Grand Theft Auto V Single GPU Performance Review Part 1

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,634
Grand Theft Auto V Single GPU Performance Review Part 1 - This is Part 1 of our full evaluation of Grand Theft Auto V's video card gaming performance. In this part we look at single-GPU performance on eight video cards at 1440p and 1080p. We discuss what settings are playable, architecture performance differences, and what settings have the biggest impact on your gaming performance.
 
There's a point in all the graphs, around the 340 second, where almost all the Nvidia cards go leave AMDs cards behind. Do you know what's happening there?

EDIT: Bah, nvm. Missed it on the article, you don't know what's happening.
 
Last edited:
Remon said:
There's a point in all the graphs, around the 340 second, where almost all the Nvidia cards go leave AMDs cards behind. Do you know what's happening there?
Click to expand...

It is a section of road that is more out in the open, with a lot less traffic, farther away from any cities or town and there is just mountains on both sides of the screen till we get to the end where we drive into another town and park our car.

There isn't anything special about the scene that we can see, and I really don't know why NVIDIA GPUs are pulling ahead there, other than perhaps they do a better job at rendering the higher quality of grass/vegetation with less traffic and less town objects. Before that, we are traveling through towns and lots of traffic. In fact, right before that, we travel down into a dense area that drops performance toward the coast through thick grass, but performance goes lower there, not higher.

I really am at a loss for what is causing that. I think when we get to doing individual game setting comparisons for performance we may find our culprit. That will tell us which visual quality setting is more efficient on NVIDIA GPUs. That bit of testing will be in the image quality part when we test image quality and what options demand more performance.
 
Nice guys, thanks for your efforts, keen to read this for my mornings work haha ;)

EDIT: I personally would have like to see single Titan X performance at 2560x1440, but all good, there might not be many people using that configuration.(i am).
 
Last edited:
I'm seeing similar results with my setup, turning down the grass setting one notch really helps out in the country.

Is part two mGPU setups?
 
Awesome, thanks for putting in the time and showing results for us Kepler users!


I'm really looking forward to the upcoming multi-GPU review.
 
does anybody have any experience running an OC'd GTX 780 at 1440p in this game?

what sort of settings can you enable?
 
spintroniX said:
sorry bud, I meant single GPU :)

the present review has performance for 1080p, but I'm curious about 1440p
Click to expand...

Shoot, my bad...for some unknown reason, I had multi-GPU in my head after I read and responded to your initial post. :p
 
@Kyle and the crew. Thanks again! These reviews keep my reading your website. :)
 
Fixed. Thanks for the extra eyes brother! - Kyle
 
Last edited by a moderator:
Great, thanks guys! that 285 Tonga is really shining in this game so far, solid price to performance, clock it up and see where it runs!
 
Any thoughts as to whether or not drivers could improve the performance of the older gen, Kepler and GCN 1.0, GPU's? Just thinking there maybe some driver optimizations that could be done to improve things? Mind you even if it is possible I guess it's not overly likely that NV or AMD would be highly motivated to do it.
 
Man... disappointed in the lack of 290x 8gb and 960 4gb cards in this review... This game would have shown differences more than any other, imo.
 
"At 1080p the GTX 970 or Radeon R9 290 are your best deals in that resolution. Of course, if you can afford it, an R9 290X would be even better at 1080p and allow higher settings."

The charts suggest that the GTX 970 is better at 1080p....with the 290X lagging behind with the same settings.
 
jamesgalb said:
Man... disappointed in the lack of 290x 8gb and 960 4gb cards in this review... This game would have shown differences more than any other, imo.
Click to expand...

Maybe two 8GB 290Xs at 4k might be somewhat useful but a single 8GB 290X is silly. They're not hitting a VRAM limit at 1440p and at 4k a single card won't have the performance to hit a VRAM ceiling.
 
jamesgalb said:
Man... disappointed in the lack of 290x 8gb and 960 4gb cards in this review... This game would have shown differences more than any other, imo.
Click to expand...

Don't have what we don't have, I've been trying to get 8GB 290X's for months, every manufacturer keeps telling me they are out of stock for reviewers. I have our first 4GB GTX 960 on its way for a full evaluation.
 
Good of you to do such granular reporting over multiple articles.

It would have been interesting to see how 4xMSAA or DSR/VSR stacked up.
Especially considering the extra vram use.
No doubt you will cover this later anyway.

I found the crawlies too distracting in this game so use VSR 1440p with FXAA on my 290x and found it surprisingly good performance in comparison to 1080p + FXAA.
But I get majorly CPU bottlenecked at times which holds my GPU back, so I suspect a faster CPU wouldnt be helped so much by DSR/VSR as other faster settings may already be in use, keeping the GPU occupied.
(I'm running a 2500K @ 4.3GHz and sometimes I get an almost flatline max CPU use on 2 or 3 cores!)

I am especially looking forward to seeing results on which settings affect CPU use the most.
 
i would love to see a cpu comparison for this game. Rumors are it does better with more cores and seems to be the firs i have heard of doing so. I think it would make a great article
 
primetime said:
i would love to see a cpu comparison for this game. Rumors are it does better with more cores and seems to be the firs i have heard of doing so. I think it would make a great article
Click to expand...

Agreed, would also love to see this. If and when you do can you please include a Phenom x965, maybe clocked to 4ghz or so as comparison?
 
schizrade said:
Any titan black users? Does the added vram help the Kepler chips?
Click to expand...

I wouldn't expect so, seems to be more architectural. Odd's are the drivers are optimized for Maxwell hardware and the tweaks haven't been applied to Kepler gear as of yet.
 
My 7950 plays perfect for me, I don’t notice any hiccups during game play at all with everything at very high.
Im glad to see the game needs high end cards to run maxed out, but still looks and plays very well on older tech imo.
Hats off to RS.
 
Running the game with two 670's in SLI on high and very high settings, grass is set to ultra and able to maintain 60FPS. I did have to turn down population density and population variety to stay somewhat in spec for the memory.
 
Last edited:
I'm disappointed to not see any Titan X results. It's easy to expect them to be better than the others but it would be good to quantify how much better.
 
So, is this the first game in a while which "legitimately" pushes the throttles of GPU's?
 
Kor said:
I wouldn't expect so, seems to be more architectural. Odd's are the drivers are optimized for Maxwell hardware and the tweaks haven't been applied to Kepler gear as of yet.
Click to expand...

Yeah that is what I was taking away from it. I did find some guy that was benching tons of cards on this and black fares well, just behind the 980. Maxwell is just more consistent.
 
Would love to see OC results for these cards. I know the 780 benefits quite a bit from a good overclock.
 
When the 4K review comes, I would like to see SLI / XF compared with a single card with Gsync / Freesync.
 
primetime said:
i would love to see a cpu comparison for this game. Rumors are it does better with more cores and seems to be the firs i have heard of doing so. I think it would make a great article
Click to expand...

Techspot already did it:

http://www.techspot.com/review/991-gta-5-pc-benchmarks/page6.html

It looks like the game tops-out at 4-5 principle cores used, which is why the FX 4 to 6 to 8 cores spans less than 20% performance increase.

That also explains why the Phenom II 980 4-core is competitive with lower-clocked 6 and 8 core processors.

The conclusions page also supports this observation, when the reviewer reported 60% CPU usage on a n 8350E, and 90% utilization for the Core i3. That SCREAMS 4-5 principle threads.

Also, good to see the i3 staying competitive. Looks like we can still highly recommend it for value gaming PC builds :D

Also, on-topic: thanks [H] guys for pushing these cards to the limit. Site like Techspot don't even bother trying that hard, and I really only read their reviews because they have regular CPU benchmarks.

Can you believe Techspot recommended 1080p NORMAL TEXTURES for the GTX 960?

Those using graphics cards with a 2GB memory buffer will be best off using the 'normal textures' with FXAA. When gaming at 1080p with these settings we found affordable graphics cards such as the Radeon R9 270 and GeForce GTX 660 Ti, GTX 760 or GTX 960 were able to deliver smooth playable performance.
Click to expand...

And then if you look THEY NEVER TESTED THE 960, they just made assumptions based on the results of other cards. So yeah, they obviously didn't put any effort into this :(
 
Last edited:
You must log in or register to reply here.
Back
Top