[TECH Report] As the second turns: the web digests our game testing methods

[blackened23]

I don't know what you just said.

Some type of subliminal message urging you and everyone to buy more nvidia stuff. 2 more 690s and a tegra 3 tablet should do it.

 

[AnandThenMan]

I think the people with the obvious bias towards either brand choose to ignore this in order to advance their agenda.

A certain member here outright admitted they are 100% Nvidia biased, so why even bother taking anything they post seriously? This entire thing is absolutely ridiculous, but I have to hand it to Nvidia, they are a marketing force. The minions eat it right up, my favorite is the guy who says he sees stuttering going from a GTX580 to a 7970.

 

[AdamK47]

I'd also like to know what Railven said about me that was so spot on. Searched and couldn't find anything.

 

[SirPauly]

Some type of subliminal message urging you and everyone to buy more nvidia stuff..

Sarcastic conspiracy theory:


Indeed! Think the unsmooth frames that do happen with nVidia products are subliminal messages, specifically with AFR and controlled to more levels by their frame metering and called this micro-conditioning years ago! It isn't their hardware, software prowess or their pro-active nature that creates the biased, loyalist nVidia mind-set.

 

[blackened23]

I was just joking, I don't disagree about nvidia's hardware and software prowess. I was actually impressed by how much the latest drivers improved Far Cry 3 in SLI - game plays great for me now.

 

[SirPauly]

I was just joking,

So was I!

 

[railven]

I'd also like to know what Railven said about me that was so spot on. Searched and couldn't find anything.

I think it was a miscommunication. I don't even know who you are! And I don't mean that rudely, haha.

 

[Homeles]

A problem where a malicious attack can destroy your system, or a problem that you most likely can't even detect. You need to realize I'm not attacking nvidia or defending AMD. I'm just drawing attention to the absurd amounts of effort certain members of this forum devote to downplaying nvidia problems and blowing AMD issues out of proportion.

Now they will reply by saying that they can detect the problem and that's why they only buy nvidia cards. They will even say that this had been a problem for years. All while people who actually use AMD GPUs have been happily playing games with no problems.

AMD has had security issues too, such as their lack of ASLR support in EMET earlier this year.

http://www.anandtech.com/show/5966/uscert-takes-amd-to-task-on-driver-security

Security vulnerability issues happen from time to time for every single software and hardware company in existence. That is just how computers work — I'm glad that you've given me the opportunity to enlighten you.

Citing one issue by its lonesome and implying that the company with that is less competent than another is completely inane. Like the people pointing out Nvidia shimmering issues that happened a good 6 years ago. What else has happened since then? If you are incapable of producing a series of failures over a period of time, your argument is not worth the pathetic amount of storage space it takes up on these servers.

 

[Keysplayr]

I don't know what you just said.

Dude, it was from my smartphone and it turned stupid. Scouts honor. LOL I'll fix it.

 

[Keysplayr]

Some type of subliminal message urging you and everyone to buy more nvidia stuff. 2 more 690s and a tegra 3 tablet should do it.

Yes. Lets make it more about Keysplayr and less about the subject. That'll work.

 

[Keysplayr]

A certain member here outright admitted they are 100% Nvidia biased, so why even bother taking anything they post seriously? This entire thing is absolutely ridiculous, but I have to hand it to Nvidia, they are a marketing force. The minions eat it right up, my favorite is the guy who says he sees stuttering going from a GTX580 to a 7970.

Which member? And show us the quote.

 

[Final8ty]

Yep. This happens in all fields. That's why the negative results from blind testing are interesting to me. Sure, something's going on with the measurement but as far as anyone can show, what's being measured is tough to perceive without someone telling you beforehand what you're supposed to see. This is perfectly consistent with all of the benchmarks out there, too - as others have pointed out, the measured problems with nV 5xx cards were never seen before in actual game play testing.

And yeah, I know we'll be accused of being blind. It's the same approach audio marketing takes when they want to sell $15,000 power cables (http://bmicable.com/oceanic_statement) and $200 magic stones (special sale! - http://www.musicdirect.com/p-7506-shakti-electromagnetic-stabilizer-stone.aspx) which make music sound better. Sure, you'll hear a difference with these just so long as you believe the hype. When you don't know if they're installed or not, suddenly the difference goes away. There's a reason people spend so much money (and give away so much "free" hardware) for marketing.

LOL! great post and personally i only have the FPS counter on for a little while then turn it off and it only goes back on in any particular game if i see and think i have an issue.

 

[Rikard]

One thing that keeps bothering me is that despite having the same card as TR tested in their review I cannot reproduce their results, and I tried now for almost a month (OK I did other stuff too...)

I believe I have identified where on the map they are, and have a save with similar environmental conditions etc so I should be able to get their value of 31 ms in the 1% extreme. Instead I get 23 ms, corresponding to an instantaneous FPS of 43. When I am indoors I get about 19 ms in the 99th percentile, which is similar to their Whiterun scores, so the real question is why do they have so bad results outdoors which I cannot reproduce? Then, if I use Radeon Pro to fix the supposed microstuttering I get 19 ms also outside (which corresponds to 52 FPS instantaneous). I said before that I do not appear to be very sensitive to microstuttering, but is anybody really if you have one frame out of hundred dropping from 60 FPS to 43, or 52 for that matter?

Contrary to some people (mostly on other sites) I am not saying that TR are deliberately sabotaging their tests, but they seem to have a problem specific to themselves, and since I cannot reproduce their poor performance I still think we need more reviews on this if we should draw any conclusions from it. There could be something else in their setup that makes their results worse than what I can obtain. For example, I wanted to record a video like they did, but even when saving it on a separate drive the video recording itself caused stuttering.

And like I said in my previous post, I noticed some stutter in a few pathological cases that FRAPS are not picking up, so I think we need to go beyond what TR was doing. I find the PCPER method interesting for that purpose.

 

[f1sherman]

"cannot reproduce their results"...

you have same test scene/benchmark run like TR?

WTH

And like I said in my previous post, I noticed some stutter in a few pathological cases that FRAPS are not picking up,

No one responded to you because you dont seem to understand:

FRAPS fps curve is not supposed to pick up anything latency problematic. It's fraps frame-times(from log) that pick up AMD's issue.

EDIT: Fraps FPS curve with much longer polling time than typicaly frame time that is. If you are saying you have stuttering that even frame-times are not picking up that would be interesting.

 

[Rikard]

"cannot reproduce their results"...

you have same test scene/benchmark run like TR?

WTH

Like I said, I think I reproduce their conditions quite accurately, so yes I do.

No one responded to you because you dont seem to understand:

FRAPS fps curve is not supposed to pick up anything latency problematic. It's fraps frame-times(from log) that pick up AMD's issue.

I am talking of frame times. Please read before you post next time.

 

[f1sherman]

Like I said, I think I reproduce their conditions quite accurately, so yes I do.

LOL

I am talking of frame times. Please read before you post next time.

Fair enough.

I don't believe that you have stuttering with smooth frame-time distribution.

 

[Hitman928]

Fair enough.

I don't believe that you have stuttering with smooth frame-time distribution.

I personally have experienced the same and have posted videos/graphs showing it.

 

[f1sherman]

I personally have experienced the same and have posted videos/graphs showing it.

Stuttering with smooth, horizontal for most parts - delta function like behavior of frame time distribution? Similar to 660 Ti above... and yet stuttering?

I'd love to see that.

 

[Rikard]

Stuttering with smooth, horizontal for most parts - delta function like behavior of frame time distribution? Similar to 660 Ti above... and yet stuttering?

I'd love to see that.

I think Hitman will give you the link now any second for you. Since that stuttering is quite severe and still giving constant frame times in FRAPS I suspect it has a very different cause, but that does not make it any less important.

 

[f1sherman]

That would only be possible with frame-skipping.
CPU bottleneck perhaps.
And 60% CPU load, even across single core, does not mean it's not CPU bottleneck.

That's why PCPER idea is a huge step forward. Nothing goes undetected.

 

[Hitman928]

https://docs.google.com/open?id=0B11hajGAJWuVdXYwV3p2WVlBQmM

Download video and watch at it's native res (i.e. don't force it to full screen) for best viewing.

 

[f1sherman]

Hitman, can you post frame-times while you're at it?

BTW that wouldnt be Radeon Pro, would it?

I don't recall Afterburner's limiter producing such a flat curve.

And I did watch: Definitly frame-skipping

 

[Hitman928]

Hitman, can you post frame-times while you're at it?

BTW that wouldnt be Radeon Pro, would it?

I don't recall Afterburner's limiter producing such a flat curve.

And I did watch: Definitly frame-skipping

Those are frame times, unless you mean the raw data . . .

Yes, it's radeon pro.

I've never heard/seen anyone claim that using a frame limiter caused frame skipping. . .

 

[f1sherman]

Nvm raw data then...

Just use Afterburner if you really need limiter.

It seems that hard fps averaging, i.e. strict wait for frame draw that Radon Pro does has its deficiencies.
And thx for being honest

 

[Hitman928]

https://docs.google.com/spreadsheet/ccc?key=0Al1hajGAJWuVdDQydm00OUNwa2pCZVlmRTRmUGJJS1E

raw data ^

I'll give Afterburner a try later on tonight if I have time.