• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Grand Theft Auto V Performance Analysis

This game is worth upgrading your rig for. The eye candy is overwhelming. Best looking sandbox game out there.
 
I think the AMD R9-295x2 tests are wrong. The R9-295x is two cards and crossfire doesn't work in windowed mode. All of the settings show that resolutions less than 4k were windowed so you only used one of the GPU's for those tests.

--Rick
 
I think the AMD R9-295x2 tests are wrong. The R9-295x is two cards and crossfire doesn't work in windowed mode. All of the settings show that resolutions less than 4k were windowed so you only used one of the GPU's for those tests.

--Rick
we used fullscreen of course. the settings screenshots are just to illustrate what settings are available (and i used windowed to make it easier to screenshot)
 
I figured as much but the numbers look really weird in that the 290 and 295 aren't as far apart as I would have expected unless your doing 1440p or 4k so I had to ask. Thanks for the quick response.

I think AMD has some work todo on their drivers.

--Rick
 
Last edited:
I figured as much but the numbers look really weird in that the 290 and 295 aren't as far apart as I would have expected unless your doing 1440p or 4k so I had to ask. Thanks for the quick response.

I think AMD has some work todo on their drivers.

--Rick
yeah at lower res CF doesn't seem to scale a lot
 
W1zzard did you set the texture on very high on GTX960 with 2GB yet you still get that 75fps@1920x1080 ? didn't that 2GB run out , I have GTX680 which suppose to performe better than GTX960 yet I'm unable to get 75 as an avarge .

EDIT : Ok I see you used in game benchmark for that FPS.
 
Last edited:
Oh yeah, I forgot to post the settings indeed. It's very high on everything, no MSAA, just FXAA. MSAA really doesn't make sense with its huge performance hit.
I'm guessing all advanced settings were off?
 
If we were living in the GTX280 days, and GTAV was released, I am pretty sure people would say "Performance sucks, what a sloppy port.". GTA4 in max settings imo, was ahead of its time.

GTA3, Vice City and SA were decent ports as well imo.


Quite the contrary, my 1100T and 7970 overclocked are running it on Very High settings with 8X AF and maintaining 40-60FPS with the average closer to 60. My CPU is probably the biggest hurdle currently.

Fraps Benchmark, being as hard on it as I can, driving really fast with a fresh game run, around downtown. Vsync on.

2015-04-17 00:09:47 - GTA5
Frames: 11724 - Time: 240000ms - Avg: 48.850 - Min: 30 - Max: 61

I can provide the csv files too.
 
Last edited:
my mate said his 970 is hitting the 500mbgate vram all the time and crippling his fps.

did you notice much stuttering with that too or did you limit the settings so prevent it happening?

What settings is he using, and what are the rest of his specs? Because I've played a few hours with all the settings maxed @1440p and basically been pegged at 100FPS(vsync on, 100Hz monitor) and no stutter.

well some people expect £600s worth brand new gpu hardware to be able to max out new games so they crank it all to 11.

From his post it sounds like he is using a single 970, which $600 and definitely isn't going to max this game out(well maybe close to max @1080p).
 
2 thread game? I won't be buying this crap.
 
2 thread game? I won't be buying this crap.

Umm...look at the results, it is playable on 2-core CPUs, but the game definitely benefits from 4-Cores. This is probably the first game to actually take advantage of more than 2 cores.

That is why the i3(2c/4t) 3.8GHz is outperformed by the i5(4c/4t) 3.0GHz by almost 10FPS.
 
Wow, you praise GTAIV's port yet everyone at the time it came out said it was one of the worst Rockstar did. Heck the port was so bad you couldn't really play it on anything less than quad cores. I remember playing it on a X2 5000+ and it was utter crap. Then I popped in a Phenom X4 and it went from a sloppy mess to actually playable. Sure it looked better either way but the game was not playable on dualies.

You look at it now and would say it was awesome that it used all the cores. Trouble was, that wasn't how it was back then. It wasn't because Rockstar designed it from the start to use quads on PC. It was because the consoles were at least triple cores. They didn't want to take time to translate it so they did the bare minimum. If it didn't work right, they didn't care. People would be forced to upgrade to quads to get it to be somewhat playable. Not to mention it was just plain buggy anyways.

LA Noire was another horrific port but it wasn't as messy as GTAIV. Its just more in dire need of a HD texture pack.

I suspect GTAV got a better port due to the fact the consoles are all x86 right now. Unlike GTAIV which was on RISC hardware, they wouldn't have to invest in translating it to x86 because PS4 and Xbone both already are on x86. Probably why it paid off to do a PC version first internally then scale it down for PS4/Xbone.
 
What settings is he using, and what are the rest of his specs? Because I've played a few hours with all the settings maxed @1440p and basically been pegged at 100FPS(vsync on, 100Hz monitor) and no stutter.

From his post it sounds like he is using a single 970, which $600 and definitely isn't going to max this game out(well maybe close to max @1080p).

when did £600 mean $600?

£300 each 970s dude, and 2 fail to max gta 5 at 1080....

all max with no aa and it is a "stuttery mess" he said. but he is also thankful for the ram bar so he can drop settings accordingly. other than that i know not what settings he used as that is all he has said about it.

shame really as having seen it run on a rog swift with 2 980s it does look sweet at 144hz.
 
when did £600 mean $600?

£300 each 970s dude, and 2 fail to max gta 5 at 1080....

all max with no aa and it is a "stuttery mess" he said. but he is also thankful for the ram bar so he can drop settings accordingly. other than that i know not what settings he used as that is all he has said about it.

shame really as having seen it run on a rog swift with 2 980s it does look sweet at 144hz.


Based on the overwhelming reviews and testing from users and some sites it seems your friend is having other issues. Anyone thankful for a RAM bar instead of running FRAPS, Afterburner, or some other tool is plain silly.

60Hz and up will make little difference in GTA, you are better off running lower FPS with higher draw distance.
 
which reviews maxxed all the settings out and then ran their tests?

from what i have seen all reviews tailored the settings to show most cards in their best light and none tried to use the game as the new crysis.
 
which reviews maxxed all the settings out and then ran their tests?

from what i have seen all reviews tailored the settings to show most cards in their best light and none tried to use the game as the new crysis.

Well first you have to understand that at 4K AA isn't really needed, so your premise is flawed to begin with, at 1080 my single 7970 is pushing it nicely with advanced settings turned on, and considering its about 2/3 of what a 970 can handle there is no reason a 970 shouldn't be able to run almost maxxed settings at 1080 with 2X AA or something.

But back to the idea that we need a "new crysis" thats a load of crap for the time being, once the modding community gets their hands into GTA5 I fully expect it will be able to put a hurting on QuadSLI Titans, or Xfire 295X2's.

http://www.extremetech.com/wp-content/uploads/2014/04/4K-NoMSAA.jpg

http://www.extremetech.com/wp-content/uploads/2014/04/1080p-NoMSAA.jpg

http://www.extremetech.com/wp-content/uploads/2014/04/1080p-8xMSAA.jpg

http://www.extremetech.com/wp-content/uploads/2014/04/4K-2xMSAA.jpg


A run with all settings maxxed, it kept giving me warnings so I had to use CCC to enable AA and AF, and pushed Tessellation to use application settings.


2015-04-17 19:42:23 - GTA5
Frames: 10260 - Time: 240000ms - Avg: 42.750 - Min: 27 - Max: 61
 

Attachments

Last edited:
thanks for the effort steevo :)

but i think your assumption has somewhat missed my point. what is needed or not is a person opinion to each user, personally for me i hate aa as i find it just muddies the crispness of the images. to use the benchmark as a tool for comparison, one which will have legs going forward you want to use all the eye candy it has to offer so as to push the performance of the hardware to its limits. like people did with crysis and why it was a title used in reviews for a few years. now the flip side to that is no game wants to be the next crysis as it was a sales failure because of the fact it was so demanding.

much like how all the reviews which i have read have run card v card tests at reduced settings so as to show that hardware in a better light. i mean joe bloggs looks at those charts and he will think take those numbers as fact. were the settings all to 11 he would see lower numbers and think either the gpu are not so great or the game is a pig (like they did crysis).

imo to think the pc gaming world does not need a new crysis is silly. but i also do think this has kinda fell into that role atm. it is the new poster child for the "pc master race" banner carriers as it rightly should be i guess. its huge sandbox world, breath taking 4k eye candy and endless fun in mp are everything pc gaming is about. throw mod support on top and i can see many people losing weeks if not months to it. as we have already seen with many videos it puts the consoles to shame on high end pc hardware and really shows the future of what, i hope anyway, video games can do.

tahiti cards are still great cards for 1080 imo, was i still running 1200p i think i might still have mine tbh. after watching that tb vid i think i will try to talk my mate into moving his install to his revo drive as it is faster than his ssd. tb had what looked like massive drops in load and fps on his sli run which i would have to guess is a data bottleneck from his ssd too. he was getting lower lows in his vid with sli than the single 980 so it really does show that the game still has some issues.

as for skyrim and mods...that game was pushing dx9 near its limits out the box. the mods just show why dx9 needed to be replaced by a more efficient api imo.
 
Wow, you praise GTAIV's port yet everyone at the time it came out said it was one of the worst Rockstar did. Heck the port was so bad you couldn't really play it on anything less than quad cores. I remember playing it on a X2 5000+ and it was utter crap. Then I popped in a Phenom X4 and it went from a sloppy mess to actually playable. Sure it looked better either way but the game was not playable on dualies.

You look at it now and would say it was awesome that it used all the cores. Trouble was, that wasn't how it was back then. It wasn't because Rockstar designed it from the start to use quads on PC. It was because the consoles were at least triple cores. They didn't want to take time to translate it so they did the bare minimum. If it didn't work right, they didn't care. People would be forced to upgrade to quads to get it to be somewhat playable. Not to mention it was just plain buggy anyways.

LA Noire was another horrific port but it wasn't as messy as GTAIV. Its just more in dire need of a HD texture pack.

I suspect GTAV got a better port due to the fact the consoles are all x86 right now. Unlike GTAIV which was on RISC hardware, they wouldn't have to invest in translating it to x86 because PS4 and Xbone both already are on x86. Probably why it paid off to do a PC version first internally then scale it down for PS4/Xbone.

And I praised it back when it was released too. Why? Because, just like GTA V, it played on a wide range of hardware. Like I said, my first play through was on a X2 4400+ and a HD3850. The CPU was over 3 years old at the time and the GPU was 2 years old. Not only was it playable, but it looked better than the console versions. To say it was unplayable on dual-cores is total BS. The only thing that made it unplayable on dual-cores was cranking the settings up. If you left them at reasonable levels(levels still higher than the levels used on the consoles) the game played just fine on dual-cores.


Rockstar said from the beginning that maxing the game wasn't something the current hardware at the time could do. They did this on purpose. To allow the game to grow with hardware. This is something we should have been applauding, it is something I wish more game developers would do, but instead everyone bitched because the game wouldn't run maxed out on their high end rig. But it would run, and look damn good, if everything was at about 75% or high settings(not very high, just high). The ironic thing is, GTA IV would have likely not receive any flak about being a bad port or "optimized" if they had just limited the settings available. The game would have still looked fantastic in comparison to the consoles and yet it would have run on high end rigs.

when did £600 mean $600?

£300 each 970s dude, and 2 fail to max gta 5 at 1080....

all max with no aa and it is a "stuttery mess" he said. but he is also thankful for the ram bar so he can drop settings accordingly. other than that i know not what settings he used as that is all he has said about it.

shame really as having seen it run on a rog swift with 2 980s it does look sweet at 144hz.

Well $600 means £600 when two GTX970s cost $600.

Oh, and 2 don't fail to max GTA 5 at 1080, in fact they max it at 1440p. I've been playing it on 1440p, max settings(FXAA, no MSAA), at 100Hz and it is smooth as can be.

I don't care what "your friend" says, it isn't a stuttery mess. I don't think W1z would recommend SLI 970s for 4k if it was a stuttery mess at 1080p...
 
+1
But i wish they added much better car damage models along with much better physics ( in the options ) more to what GTA 4 has. As to me it just looks pretty and with some laughs in SP and that's about it in fact i believe they have taken a step backwards to stop people bitching when they cannot max out their game lol.

As it is it's a $20 with addons aand it's the 1st GTA game i have done this with
 
on a X2 4400+ and a HD3850.

Well $600 means £600 when two GTX970s cost $600.

I don't care what "your friend" says, it isn't a stuttery mess. I don't think W1z would recommend SLI 970s for 4k if it was a stuttery mess at 1080p...

so you was one of the people who found gta 4 fine on a dual core too?

interesting as i found it to be a laggy mess on faster c2d than any amd x2 and it was only really playable for me with an x3 or quad core.

oh so you can read and not just look at the posts and make your own version of what it says, i am glad to see that at least...

because you seem to be one of the lucky ones who has not seen the last 500mb of ram issue it does not exist??

does that mean it aint really raining here if the sun is shining on your door too? lol

seeing as how w1z has already cleared up what settings the tests were ran at, and they were not all the way up it does not shock me to hear it can play the game at high res. but the kicker is this, aint nobody in their right mind going to go out today and pay £600 (less now as they have cut prices since the 500mb-gate) for 2 of these to not be able to run max settings. makes more sense to save £150 and take the 295 as that way you get about the same frames (more at higher res like 4k) and you can turn the setting up to fill all the vram without worry.

all that being said i care not what you say either, i will see what fell is on about this weekend and we can laugh at his decision to buy gimped hardware.
 
I built a quad core machine and a 1GB 4850 just to run GTA4, most of my current upgrades from my 5870 were to run GTA5 as well as a few other games.

Also don't do jumps on sport bikes, or you die.
 
so you was one of the people who found gta 4 fine on a dual core too?

interesting as i found it to be a laggy mess on faster c2d than any amd x2 and it was only really playable for me with an x3 or quad core.

Once you turned off that stupid game recording feature, and adjusted the settings to medium-low, yeah it played just fine on a dual core and looked awesome compared to the console versions. Heck, they even put out a nice little guide on what the console equivalent settings where. First off the consoles were only rendering at 720p, so since I was able to run it at 1366x768, I was already ahead of the game. Then the consoles had the view distance set to 21 out of a possible 100! I was able to run it at 25. Detail distance on the consoles was only 10! My rig let me run it at 20. Render quality on the consoles was low, I could do medium. It goes on like this for all the settings. Basically my 3 year old dual-core and 2 year old GPU could run GTA IV at higher settings than the consoles. That is pretty impressive. Oh, and the consoles only got 20-30FPS, I was steady over 30.

If you didn't go crazy with the settings, and actually used reasonable settings, the game was playable on some rather low end hardware. But some people seem to think that if you can't run max settings on shit hardware the game must be an unoptimized crappy port...

oh so you can read and not just look at the posts and make your own version of what it says, i am glad to see that at least...

because you seem to be one of the lucky ones who has not seen the last 500mb of ram issue it does not exist??

does that mean it aint really raining here if the sun is shining on your door too? lol

No, I just don't blame every performance problem on it because it really doesn't cause as much stutter as people say it does.

Again, W1z wouldn't be recommending SLI 970s for 4k at max settings(no MSAA) if there were noticeable stuttering, and if we believe "your friend" stuttering so bad at 1080p that the game isn't playable.

Oh, and if you don't believe W1z, the guys over at NCIX did a performance analysis too. And you guessed it, they also recommend SLI 970s(or a Titan X) for 4k. Man, you'd think if the 970s and the 0.5GB RAM issue was so bad, either W1z of the guys over at NCIX would have experienced...right...right?!? But "your friend" says his 970s are to blame, so it must be. Definitely no chance of it being some other problem. Those professionals, that deal with graphics card testing FOR A LIVING, have no clue what they're talking about. But "your friend", he's the source we should believe...:rolleyes:


seeing as how w1z has already cleared up what settings the tests were ran at, and they were not all the way up it does not shock me to hear it can play the game at high res.

You make it sound like he ran it on low settings. The only settings not maxed out were MSAA(already explained why), Reflection Quality, Grass Quality, and PostFX. And the only one of those that affect memory usage is Reflection Quality, and it doesn't make a big difference(and the visual difference is unnoticeable when playing the game).

And again, he ran it at 4k using almost maxed out settings. Your claiming it is a stuttery mess at 1080p! Even at absolute max settings, even with some MSAA, SLI 970s should be running just fine. As Steevo pointed out, it seems your friend is having other issues.

aint nobody in their right mind going to go out today and pay £600 (less now as they have cut prices since the 500mb-gate) for 2 of these to not be able to run max settings. makes more sense to save £150 and take the 295 as that way you get about the same frames (more at higher res like 4k) and you can turn the setting up to fill all the vram without worry.

The 295x2 has its own set of problems. The first one being, at least here in The States, it is about $50 more than a pair of GTX970s. Sure the 295x2 is about 15% faster at 4k than a pair of GTX970s if it stays at full speed constantly. Which is can do in an open test bench, but once I put my 295x2 in my case it would throttle after about 15 minutes of gameplay and framerates would drop 10-15FPS. I tried turning the fan around to pull in cool air, and managed to stop it from throttling as quickly, but it still throttled after 30-40 minutes of gameplay, and I don't think the rest of the components in my case were too happy with 500w+ of heat being dumped in the case. Plus when using the 295x2 my room got uncomfortably hot, which hasn't happened since I had my GTX470s...
 
Thank you for the review - I have a query - what was the CPU usage while playing GTA V ? Far cry 4 shows 70% cpu usage on i5 4670K.

The reason to ask this question is that i am doubting that my 4 Cores CPU will hold good in future, i might have to pull the trigger for 8 cores CPU in a year or 2.
stRjVi3.png


i7 4790K @4,7GHz, 8GB DDR3 RAM @2133MHz, GTX680 @1275MHz.

Game settings: 1080p, "normal" texture quality and all other settings at "high", FXAA, no MSAA. 85-110FPS while driving in city.
 
Is this comparison test sponsored?
 
Is this comparison test sponsored?
Yes, I've sent a crate of green bananas to W1zzard. Not sure if he received them though. Just to be fair, i wanted to send him a crate of red bananas too but couldn't find any in my local shop.
 
Last edited:
Back
Top