msaa looks far better than fxaa. txaa doesnt really help much over msaa. you can tell by the classic thickening effects on the foliage fxaa always has.
 
Game runs awesome on my GTX 980 with an i7-4770k CPU at 1440p. I just wish I could drive a car and chase down and shoot a guy on a motorcycle without giving it 20 tries and thinking of abandoning the game because it's not fun anymore.
 
Top Bottom