New testing method intro'd by TechReport/HD7950 VS GTX660ti

[Novuake]

Hi forum.

I would like to start a discussion on the topic brought up by TechReport concerning an FPS latency issue found on the HD7950 and possibly on all GCN cards.

Here are the articles and followups.

http://techreport.com/review/23981 [...] -revisited
http://techreport.com/review/24022 [...] -windows-8
http://techreport.com/review/24051 [...] peed-video
http://techreport.com/blog/24133/a [...] ng-methods

Now the internet seems to be only catching onto this now, TechReport has been using this method of testing for almost a year now.

Nvidia fanboys are loving it and AMD fanboys are outraged.
Personally I believe this testing method is still VERY new and not a good measure of real world smoothness just yet but I would very much like to see this investigated by Tom's and other sites. I also believe this is still something to be considered when purchasing these card.

AMD has made an official statement concerning this, vowing to fix this issue in future driver updates and CAPs(as the one for Borderlands 2).
IMO GCN is still a very new architecture and AMD is short n resources compared to Nvidia, they still need to figure out the nuances of GCN.

What is your opinion?
Which card is better for you do you think and why?
Does the FPS latency issue effect your choice in graphics card?

Please keep this a rational discussion void of fanboy ravings.

EDIT : Dag nabit! Should have made this a discussion... Can a moderator please change this if possible?

Another link thanks to renz496 :
http://www.pcper.com/reviews/Graphics-Cards/Frame-Rating-New-Graphics-Performance-Metric

EDIT : 17-01-13. Latest update from TR. The release of Cat 13.2 is improving the latency issue.
http://techreport.com/review/24218/a-driver-update-to-reduce-radeon-frame-times

Another update from PCPER thanks renz496 :
http://www.pcper.com/reviews/Graphics-Cards/Frame-Rating-Part-2-Finding-and-Defining-Stutter

As I said in my reply, as far as I can see the only way to clearly really convey frame latency to the consumer is via high-speed video. Although this still does not make it very simple to understand the real word effect.

 

[redeemer]

Yeah anyone remember the VSYNC issues with the 600 series?? No need to panic that what driver updates are for.

 

[redeemer]

The Techreport has always been biased anyhow. The funniest things for me is that most of the reviewers compared a 600 series that has autoboost to a stock 7000 series card from AMD. Then everyone realized what happens when you overclock a Tahiti.

I honestly prefer Guru3d and Xbitlabs for reviews

 

[Proximon]

The title of this topic has been edited by Proximon

Sorry, looks like a forum bug is preventing the thread type change. It's usually easy to do. Ask me again later

 

[bigshootr8]

But as a reader of articles you have to kind of go through the fluff per say and understand what you can take away from it to sway one way or the other. I've seen the article for the 660Ti beating the 7950 however I know longevity wise the 7950 will probably hold better. For me I play a lot of nvidia games so I tend to like having a card that has a more full feature set for the games I play. The only feature that amd offers that I would desire is eyefinity so I feel I get a more complete product with nvidia at the moment. However if I wasn't playing games with physx I would probably be rocking a 7970.

 

[Novuake]

The title of this topic has been edited by Proximon

Sorry, looks like a forum bug is preventing the thread type change. It's usually easy to do. Ask me again later

Thanks for the effort, will PM you later!

 

[Novuake]

But as a reader of articles you have to kind of go through the fluff per say and understand what you can take away from it to sway one way or the other. I've seen the article for the 660Ti beating the 7950 however I know longevity wise the 7950 will probably hold better. For me I play a lot of nvidia games so I tend to like having a card that has a more full feature set for the games I play. The only feature that amd offers that I would desire is eyefinity so I feel I get a more complete product with nvidia at the moment. However if I wasn't playing games with physx I would probably be rocking a 7970.

A very legit reason to prefer Nvidia.

 

[Novuake]

The Techreport has always been biased anyhow. The funniest things for me is that most of the reviewers compared a 600 series that has autoboost to a stock 7000 series card from AMD. Then everyone realized what happens when you overclock a Tahiti.

I honestly prefer Guru3d and Xbitlabs for reviews

I actually like TR a lot, but the selection of games used there are a little biased with the exception of Skyrim.
I respect TR for their innovative idea though.

Another site that I like for reviews are Tweaktown although they always neglect comparing OC with OC.

 

[bigshootr8]

I would almost prefer if sites had a overclocking side of cards seperate from cards that are not overclocked. You get so much bias when you throw in well I have x card and when I overclock it goes beyond what the z brand goes.

 

[Novuake]

I would almost prefer if sites had a overclocking side of cards seperate from cards that are not overclocked. You get so much bias when you throw in well I have x card and when I overclock it goes beyond what the z brand goes.

Yeah and its quite a lot of effort to try and compare the numbers if they are not in the same review thank to so many variables.

 

[Mousemonkey]

Yeah and its quite a lot of effort to try and compare the numbers if they are not in the same review thank to so many variables.

One of those being the country that one may be in...

 

[bigshootr8]

Oh yea most certainly there are some countries that have cards that are crazy mofo expensive.

 

[Novuake]

One of those being the country that one may be in...

Precisely, pricing, room temps, availability. SO MANY!!!

 

[redeemer]

what about Kepler autoboost?

 

[Mousemonkey]

Oh yea most certainly there are some countries that have cards that are crazy mofo expensive.

And then there are other things that one might want to do with their cards (F@H for instance, shameless plug ).

 

[bigshootr8]

lol monkey stirring up the drama in the pot with the brands

 

[Mousemonkey]

lol monkey stirring up the drama in the pot with the brands

How so? It was like that when I got here.

 

[redeemer]

But as a reader of articles you have to kind of go through the fluff per say and understand what you can take away from it to sway one way or the other. I've seen the article for the 660Ti beating the 7950 however I know longevity wise the 7950 will probably hold better. For me I play a lot of nvidia games so I tend to like having a card that has a more full feature set for the games I play. The only feature that amd offers that I would desire is eyefinity so I feel I get a more complete product with nvidia at the moment. However if I wasn't playing games with physx I would probably be rocking a 7970.

How many Physx titles are you enjoying anyway? Yes there are review where 660ti is a head, but what happens when you overclock. You stated that you want to keep overclocking spererate but that is impossible because of Keplers autoboost feature. The only thing I use Nvidia for is 3D and cuda btw!

 

[bigshootr8]

How many Physx titles are you enjoying anyway? Yes there are review where 660ti is a head, but what happens when you overclock. You stated that you want to keep overclocking spererate but that is impossible because of Keplers autoboost feature. The only thing I use Nvidia for is 3D and cuda btw!

I play all the big ones >< so I'm a awful person to ask.

Batman Arkham Asylum
Batman Arkham City
Borderlands 2
Metro 2033

 

[Mousemonkey]

I'm actually a big fan of the work TR does and I wish that other sites would at least adopt their philosophy if not their precise methodology (I would hope other sites would offer their own flavor of frametime analysis and not merely repeat what TR is already doing). And obviously TR uses a pretty limited suite of games - I find it to be a balanced suite, but it's way too small.

More work needs to be done on conveying what some of their info, like 99th percentile frametime, means to end-user experience in specific terms, but I hope GPU reviewers are taking notice of what is going on over there... average FPS does NOT tell the whole story.

My real wish that folks like [H]ardOCP and others who comment on "smoothness" would start putting their money where their mouth is and actually make an attempt to measure smoothness. Every SLI or CF article [H] has written for at least a year makes some mention of how SLI "feels" better. However, IMO they might as well be saying the sky looks blue but is really purple, because such a conclusion often flat out contradicts their data.

I am currently running two SLi rigs, do I count?

 

[bigshootr8]

Yea there would need to be more of a closer eye on things to justify saying it was smooth and non jittery.

 

[Novuake]

I am currently running two SLi rigs, do I count?

I think you count. What is your experience? Have you had similar experience with CF?

 

[darksparten]

A very legit reason to prefer Nvidia.

PhysX isn't exactly a legit reason, exactly 27 games use PhysX and 4 of them are triple A so to speak(batman/borderlands, etc.).

 

[Novuake]

PhysX isn't exactly a legit reason, exactly 27 games use PhysX and 4 of them are triple A so to speak(batman/borderlands, etc.).

ANY rational reason to prefer Nvidia is a legit reason.

 

[redeemer]

I play all the big ones >< so I'm a awful person to ask.

Batman Arkham Asylum
Batman Arkham City
Borderlands 2
Metro 2033

Metro must decimate your card, Last Light hopefully has better coding