Sign in with
Sign up | Sign in
Your question

Higher Frames Per Second Does Not Mean Smoother Gameplay - A must read

Last response: in Graphics & Displays
Share
a c 171 U Graphics card
December 16, 2012 7:43:27 AM

I have been looking at getting a gtx660 or hd7870 and had almost certainly decided on the 7870 as it seemed faster overall, if you only look at FPS. But just recently I have stumbled across some articles about this new development. the first article i have linked gives a good visual display of what is going on. It is good to read all the articles i have linked to give you a broader idea of what is going on. But there you have it, a card with higher average FPS may not give you smoother gameplay.

http://techreport.com/review/24051/geforce-versus-radeo...
http://techreport.com/review/23527/review-nvidia-geforc...
http://techreport.com/review/23981/radeon-hd-7950-vs-ge...
http://www.mvps.org/directx/articles/fps_versus_frame_t...
http://www.tomshardware.com/forum/380802-33-frames-meas...

I am still a bit curious about the matter as the tests would be without vsync, and also looking at Toms recent article about crossfire/sli stuttering shows all fromes sync perfectly when a frame rate cap is on so long as the frame rate doesnt dip below the cap. So having vsync on would differ the results possibly. But at this stage, it looks like the gtx660 vs 7870 battle http://techreport.com/review/23527/review-nvidia-geforc... is much closer and the geforce takes a very slight edge in regard to gameplay smoothness. Looking for ward to hearing on what you all think.
a b U Graphics card
December 16, 2012 7:46:20 AM

I will def read these. But I have heard that AMD cards have been a bit stuttery recently for some reason (I assume drivers). I have a MSI GTX 660 and it performs FLAW!LESS!LY! (306.97 Driver)
m
0
l
a c 271 U Graphics card
December 16, 2012 11:34:16 AM

The first article just reinforces why I don't really want an AMD/ATi card in any of my rigs, ever! :lol: 
m
0
l
Related resources
a b U Graphics card
December 16, 2012 11:47:13 AM

It's a good thing mods are here... all-out nVidia vs AMD warfare in 3... 2... 1...

But seriously, should be some interesting discussion here. I'll be really interested to see what the AMD fans come back with. Too bad frame latency benchmarking didn't begin years ago.
m
0
l
a c 271 U Graphics card
December 16, 2012 12:22:59 PM

sam_p_lay said:
It's a good thing mods are here... all-out nVidia vs AMD warfare in 3... 2... 1...

But seriously, should be some interesting discussion here. I'll be really interested to see what the AMD fans come back with. Too bad frame latency benchmarking didn't begin years ago.

I've always been an exponent of minimum frame rates as that's what I've always believed that's what produced the jerky motion as seen in some of those vids so its a bit of a surprise to see that it may actually be caused by the sudden but consistent highs that result in the higher average fps rate. I wonder how much that has to do with the fact that a few years ago the average refresh rate of LCD's was 55/60hz?
m
0
l
a b U Graphics card
December 16, 2012 12:31:50 PM

The problem with minimum frame rate is that it's still an average taken over the course of a second. People assume that a minimum of 20fps means the longest frame render was 50 milliseconds, when in fact it simply means the second with the longest average frame render time rendered 20 frames.

That 20fps could have been a mix of 30 millisecond frames (~33fps) and 70 millisecond frames (~14fps). That's the problem with frames/second as a whole (even minimums) - it's always a mean average that smooths out dips and spikes.
m
0
l
Anonymous
December 16, 2012 1:25:56 PM

Haha, well my two 5850 in x-fire still kick nicely :)  FarCry3 on Ultra 2xAA, no probs at all :) 
m
0
l
a b U Graphics card
December 16, 2012 1:33:14 PM

Anonymous said:
Haha, well my two 5850 in x-fire still kick nicely :)  FarCry3 on Ultra 2xAA, no probs at all :) 


I almost did that but decided to take a 5970 instead (effectively the same thing but in one card). That card is dead now though. No Far Cry 3 for me until I get a proper replacement next year.
m
0
l
December 16, 2012 1:38:24 PM

Will just add.

Hard drive access causes drops in performance. For an extreme experience of this, scan your hard drive with your AV. Now, play a game.
m
0
l
a b U Graphics card
December 16, 2012 3:03:53 PM

MajinCry said:
Will just add.

Hard drive access causes drops in performance. For an extreme experience of this, scan your hard drive with your AV. Now, play a game.


Pro tip: don't leave a virus scan running while gaming.

Also, SSD.
m
0
l
December 16, 2012 3:47:29 PM

I'm spoiled by higher FPS. When I had a single 7970, the FPS on most games was high but seemed kind of jerky...not sure how to give a technical explanation lol. When I went to 2 7970s it seemed to be quite a bit smoother, but then I started having micro stutter. Now that I'm on 3 cards, all games play smooth as butter.

I will say that arguments on FPS are the same kind of arguments as amd vs nvidia. It all comes down to whatever setup the user prefers the most.

I was playing borderlands 2 on the pc, but then did some stupid attempts at overclocking my CPU and fried it. So while my computer was down, I played the game on my ps3. After an hour of playing, I had to put it down...the FPS on ps3 was making me feel sick. So yeah, I'm now spoiled and now my ps3 is at work since I no longer want to use it lol
m
0
l
a c 191 U Graphics card
December 16, 2012 4:10:26 PM

I have reason to believe my HD7870 (and HD7770) do not always maintain clocks at 3D levels, because I see a different kind of jitter from time to time that indicates a FPS drop. I think it's driver-related, because this behavior improved a lot after the last Catalyst update I installed (12.11). Still, in looking at the videos, I saw BOTH cards "jumping," just that the nVidia card did smaller jumps more often; yes it was a little smoother, but nothing to make me scrap my HD7870. I suspect that a settings change somewhere could make a pretty big difference, and believe that should be explored before coming to final conclusions. Maybe there is a specific [type of] detail setting that AMD cards (or their drivers) don't handle well, and I'd like to know what that is.
m
0
l
a c 171 U Graphics card
December 16, 2012 5:51:56 PM

sam_p_lay said:
It's a good thing mods are here... all-out nVidia vs AMD warfare in 3... 2... 1...

But seriously, should be some interesting discussion here. I'll be really interested to see what the AMD fans come back with. Too bad frame latency benchmarking didn't begin years ago.

Myself not being a fanboy of either and have owned both over the years, in the same price range there still isnt much difference between ati/nvidia, even with this new benchmarking method. the gtx660 vs 7870 battle for example, amd gives more fps but loses ever so slightly in 99th percentile frame latency, both cards are the same price where i live and there isnt nearly enough difference shown to swing a buying decision either way. I am thinking for the moment though, i might be going with the 660 because any kind of lag/stutterring is of more importance than maximum average FPS to me.

And if you are playing with vsync on and are hitting your vsync cap most of the time, you probably wont see any stuttering like they showed in the articles i linked, as Toms microstuttering article described. I'd like someone like the good people at Toms to clarify Techreports results, and do some tests with vsync/tripplebuffering/framerate caps on etc.
Is it a driver issue that can be fixed? or is it something lacking in AMD's hardware that nvidia has? something simple like an output buffer to the monitor?
m
0
l
a b U Graphics card
December 16, 2012 7:15:14 PM

Well I've actually spoken to Chris Angelini about this, and he didn't mention anything about flaws in the methodology or any obvious problems. On the contrary infact he was pretty impressed with Scott's work.

With regards to testing using a framerate cap, did you see the results with RadeonPro in the 7990 review? Massive improvements, just a bit complicated and time-consuming having to recalibrate it for each game (and actually you'd need to recalibrate for the same game each time you adjusted your settings and affected framerates). Still, Mautari did some really impressive work there, makes you wonder why AMD were unable to do this for their own products (they need to hire this guy!).
m
0
l
!