GTA V on PC

[DavidGraham]

The game runs perfectly well with Max everything (Basic and Advanced settings) @720p on my 660Ti 3GB. it consumes 3.4GB of video memory though, without High resolution shadows it falls back to a more manageable 3.1GB! Can't think of the memory consumption @1080p!

 

[orangpelupa]

Yikes, those with 1GB VRAM are doomed

 

[psorcerer]

http://www.eurogamer.net/articles/digitalfoundry-2015-grand-theft-auto-5-pc-performance

Sorry to ruin it, but 59 avg and 42 min = 30Hz (with vsync)
And I would not believe Eurogamer even one bit, just out of curiosity tried that on a much more capable rig and got 50 ms frames (albeit rarely) in forest area immediately.
Not to mention that the lighting and overall IQ of this game is somewhere in the league of early PS3 games (although they did add some newer effects like better shadows).
But lighting, that lighting, my eyes bleed when I see any cutscene with closeups.

 

[Silent_Buddha]

Ummm, either you posted in the wrong forum about the wrong game. Or you're deliberately trolling.

There's plenty of people on this forum running the game at far higher IQ than the PS4 version at a locked 60 hz.

Regards,
SB

 

[psorcerer]

higher IQ than the PS4 version at a locked 60 hz

On GTX750 + i3?

 

[psorcerer]

And here I have an example of frame times in mountains region.
It's i7 + GTX970 1080p
And in the city it's indeed 60fps locked, no problems.

 

[tongue_of_colicab]

Yikes those with 1GB VRAM are doomed

What card with only 1gb memory is fast enough to run this game at playable framerates anyway?

 

[orangpelupa]

Radeon HD 7770 should be powerful enough and it's 1GB.

 

[pjbliverpool]

Sorry to ruin it, but 59 avg and 42 min = 30Hz (with vsync)

So you either lock to 30fps and ramp up the graphics settings beyond the console settings until you're running just fast enough to maintain that 30fps (perhaps very minor occasional dips) or you run with vsync off and enjoy the much higher frame rates. Yes there may be some tearing but I think people often make way too much of an issue out of that. Certainly different games at different frame rates suffer from it to different degrees. I often play without vsync and don't notice any tearing (or at least only occasional, minor tearing which is easily countered by the constant much smoother frame rate).

And I would not believe Eurogamer even one bit, just out of curiosity tried that on a much more capable rig and got 50 ms frames (albeit rarely) in forest area immediately.

Based on that frame time chart showing the average frame rate to be 46fps then clearly you're not running at the same settings that DF is using for the 750Ti.

 

[Deleted member 11852]

Grabbing a fast bike and hammering around in first person with without a doubt my favourite way to kill time in GTA V (albeit on PS4). The sense of speed is tremendous

 

[Malo]

I agree. I spent half of last night speeding around on various bikes. At 60fps it's fantastic weaving in and out of traffic.

 

[psorcerer]

So you either lock to 30fps and ramp up the graphics settings beyond the console settings

Won't help much, because even "all Very High" does very little difference to PS4 (and to framerate).

Yes there may be some tearing

There is a lot of tearing, it is indeed less noticeable in gameplay, but it's horrid in cutscenes (where camera rotates and changes planes frequently).
My frame times in the city are usually below 14 ms.

then clearly you're not running at the same settings that DF

Yeah? No, they clearly did not check it in mountain/forest regions. And them being biased fanboys I'm not even surprised.

 

[pjbliverpool]

Won't help much, because even "all Very High" does very little difference to PS4 (and to framerate).

In who's opinion? No offense but I'll make my own judgement on what I consider worthwhile graphical upgrades thanks. Good AA alone could make a huge difference..

There is a lot of tearing, it is indeed less noticeable in gameplay, but it's horrid in cutscenes (where camera rotates and changes planes frequently).
My frame times in the city are usually below 14 ms.

I can live a bit a bit of tearing in cut scenes if it means around double the frame rate in gameplay.

Yeah? No, they clearly did not check it in mountain/forest regions. And them being clearly biased fanboys I'm not even surprised.

I'm afraid I'm not buying that. You're suggesting that in those areas the 970 can't get higher than 46fps average at only High settings, 1080p + FXAA. If that's the case the game would be literally unplayable at those relatively low settings on anything slower than s GTX680. And every review I've seen so far completely contradicts that.

 

[homerdog]

I find that almost 100% of the time, the first one to use the word Fanboy in a thread is in fact just that.

 

[psorcerer]

Good AA alone could make a huge difference..

No, it cannot. AA* is obviously not the technology that moves realtime computer graphics forward. Otherwise we would still be using FFP instead of programmable hardware.

I can live a bit a bit of tearing in cut scenes if it means around double the frame rate in gameplay.

It doesn't matter too much, what matters is how Eurogamer guys bash tearing all the time. So, for them it should be a huge problem.

You're suggesting that in those areas the 970 can't get higher than 46fps average at only High settings, 1080p + FXAA.

MSAAx2

would be literally unplayable at those relatively low settings on anything slower than s GTX680

And why is that? Are you judging from other games performance? Because if we rely on other games typical performance then city-scapes must have lower framerate compared to mountain/vegetation types, not to mention that the city is quite dense in GTA5 and mountains are not.

* it obviously means the brute force, hardware-based multisample AA and not analytical or morphological ones.

 

[pjbliverpool]

No, it cannot. AA* is obviously not the technology that moves realtime computer graphics forward. Otherwise we would still be using FFP instead of programmable hardware.

Well in my opinion it can and does. Prettier pixels are great but a clean image makes a world of difference, especially when the image fills a large proportion of your FOV as it does on many PC setups.

It doesn't matter too much, what matters is how Eurogamer guys bash tearing all the time. So, for them it should be a huge problem.

I agree that they are overly harsh on tearing and I think are largely responsible for the current obsession with vsync'd 30 or 60 fps. Go back 5 years and no-one really cared that much whether the frame rate was locked or not. Higher was better. I don't think it's quite that cut and dry but neither is it absolutely necessary to vsync to either 30 or 60 fps - IMO.

MSAAx2

DF are testing with only FXAA engaged on the 750Ti and apparently MSAA is quite a performance hog in this game.

And why is that? Are you judging from other games performance? Because if we rely on other games typical performance then city-scapes must have lower framerate compared to mountain/vegetation types, not to mention that the city is quite dense in GTA5 and mountains are not.

Because the 680 is at best 2/3rds the speed of a 970 and thus anything slower than that (2/3rds of 46 fps) would be averaging below 30fps in those sections and thus unplayable.

 

[psorcerer]

I agree that they are overly harsh on tearing

Yet they say that 59 average is "like 60" with straight face.

MSAA is quite a performance hog in this game

I would not say that 14 ms is bad performance. And it's still with MSAA engaged.
I would say that the game just has some problems in vegetation areas, probably poorly implemented AOTC, but who knows.

Because the 680 is at best 2/3rds the speed of a 970

If the pure hardware speed was any reference to actual DX11 one, we would not need all the new APIs. Sady it's not even close. Most of the speed difference comes from the driver and API and not the hardware. Unless you're using some very hardware (or better yet memory bandwidth) dependent settings like resolution or MSAA. I would guess that GTA5 performance is also very CPU (mainly cache latency) dependent because it was always the case with Rockstar games.

 

[Arwin]

Wasn't MSAA disabled primarily because of memory cost?

 

[Deleted member 13524]

What card with only 1gb memory is fast enough to run this game at playable framerates anyway?

HD5970



Does anyone else notice a huge difference in framerate between the urban areas and the rural ones?
The vegetation in the rural areas make a huge impact on my 290X if shadow quality, shadow MSAA, etc.are maxed out. I get between half and 1/3rd of the performance I get in the city.

 

[Daozang]

HD5970



Does anyone else notice a huge difference in framerate between the urban areas and the rural ones?
The vegetation in the rural areas make a huge impact on my 290X if shadow quality, shadow MSAA, etc.are maxed out. I get between half and 1/3rd of the performance I get in the city.

I did notice that some times my frames dip bellow 60 outside the city for a second and back to normal again.
To be honest, I just cranked everything up and the game runs great with vsync on, so I can't say if I could avoid those small dips with some optimization.

Edit
It seems that I had GeForce experience recommended settings enabled (and I don't even remember doing it)...
So all though it is cranked pretty high, not everything is on "ultra".