Recorded at 120 FPS Recorded at 240 FPS Using high speed video, the stutters/hitches of each card can now be seen clearly. Full report: http://techreport.com/review/24051/geforce-versus-radeon-captured-on-high-speed-video Here's the frame time plot:
Thank you taufikp! I think I made the better decision to get the gtx 670 rather than the HD 7970. I moved the mouse to give you rep....but....lol.
I do not doubt the results, but I have to wonder if there was something about the system that favored the nVidia card over the ATI one. Testing hardware is difficult at best due to all the variables that can come up, and GPU's are no exception. If the same system and drives are used for testing both cards, were proper precautions such as reloading the entire OS taken to ensure that there were no left over drivers to conflict with the new drivers etc. It would also be helpful to post the exact system configs along with any and all other software installed and used during the tests such as MSI Afterburner etc. Having information as the the state of the GPU and the CPU during the test would also be helpful in determining if the results can be validated or not. I read the article that you linked and it does little to explain system config or any other software that was involved so determining the bias of the article is impossible without all the other factors.
That's a fair observation. I remember when the performance was best when it was Intel CPU + ATI card and AMD CPU + Nvidia card....weird.
@Gearjammer The article in the OP is (another) follow up to their original more indepth article here.
GCN is a brand new architecture, was obvious that they will need some time before reach a certain maturity. It's like for the cars or everything else, the first models have always a bit more problems, but from what i saw AMD is aware of the problem and working on it for the next catalyst builds : http://techreport.com/news/24136/dr...ame-latency-troubles-to-be-fixed-with-updates
Thanks MJP, it wasn't really clear from the OP that this was the case. I read the entire article you linked and I see that this is definitely an issue and according to the link that HKF0x posted, AMD is working on it as we write. It is always helpful when posting benchmarking links to link the main article and then any article that shows specifics so that there is less confusion in the future taufikp, but good find none the less,
Actually there are several posts in this forum, and all related to that Radeon vs GeForce 'spike' frame time articles. I also bought my GTX670 (originally I wanted to buy an HD7900 series) based on those articles' conclusion List of the articles I read: - A New Look At Game Benchmarking : http://techreport.com/review/21516/inside-the-second-a-new-look-at-game-benchmarking - HD7950 vs GTX660 Ti Revisited : http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited - HD7950 Stumble In Windows 8? : http://techreport.com/review/24022/does-the-radeon-hd-7950-stumble-in-windows-8/8 And then this article was published after I bought the GTX670! >.< - Radeon Driver To Be Tweaked: http://techreport.com/news/24136/dr...e-radeon-frame-latencies-in-series-of-updates
A new driver for AMD Radeon cards, 13.2 beta, has been released. Here's the full report of how this new driver successfully reduce the frame-time latency in Radeon cards: http://techreport.com/review/24218/a-driver-update-to-reduce-radeon-frame-times
Glad to see AMD is addressing this issue and I am looking forward to further developments in this area. This will likely impact our abilities on track as it will present smoother more consistent frames and hopefully will not affect the control input lag. If all goes well and in the direction I would like to see, it would negate the need to use things like vsync to smooth out the graphics.
Exactly. I'm going to reinstall my HD5770 to try this 13.2 beta driver and doing another benchmark. I'm glad that AMD has taken the frame-time latency report seriously. Since I'm planning on buying an HD7900 series card as well, this driver improvement is very good news indeed.
Well that's it then, its just tearing from having vsync off, stick it on and is smooth, well it is on mine anyway?
I get jerky fps on my GTX 460 SLI when i have no vsync video off and the fps is anything below 140. As the average fps is lowered, around 60 fps the game is pretty much an unplayable mess of jerky unsmooth gameplay. I have the settings lowered to the point to supply me with 140+ so it's smooth and enjoyable but i miss the graphics. Thanks taufikp, i found that very insightful read. I used to brush over the types of reviews that used frame timing because i was lazy to wanting to take the time to understand it fully but now i do and makes a lot of sense to use them in the future for my own means of comparing cards. Please do let me know your results with the new driver, i'm still sticking to my GTX 670 though because i had a lucky bargin but if the results are good and smoothness in rf2 it would mean ATI...sorry...AMD are back on track and it makes complete sense for up-graders to get a HD 7970 over a GTX 680 at such a lower price.
For jerky movement in rF2, try to disable the Replay feature first. In my system, disabling Replay = silky smooth movement, but of course now I don't have any 'documentation' of my race If you don't like that, then this next solution is worth to try: if you have more than 4 GB of RAM (system RAM not video RAM), you can edit the PLR file and change the line that enabling you to save the whole Replay data into RAM first before writing them into the disk after the race is over.