nvidia driver frame latency issues

[BallaTheFeared]

You guys are acting pretty silly,


http://alienbabeltech.com/main/?p=33257&all=1





It's not what I wanted, but I understand frametime testing is more work than normal simple fraps. I do not like the canned benches, but I can appreciate the effort.

 

[SirPauly]

To be fair more investigation is needed, just like for amd.

Indeed! With more investigations, reviews, articles and discussions that go beyond frame-rate is a service to the gamer, community, companies and their product offerings. I believe nVidia and AMD take seriously about their products and their experiences, desire to improve them for their current and prospective customers.

 

[thilanliyan]

Note:I use Intel CPUs only...as they are far better than anything AMD produces.

Shhhhhhh!!! My FX might hear you!! It still doesn't know that FX doesn't mean what it used to. Lol

 

[Rikard]

You guys are acting pretty silly,


http://alienbabeltech.com/main/?p=33257&all=1





It's not what I wanted, but I understand frametime testing is more work than normal simple fraps. I do not like the canned benches, but I can appreciate the effort.

Thanks, that study agrees well with my finding that I only see this problem on Hitman:Absolution. (That was on 12.11b11, could have been improved by now.)

 

[SirPauly]

I wouldn't consider it bashing really, more of a raising awareness sort of thing.

Imho,

Personally don't have any problems for pro or con awareness. The problem with forums at times is simply the personal attacks and accusations of agenda mongering or one-sided mongering instead of the points to engage for discussion.

Actually like your posts and that you don't personally attack!

 

[VulgarDisplay]

You guys are acting pretty silly,


http://alienbabeltech.com/main/?p=33257&all=1





It's not what I wanted, but I understand frametime testing is more work than normal simple fraps. I do not like the canned benches, but I can appreciate the effort.

I was reading that article thinking it was amazing that ABT was actually producing something unbiased and worthwhile. Then I got to the conclusion and my mind was blown.

From just looking at eleven benchmarks including nine games, we see that the HD 7970 at GHz edition clocks has more difficulty providing even frame rates. The GTX 680 does a better job of delivering a smoother experience in our benchmarks despite the HD 7970 generally having higher fps (framerates).

This statement in the conclusion made no sense after reading the article so I went back and checked the benchmarks. All told the Radeon was as good or better in 6 of the tests while providing faster FPS, 2 of their runs were awful on the 7970 and the gtx680, and 3 of the tests showed the the gtx680 being better and one was just a synthetic benchmark.

What a load of garbage. Their review basically states that the 7970 is as good or better in most of the benchmarks, and then they say that the 7970 has difficulty providing a good experience. This fully confirms to me that ABT is a steaming pile and will not get any more clicks from me.

 

[BallaTheFeared]

I can't pretend to speak for the author, but perhaps eliminating everything but the results might make it more clear how he came to that conclusion.

Although there is some variation, the runs and gameplay on this older DX1o game are nearly identical. The built-in benchmark is also generally more demanding on PC resources than the game itself.

The HD 7970 at GHz speeds is slightly faster, however the mimimums and average are higher; the amount of spikes are similar and neither runs are particularly smooth. Let’s move on to Far Cry 2.

Neither of them are particularly smooth, however, the Radeon has higher spikes more often and you can often see the jitter.

There are more spikes with the GTX 680 although the apparent framerate appears to be nearly the same. The reason jitter is not so apparent in this benchmark is because most of the spikes occur below 16.7ms.

Both of these video cards run with noticeable jitter in sections of this benchmark. Overclocking to GHz edition speeds does not change much with the HD 7970.

There is really little difference although the GeForce has a single spike that is 4ms higher than the Radeon’s. Overclocking the HD 7970 to GHz edition clocks brings some benefits besides increasing the framerate.

Neither of these video cards suffer from much jitter in this benchmark.

As evidenced by the frame times charts, there is a lot of difference in smoothness experienced by the gamer between playing Hitman: Absolution on a GTX 680 versus playing it on a HD 7970. The benchmark may be more demanding than the game, but the jitter is still evident on the Radeon. Even though the HD 7970 is faster in raw framerates, the frame delivery is uneven and it detracts from gameplay. We will continue to monitor AMD’s drivers in this particular case.

There is quite a bit of difference in the evenness of frame rate delivery between the two cards. We will will note Crysis 2 for later to see if AMD makes any changes in their drivers.

You'd have to ask the author why he felt what he felt in his conclusion.




For me personally the use of canned benches (many of which have their own profile) makes the results less meaningful. However I understand it's no easy task to do these reviews, and thus far despite the canned nature it has given us the most comprehensive list of games yet afaik. I hope he continues to deliver these reviews, with the greater hope that he moves to at least a few non canned runs in some of the more popular titles.

 

[BrightCandle]

This statement in the conclusion made no sense after reading the article so I went back and checked the benchmarks. All told the Radeon was as good or better in 6 of the tests while providing faster FPS, 2 of their runs were awful on the 7970 and the gtx680, and 3 of the tests showed the the gtx680 being better and one was just a synthetic benchmark.

In terms of smoothness and stuttering I actually make it 6 wins to the GTX 680 over hthe 7970. Its not about frame rates its about the 99% and 99.9% points as well as if there is any microstutter present and how severe it is. But 2 of those looses for the 7970 are unplayable messes, the frame times graphs are awful. Yet that never happens on the 680, it has 1 that hits noticeable levels but never does it get as severe as the 680.

So I don't know how it is you got to your conclusion, but I agree with ABT on the basis of the data presented, the 680 is superior in smoothness and when it looses it looses on FPS and not on smoothness.

 

[VulgarDisplay]

In terms of smoothness and stuttering I actually make it 6 wins to the GTX 680 over hthe 7970. Its not about frame rates its about the 99% and 99.9% points as well as if there is any microstutter present and how severe it is. But 2 of those looses for the 7970 are unplayable messes, the frame times graphs are awful. Yet that never happens on the 680, it has 1 that hits noticeable levels but never does it get as severe as the 680.

So I don't know how it is you got to your conclusion, but I agree with ABT on the basis of the data presented, the 680 is superior in smoothness and when it looses it looses on FPS and not on smoothness.

I drew my conclusions based on what the author said about the results after each benchmark run.