GTA V PC Performance Thread

True. FXAA and DSR work great together.

I wonder why so many people are using costly MSAA. I hardly see anyone using DSR and FXAA.
Because I tried it and decided that 1080p w/MSAAx4 is better than 1440p w/FXAA. The game's aliasing issues are fairly severe, so I think a dedicated AA solution is just more effective unless you can downsample from much higher(which is obviously enormously costly).
 
This game takes forever to load into SP on my SSD.

It's weird because after I installed it yesterday it booted up quick, but now it's slower than my PS4 copy.
 
Can anyone help me out. I've got an i5-4670k @3.8ghz with a 780ti and 8gb ram I have reduced my settings as low as possible but i still cannot get a solid 60fps, its just constantly falling below. The only thing i can think of would be the hard drive possibly as it is about 3 years old, however i dont have any problems in other games and am unsure how to test this.
Any help would be much appreciated
 
Ok, yeah DSR is the way to go in this game. FXAA at 1.78x or 2x DSR gets rid of almost all the jaggies and creates a smooth detailed image. Hell, even with no FXAA or MSAA, just downsampling seems to get rid of 90%+ of the jaggies, far more than MSAA 4x does at 1080p.

Running at 2650 x 1600 with no AA, reflections down to high, reflection MSAA off, tessellation down to high, particles and post fx down to high, soft shadows on softest, but with long shadows and high res shadows set to on; I seem to get fairly constant 60fps even with police chases through the city and lots of destruction. Might mess with the settings more, but I think this might be my go to settings for now.
 
computerbase benchmarks:

http://www.computerbase.de/2015-04/...prozessoren-und-kantenglaettung-im-vergleich/

Apparently AMD's CHS is better than PCSS and everyone should be using it, even people with Nvidia GPUs. According to them PCSS is missing shadows, causes crashing (on AMD) and performs worse.

They also take a look at performance scaling with MSAA and it's not pretty.

I wasn't expecting this to be a good port, very tempted to buy it even though I already have it on PS4.
 
I have FXAA + TXAA + 4x MSAA enabled purely to try and get the most coverage. Temporal aliasing in games like this drives me insane, so TXAA is nice and the general blanket blurring with FXAA helps a tiny little bit with the overall image too.

Yeah, it won't produce the sharpest screenshots, but I've found myself caring a lot more about in-motion image quality these days and stuff like TXAA is nice for that when you have games like GTAV which have some rather stubborn aliasing. I don't always use it, but it's a nice option to have.
 
Can anyone help me out. I've got an i5-4670k @3.8ghz with a 780ti and 8gb ram I have reduced my settings as low as possible but i still cannot get a solid 60fps, its just constantly falling below. The only thing i can think of would be the hard drive possibly as it is about 3 years old, however i dont have any problems in other games and am unsure how to test this.
Any help would be much appreciated

Describe the probably more specifically. By falling below do you mean it's at 50? 30? Is it a steep drop and recovering or a steady low?
 
Okay, after setting vsync to on in the nvidia panel, the game no longer stutters and does sit at 60fps. Able to crank everything up to max other than MSAA, tesselation, grass, and smooth shadows, along with the advanced settings in distant shadows. Only drops frames due to loading cars and such, may turn down those values a bit just to keep it at 60. Satisfied! Looks amazing even with only FXAA
 
computerbase benchmarks:

http://www.computerbase.de/2015-04/...prozessoren-und-kantenglaettung-im-vergleich/

Apparently AMD's CHS is better than PCSS and everyone should be using it, even people with Nvidia GPUs. According to them PCSS is missing shadows, causes crashing (on AMD) and performs worse.

They also take a look at performance scaling with MSAA and it's not pretty.

I wasn't expecting this to be a good port, very tempted to buy it even though I already have it on PS4.
Why the fuck would they do their CPU comparisons at 4K? Fucking stupid as hell. All that shows is that there is an obvious GPU bottleneck. Shame, cuz I'd like to see how much clock speed affects performance at resolutions most of us will be playing at.

I'll check out AMD's soft shadows, though. Didn't realize they were Nvidia-compatible.

And the MSAA scaling is pretty reasonable, really.
 
This game takes forever to load into SP on my SSD.

It's weird because after I installed it yesterday it booted up quick, but now it's slower than my PS4 copy.

From my experience I had super long/endless loading times when having the game window out of focus, don't know if that is it.
 
Describe the probably more specifically. By falling below do you mean it's at 50? 30? Is it a steep drop and recovering or a steady low?

Its very constant stuttering. It's actually quite hard to explain, i guess you could say it is very interrupted. One thing that seems to help is rebooting the game constantly until u eventually have a better experience.

Going to try an older driver now just to see the results
 
From my experience I had super long/endless loading times when having the game window out of focus, don't know if that is it.

Just figured it out. Issue started after I added a shortcut to my somewhat massive music folder on my external drive. After I deleted the shortcut it loaded quick.
 
Hmm, I had to turn off hardware restriction to get the game how I wanted it to look (at least on part with PS4) but I got it there with about 2.15 ish GB of video memory. My graphics card only has 2GB, but my theory is I'll rarely (if ever) truly max that out, and Rockstar's estimate is simply that, an estimate.

Was running at 60FPS with no problems, so far so good.
 
Because I tried it and decided that 1080p w/MSAAx4 is better than 1440p w/FXAA. The game's aliasing issues are fairly severe, so I think a dedicated AA solution is just more effective unless you can downsample from much higher(which is obviously enormously costly).

Ok, yeah DSR is the way to go in this game. FXAA at 1.78x or 2x DSR gets rid of almost all the jaggies and creates a smooth detailed image. Hell, even with no FXAA or MSAA, just downsampling seems to get rid of 90%+ of the jaggies, far more than MSAA 4x does at 1080p.

Running at 2650 x 1600 with no AA, reflections down to high, reflection MSAA off, tessellation down to high, particles and post fx down to high, soft shadows on softest, but with long shadows and high res shadows set to on; I seem to get fairly constant 60fps even with police chases through the city and lots of destruction. Might mess with the settings more, but I think this might be my go to settings for now.

Now this is interesting :)
 
These are my settings

kJf2G6f.jpg

For now it is rock solid 60FPS on i7 2600K Stock, 8GB DDR3, GTA 760 2GB factory overclocked. No jaggies, DSR works like a charm (in Borderless mode because my stupid monitor doesn't support full screen for some reason) and all of that while i broke Memory limit for 500MB. I will see later if there will be any issues.
 
computerbase benchmarks:

http://www.computerbase.de/2015-04/...prozessoren-und-kantenglaettung-im-vergleich/

Apparently AMD's CHS is better than PCSS and everyone should be using it, even people with Nvidia GPUs. According to them PCSS is missing shadows, causes crashing (on AMD) and performs worse.
I read this yesterday and found their argument a bit silly. They say CHS is better because it's "more detailed" -- by which I can only assume they mean "sharper" -- when the whole point of technologies like this is to make shadows progressively further from the casting object less defined. And that (the variable degree of hardness, and in particular, and perhaps surprisingly, the maximum softness!) is what makes it somewhat expensive to render, so it's not surprising that the PCSS implementation is more expensive.

Now, if there are actually missing shadows than that is a separate issue, but calling a soft shadow technology better because it's sharper is just off.

Why the fuck would they do their CPU comparisons at 4K? Fucking stupid as hell. All that shows is that there is an obvious GPU bottleneck.
Yeah, the only site which always does GPU benchmarks right is gamegpu. Edit: or actually no, they did it at 1080p.
 
Apparently AMD's CHS is better than PCSS and everyone should be using it, even people with Nvidia GPUs. According to them PCSS is missing shadows, causes crashing (on AMD) and performs worse.

PCSS does more than AMD's version. You can clearly see in the screenshots that AMD shadows are much harder edged. The problem with PCSS is that it needs to be configured to work with most cases but there can be some where it looks worse. I think it does look better most of the time though.
 
So after messing around with the page file some more I'm now able to play the game for an hour or so before it reaches a critical limit. I've never monitored memory usage before, but does this always accumulate or would it just use a set amount constantly. The way it's going for me at the moment is that it starts at 3gb and slowly works its way up to the 8gb mark without ever decreasing.
Is this an issue with the game or something wrong with my setup?
 
True. FXAA and DSR work great together.

I wonder why so many people are using costly MSAA. I hardly see anyone using DSR and FXAA.
1080p + 2xMSAA + TXAA gives me better image quality and better framerates than 1440p (through DSR) + FXAA. There's this shimmering thing going on with distant edges that kinda looks like borked DOF that MSAA seems to take care of better.
 
Tried to play just now but the performance has mysteriously just went to complete and utter SHIT. Absolutely unplayable.

Have no idea what the fuck is going on. Just sitting still in a car, looking around and it's jumping wildly from 28fps to 84fps with vsync off and this wild fluctuation happens pretty much no matter what settings I use.

What the fuck.

1080p + 2xMSAA + TXAA gives me better image quality and better framerates than 1440p (through DSR) + FXAA. There's this shimmering thing going on with distant edges that kinda looks like borked DOF that MSAA seems to take care of better.
Yup. 1440p comes off better in screenshots, but in-game, it's *far* more shimmery than 1080p w/MSAA.
 
Why the fuck would they do their CPU comparisons at 4K? Fucking stupid as hell. All that shows is that there is an obvious GPU bottleneck. Shame, cuz I'd like to see how much clock speed affects performance at resolutions most of us will be playing at.
What are you talking about? They tested it at 1920x1080. In that graph @4K means with 4 Cores (Kernen).
 
I noticed that when my framerates go quite a bit over 60fps, navigating the menus turns into quite a struggle. Forcing 60hz seems to fix that for me, unfortunately

That's fine for now, I still have to finetune the game this afternoon. But let's hope for a fix.
 
Now, if there are actually missing shadows than that is a separate issue, but calling a soft shadow technology better because it's sharper is just off.
I wonder if it's case of truly missing shadows or shadows from very thin shadow casting objects being far from receiving object.
In such case the shadow will be almost invisible, as it should.
 
I read this yesterday and found their argument a bit silly. They say CHS is better because it's "more detailed" -- by which I can only assume they mean "sharper" -- when the whole point of technologies like this is to make shadows progressively further from the casting object less defined. And that (the variable degree of hardness, and in particular, and perhaps surprisingly, the maximum softness!) is what makes it somewhat expensive to render, so it's not surprising that the PCSS implementation is more expensive.

Now, if there are actually missing shadows than that is a separate issue, but calling a soft shadow technology better because it's sharper is just off.

That's true, CHS isn't really blurring at all it seems in those pictures.

One of you guys should take a few comparison shots in different areas. I would, but don't own the game.
 
I'm using mouse and keyboard controls and every so often, it's like the game uses the keyboard buffer instead of real-time key presses - so after accelerating the car for a while, I can't turn because it keeps accelerating for about three more seconds, or the character walks around a bend but keeps rotating for a while after.

My framerate seems to be fine, everything looks smooth but the occasional movement glitch is very frustrating mid-chase or combat.
 
That's true, CHS isn't really blurring at all it seems in those pictures.
I'm pretty sure CHS works by having the original/normal blur amount farther from object and when shadow receiver is close to caster it brings PCS samples closer to each other, making shadow even sharper.

Effect of CHS would most likely be more visible with low shadowmap resolution. (quite usual that shadowmap resolution is tied to blur amount, which really shouldn't be the case.)
 
I'm pretty sure CHS works by having the original/normal blur amount farther from object and when shadow receiver is close to caster it brings PCS samples closer to each other, making shadow even sharper.

CHS would most likely be more visible with low shadowmap resolution. (quite usual that shadowmap resolution is tied to blur amount, which really shouldn't be the case.)
That's a very good point, tons of games get this wrong. Perhaps that's also where the misconception that sharper (rather than more accurate) shadows are better comes from.
 
1080p + 2xMSAA + TXAA gives me better image quality and better framerates than 1440p (through DSR) + FXAA. There's this shimmering thing going on with distant edges that kinda looks like borked DOF that MSAA seems to take care of better.

Interesting. Seanspeed mentioned the same thing. Does it get better if you downsample higher?
 
I actually just tested out CHS and whilst the performance was marginally better, from what I can tell, and it is just a quick visual test (no actual data captured and compared), but it seems like within a certain (close-ish) distance CHS doesn't really soften the shadows.

As I said, there nothing scientific about my quick comparison, I may very well be talking out of my arse, but there were instances where two different shadows appeared to be just as sharp as one-another but the object they were cast had a reasonable difference in distance.

I'd spend some time actually taking some screenshots but I'm running out of time to rush through the story, and I want to get these last 15 missions or so done.
 
So after messing around with the page file some more I'm now able to play the game for an hour or so before it reaches a critical limit. I've never monitored memory usage before, but does this always accumulate or would it just use a set amount constantly. The way it's going for me at the moment is that it starts at 3gb and slowly works its way up to the 8gb mark without ever decreasing.
Is this an issue with the game or something wrong with my setup?

Seems normal to me. I have had page file sizes over 10 GB with other games occasionally. To me it doesn't make sense to do anything to the page file except put it on your fastest drive, let Windows manage its size.
 
Interesting. Seanspeed mentioned the same thing. Does it get better if you downsample higher?
It should, but TXAA temporal component most likely is reason what truly helps with shimmering. (it also includes wider resolve, so DSR smoothness should not be very low to be directly comparable.)

Most likely DSR, TXAA + FXAA ẃould be ideal, but do not know if one can use FXAA with TXAA. (will have to test when I get home.)
 
Running 4k smooth with 2x 7970s OC and a 4770k at 4.1ghz (shitty chip)
Wouldnt say 60fps but defs around 50 most the time only issue is minor screen tear sometimes.
 
Processor (CPU) Intel Core i7 4770K HD4600
16gb Kingston Hyper-X DDR3 1600mhz
2GB Nvidia GeForce GTX 770
120GB Tungston SSD
Windows 8.1

1920x1080p @ 59hz

Only got a quick 10 minutes in this morning but it seems I can either max out all sliders with textures on normal or have textures on high and pull a few sliders down.

Couldn't get 60hz to show up, but 59 should be fine.

Weirdly even though it was specified at 59 when I did the bench mark test it was showing 30 the whole time. Not sure what's going on there. Will investigate more later.
 
Seems normal to me. I have had page file sizes over 10 GB with other games occasionally. To me it doesn't make sense to do anything to the page file except put it on your fastest drive, let Windows manage its size.
How does this game behave if there is more than enough system RAM (16GB or more)? Is it using excessive page file sizes with that amount too?
 
Seems normal to me. I have had page file sizes over 10 GB with other games occasionally. To me it doesn't make sense to do anything to the page file except put it on your fastest drive, let Windows manage its size.

So does that mean that I shouldn't worry about afterburner showing 8gb or more ram usage? I've been quitting the game and restarting every time I hit 7gb in fear of the game crashing and losing progress.
 
So does that mean that I shouldn't worry about afterburner showing 8gb or more ram usage? I've been quitting the game and restarting every time I hit 7gb in fear of the game crashing and losing progress.

MSI isn't showing you the page file but regular RAM usage. If your ram usage there is hitting the cap, you may find your game crashing with an out of memory error.
 
Top Bottom