Skr13

GPU Charts 2012

14 posts in this topic

I did some research about the latest graphics cards reviews/tests of this year and joined the results in this topic.

Average Performance 1920x1080

Nvidia Drivers Performance Index

AMD Drivers Performance Index

Rating DirectX 9 & 11 1920x1080

GPGPU Performance Index

TDP Watts

Fan Noise Idle & Load

0

Share this post


Link to post
Share on other sites

3D Mark 11 1920x1080

Sleeping Dogs FPS

Battlefield 3 FPS

Far Cry 3 FPS

Assassins Creed 3 FPS

Max Payne 3 FPS

2012 was a year full of great games, new graphics cards appeared every month and I expect 2013 will gives us more hardware and software exciting stuff. :P

2

Share this post


Link to post
Share on other sites

nice collection of data here :)

however my system still doesn't agree with any of the FC3 numbers, maybe has to do with Lucid MVP HyperFormance?

I'm running at 1920x1200 Maxed w/ 2x MSAA and 16xAF and avging 45 FPS

I wonder what their FOV settings are, because that plays a big role, mine is at 80 something.

Also I'm running WHQL 310.70 and they are running Beta 310.64, although that shouldn't play a big role.

0

Share this post


Link to post
Share on other sites

nice collection of data here :)

however my system still doesn't agree with any of the FC3 numbers, maybe has to do with Lucid MVP HyperFormance?

I'm running at 1920x1200 Maxed w/ 2x MSAA and 16xAF and avging 45 FPS

I wonder what their FOV settings are, because that plays a big role, mine is at 80 something.

Also I'm running WHQL 310.70 and they are running Beta 310.64, although that shouldn't play a big role.

Sorry can´t help you, In fact I have not yet tested the game, and I never try to "catch" reviews numbers even If I have a identical system, always will be some difference.

3Dmark, Fluidmark, Unigine Heaven and other bench programs are best to directly compare with other systems, because they do the same routine for every test, and in the other hand, games benchmark numbers depend on map, for example.

The new GeForce 310.70, according to Nvidia when using HDAO run a bit faster than the 310.64, we measure three percent extra performance, which means absolutely but only a Fps more - and in the case of the GeForce GTX 680th »PcGamesHardware«

Anyway, it is always better if you have more fps than charts, if were the opposite could be frustrating ...

0

Share this post


Link to post
Share on other sites

Thanks for the info there Skr13, quick overview to reflect the performances as things stand currently :)

One main aspect of peoples decision when choosing AMD or Nvidia is how smooth/playable the gameplay is when the your getting low FPS. The effect is only with SLI/CF: I remember way back in March the perceived smoothness was in Nvidia's favour as 30fps (that review I recall with with BF3) is playable without issues; however AMD needed 50~60fps to match the same level. I haven't come across of an update of this situation as described by HardOCP - maybe if someone has CF setup they could post their view...

source

0

Share this post


Link to post
Share on other sites

[uPDATE] GameGpu 2012 Games Performance Overview

I will not provide the usual translation and personal comment due large number of charts/pictures, and diverse information about many games, but I recommend everyone to take a look at their charts.

They made the tests with latest games Patches/Updates available and most recent graphics card drivers.

GameGpu!

»LINK«

1

Share this post


Link to post
Share on other sites

Thanks for the update Skr13. Totally understand the difficulty providing your own commentary for all those games and appreciate it when you can :)

Will be interesting to read other opinions here as they testing recent demanding games to give a good broad overview. Noticed BestJinjo gets around commenting there too ;)

From my initial reading and correct me if I'm wrong: I should note (In Borderlands 2) that although AMD cards can render PhysX, they can only render some elements and not more complex aspects such as cloth, dripping blood, etc that Nvidia GPUs excels on their own engine. Unless the AMD drivers and patches has changed things I'd say Nvidia running proper High PhysX on the GPU was a bit unfair as it's very taxing when AMD could be offloading more elements to the CPU. I checked some of the benches against Hardocp 1600p and they tally up despite the slight differences with the ingame settings. Regarding Farcry3 v1.04, I did play it maxed with MSAAx8 and the VRAM crept up to 1600MB but the game wasn't smooth/playable due to lack of GPU bandwidth - each GTX 680 is 205.1 GB/s.

0

Share this post


Link to post
Share on other sites

Thanks for understanding my position SMiThaYe.

I agree with you, I found unfair put PhysX games on bench plans, running from GPU is always faster and better than CPU.

I remember back when I had my HD 4870 trying to run NFS Shift along with PhysX enabled, I had poor frames and 100% CPU usage(Q9550), terrible.

GTX 460 saved my life when I bought it.

0

Share this post


Link to post
Share on other sites

From my initial reading and correct me if I'm wrong: I should note (In Borderlands 2) that although AMD cards can render PhysX, they can only render some elements and not more complex aspects such as cloth, dripping blood, etc that Nvidia GPUs excels on their own engine. Unless the AMD drivers and patches has changed things I'd say Nvidia running proper High PhysX on the GPU was a bit unfair as it's very taxing when AMD could be offloading more elements to the CPU. I checked some of the benches against Hardocp 1600p and they tally up despite the slight differences with the ingame settings. Regarding Farcry3 v1.04, I did play it maxed with MSAAx8 and the VRAM crept up to 1600MB but the game wasn't smooth/playable due to lack of GPU bandwidth - each GTX 680 is 205.1 GB/s.

Their testing is confusing. They label the graph as PhysX High but in the game control settings, PhysX is disabled. The only way to get PhysX to work with AMD cards is to use the .ini hack in BL2 and offload it to the CPU, in which case the performance hit is huge, and yet GameGPU's graphs do not show this on the Radeons. Also, FPS testing in BL2 doesn't reveal the significant stutter HD7970Ge has compared to GTX680:

More or less the moral of the story is some games run way better on NV (BL2, Assassin's Creed 3, WOW, Shogun 2, etc.) and others on AMD (Sleeping Dogs, Hitman Absolution, Dirt Showdown, Metro 2033, etc.). It's just not possible this generation to buy the best GPU for most games as was the case with 8800GTX/280/285/480/580.

We even had a situation of GTX680 dominating HD7970 for at least 10 months in BF3 and now it's reversed:

I hope GTX780 is a full-fledged GK110 so the choice for the flagship GPU next round is more clear cut! :)

0

Share this post


Link to post
Share on other sites

BestJinjo, I'd say they are either trying to confuse the readers not informing them beforehand of their methodology surrounding PhysX, or they may not be 100% aware of what they are doing in how the game works on AMD vs the full effects Nvidia is capable of.

In my experience the GTX 680 for BF3 (tested with single and SLI) doesn't show stuttering as you can see at 0:08 when the view pans, that was on 310.70s and 310.90s (same playtime) that are of identical performance.

I'm always watchful of people showing their GPU stuttering on youtube, the minority are actually demonstrating a twitchy mouse. If I had done this testing, in the next demo I would have ran in a straight line down the bank with OSD of GPU from Afterburner and PIP CPU activity to watch for spikes. Despite that, the video does show issues which would be a determining factor for would-be-buyers. Funny because I choose GTX680s because the game was much smooth at lower FPS than AMD which required FPS into the 50s to remain smooth - I've mentioned that before.

I don't like nit-picking but NordicHardware incorrectly listed the video title as frame times when he had meant to say framerate; frametime is the delay (ms) until a completed frame is rendered to the display.

0

Share this post


Link to post
Share on other sites

SMiThaYe, I think for those who enjoy PhysX effects, NV is the way to go no doubt about it. I don't like to rely on 1 website for GPU testing as sometimes they miss out on issues just like GameGPU didn't mention stuttering problems on HD7900 cards in BL2. Similarly stated, Far Cry 3 is a stutter-fest on GTX690 compared to HD7970 GE CF and most reviewed didn't show this through FPS testing alone:

http://videos.pcgame...-in-Slow-Motion

http://www.pcgamesha...hmarks-1036726/

It appears both companies have stuttering issues and it's practically impossible this generation to get the best setup as a result. I hope these drives issues get resolved by both sides, although admittedly AMD's cards have suffered more in latest games.

It just pains me to see games like Far Cry 3 with worse graphics than Crysis 1 that perform so poorly on $1000 of GPU setups:

Consolification is killing PC gaming. We keep upgrading every new generation and games just get more GPU intensive without getting much better looking. Programmers are optimizing games like crap (Hitman Absolution, Sleeping Dogs, Assassin's Creed 3, NFS:MW all have serious optimization/performance issues and not a single one of those games looks as good as Crysis 1).

I wouldn't mind Crysis 3 bringing bringing $1000 worth of GPUs to 25 fps but it better look like it's worth it :) 5 years since Crysis 1 came out and I am still searching for a game that wowed me technically/graphically-wise the same way Crysis 1 did back in 2007.

0

Share this post


Link to post
Share on other sites

Interesting watching that FC3 vs Crysis video (was that vanilla because it looked very nice) as it was clear as day how much FC3 is being held back for consoles. From the previews I was initially totally put off from buying FC3 because it looked too cartoonist from what I'm used to as a hardcore gamer, only later on I bought it and didn't think too much about the graphics after 80 hours playtime mostly mucking about. Sure there are annoyances after that much playtime.

The realistic ripples in the water are missing and dissipate far too quickly, the trees are using God-mode in FC3 and saying 'hit me hit me', water lacks enough particles to appear realistically and is very apparent on the waterfall. Isn't that hard to do now and been possible since playing Bioshock which is an amazing game worth buying on PC! But if I push it one step further to how things could be / should be now then take a look at Battleship movie water effects.

Battleship water particles

I'm sure there are plenty of games that have brought a couple GPUs to their knees but that was because of retail editions being released as full versions when in fact they are so buggy I'd call them betas and I've played hundreds of betas (and a dozen alphas). I've said it for many years to others but there has always been a rush to release titles to fit precise release dates and compromise achieving a higher quality. The main issue I have after all these years is publishers targeting Christmas and releasing a ridiculously amount of anticipated titles in such a short period that you'd think there weren't enough weeks in the year! This treats the consumers with a lack of respect as we take the game in the current state and eagerly await a patch just to play the darn game because we should be grateful. Extended release dates are very rare and would like to know what percentage of PC titles are pushed back for bug-fixing, and general optimisation.

This leads to my past wish for more flexible release dates being wiped. The drive is for constant yearly updates of stale cross-platform titles for the best returns which in fact leaves the industry in a perilous position when a big hitter endures falling sales there isn't much room to manoeuvre with less quality choices currently available - now arrives the growing popularity of free-to-play titles or crowdfunding to appeal to disappointed gamers looking for alternatives in the problems the industry is exposed to.

0

Share this post


Link to post
Share on other sites

HardOCP - 2012 AMD and NVIDIA Driver Performance Summary Review

 

We have evaluated AMD and NVIDIA's 2012 video card driver performances separately. Today we will be combining these two evaluations to show each companies full body of work in 2012. We will also be looking at some unique graphs that show how each video cards driver improved or worsened performance in each game throughout the year.

 

LINK: http://www.hardocp.com/article/2013/03/04/2012_amd_nvidia_driver_performance_summary_review/1

0

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now