Asking Ryan Smith of AT if a special examination could be done?

[wand3r3r]

Don't drag me into your FUD, thx

Ok I'll humor this.

What fud (of mine, personally) are we talking about here?

 

[SirPauly]

Some seem to think that one review in one game that is not in keeping with past reviews isn't unproven FUD.

AMD themselves respect the site enough to do internal investigations.

 

[NIGELG]

AMD themselves respect the site enough to do internal investigations.

You don't know if AMD respects the site or think they are full of garbage.The investigation by AMD would see if these results are repeatable or if it's garbage.

 

[SirPauly]

You don't know if AMD respects the site or think they are full of garbage.The investigation by AMD would see if these results are repeatable or if it's garbage.

Sure I do! TechReport's investigations has raised internal alarms at AMD and will investigate -- if they didn't respect Techreport -- they would probably do or say nothing.

 

[Schmide]

Ars on frame smoothness.

Edit: lol i glossed over the.

This story was brought to you by our friends at The Tech Report. You can visit the original story here.

 

[cmdrdredd]

This thread sums up why I never/rarely post on Anandtech. Pages and pages of AMD vs. NVIDIA flaming that never goes anywhere.

Ryan has already said that the testing will happen one day. Several people (including myself) have voiced support for it.

What other genuinely on topic posts have been made? And no, psychoanalysis of the OP and his intentions (overt or otherwise) are not on topic.

Seriously, on many other forums this kind of pointless crap would have been modded to death early on...that is, if the thread were not closed outright.

Don't you all have better things to be doing than trying to engage in futile religious debates? I ask this without a shred of sarcasm, truly.

What do you think I am? I have nothing better to do...you think I have a life? I have the internets. /sarcasm

I do agree. Some of what has been said boggles the mind.

Your problem is that it isn't just my subjective opinion, it is data from other sites as well. For example [H] did a performance review of AC3 and not once did they mention excessive unplayable stutter on the HD 79x0 cards. Also they get 20% extra performance out of a 7950 in the game compared to TR at similar settings. So stop with the BS we are all wrong, unless you think [H] and other review sites are also wrong and cannot be objective.

HardOCP uses a 4.8Ghz 2600k on a Z77 Motherboard. TechReport is on a 3820 on x790 with no overclock. The difference in clock speed is 1Ghz (3820 is 3.8Ghz turbo) so that can make a difference in framerate. Further HardOCP is using a 7970, TechReport was testing a 7950.

I agree, the other thread (while very blunt) actually provided some evidence to show NV adaptive vsync is a joke, and it died a quick death.

Then the "are we biased" died pretty quickly.

Here we have the same level or possibly even worse behavior but it just stays open.

I don't care about the individual issues, but this is a bit odd.

Did you read the vsync thread? The OP didn't understand it or deliberately tried to cause a stir. Then when it was explained he went on to make some surprisingly ignorant statements for nothing more than an argument. The whole point of THAT thread was someone was misunderstanding the point of adaptive vsync and when facts were presented as to what it does and why it actually does its job, hostilities commenced.

 

[cmdrdredd]

n/m

 

[96Firebird]

Did you read the vsync thread? The OP didn't understand it or deliberately tried to cause a stir. Then when it was explained he went on to make some surprisingly ignorant statements for nothing more than an argument. The whole point of THAT thread was someone was misunderstanding the point of adaptive vsync and when facts were presented as to what it does and why it actually does its job, hostilities commenced.

I believe it was locked because the OP was a previously banned member that the staff didn't want back. Good riddance, he didn't seem like the brightest crayon in the box.

 

[bononos]

......

Hard OCP's eyes: in the article he states on 3 separate occasions that the AMD cards are smoother then in the conclusion says Nvidia is smoother, Techreport's original article which defies all logic presented by their previous reviews that cover the 7950, a German video of far cry 3 that shows AMD actually being smoother than Nvidia, and Techreport's Skyrim example which to me only shows one hitch at 30 seconds and the rest is horrible screen tearing that is also apparent on the Nvidia side.
........

There is a newer TR article Dec11 (I think) which benchmarks newer and presumably more gpu intensive games which adds more weight to their contention that the Radeons have more jitter as compared to their Sept article which caused the initial discussion of this thread.

 

[Revolution 11]

AMD themselves respect the site enough to do internal investigations.

Respect has nothing to do with this. It is simple good business practice. If I was running a billion-dollar tech company (AMD is $1 billion?), I would investigate any flaws especially if they concern a product whose parent division is outperforming the rest of the company in terms of competitiveness. The GPU division at AMD is one area where the company remains competitive on a sustained basis.

As long as the flaw scenario seems credible, it is worthwhile investigating the problem. If it is real, better to get on it before more thorough checking is done by the rest of the community. If it is not, PR fiasco averted.

I don't think TR did a very stringent experiment or used the best methodology and scale but their data is plausible enough that AMD needs to check this out.

 

[pandemonium]

We read between the lines.

Ya, don't do that. There's no need to place assumptions into written text. It just gets everyone all fired up over conjecture, which is never a good thing.

 

[3DVagabond]

N/M. Someone beat me to it,

 

[SirPauly]

Respect has nothing to do with this. It is simple good business practice. If I was running a billion-dollar tech company (AMD is $1 billion?), I would investigate any flaws especially if they concern a product whose parent division is outperforming the rest of the company in terms of competitiveness. The GPU division at AMD is one area where the company remains competitive on a sustained basis.

As long as the flaw scenario seems credible, it is worthwhile investigating the problem. If it is real, better to get on it before more thorough checking is done by the rest of the community. If it is not, PR fiasco averted.

I don't think TR did a very stringent experiment or used the best methodology and scale but their data is plausible enough that AMD needs to check this out.

Imho,

Respect has everything to do with this. TechReport isn't some vocal forum poster.

 

[NIGELG]

Imho,

Respect has everything to do with this. TechReport isn't some vocal forum poster.

No....it's a business and anything that throws their products in a bad light merits investigation.It's not respect....you would like it to be.

It's just one or two sites' opinions.Techreport is just a review site out of many.

 

[SirPauly]

Sure thing -- AMD doesn't respect TechReport or Scott.

 

[NIGELG]

Sure thing -- AMD doesn't respect TechReport or Scott.

Did you ask them?...or do we just have your assumption that they 'respect' the site?

 

[Final8ty]

Did you ask them?...or do we just have your assumption that they 'respect' the site?

Let the market decide

 

[SirPauly]

Imho,

If they didn't respect the site -- they wouldn't bother. It's not like this is a vocal claim from an extremist making noise. Respect for Techreport may be in different forms; the editor and their work and the sheer size of their readership.

I'll take the word of Anand Shimpi when it comes to Scott:

https://twitter.com/anandshimpi/status/279440323208417282

TechReport has some surprising results based on his investigations. The awareness from this may improve AMD and nVidia products being smoother -- the bigger picture to me.

 

[NIGELG]

Imho,

If they didn't respect the site -- they wouldn't bother. It's not like this is a vocal claim from an extremist making noise. Respect for Techreport may be in different forms; the editor and their work and the sheer size of their readership.

I'll take the word of Anand Shimpi when it comes to Scott:

https://twitter.com/anandshimpi/status/279440323208417282

TechReport has some surprising results based on his investigations. The awareness from this may improve AMD and nVidia products being smoother -- the bigger picture to me.

In other words you assume and don't know for sure.

Any luck having Rage 3d investigate?

 

[RussianSensation]

2.You mean that both TR and [H] are dishonest and we should believe some random forum posters gotcha.

TR did test HD7970Ghz vs. GTX680 and 680 showed more stutter in previous testing. No one on the forums or online raised a storm about it though. Why is that? Because we know stutter happens depending on the game/game engine. In the same review HD7970 925mhz tied an overclocked 670:

Source

There are plenty of cases where NV cards stuttered more than AMD cards from the same website as well.

GTX570 vs. HD6970

or in Batman AC where GTX560Ti/570 lost badly to HD7850/7870 or even HD6870 in smoothness.

I don't recall anyone raising a storm regarding the stuttering of GTX500/600 even though it happened on many occasions in older TR reviews. You can easily find 10 games where NV will stutter more or vice versa. It could very well be the case that in those specific games GTX660Ti just stutters less. What would happen if we picked 10 entirely different games? What would happen if we tested HD7970Ghz vs. GTX680?

The results are not necessarily conclusive to start staying that "all AMD cards stutter" or "AMD had these issues for years but TR is the first site to expose this". They cannot be generalized across all games or across all AMD vs. NV GPUs because as you can see NV cards have also stuttered in games and not just GTX600 seriers. The results of HD7950 vs. GTX660Ti also contradict other website's testing that report HD7950 boost to be smoother than 660Ti. Are we supposed to take HardOCP's view over TR's, or vice versa? Their views for single GPU micro-stutter are contradictory for HD7950 boost vs. GTX660Ti. So which website is correct?

The Internet suddenly realized how TR does its testing but if people have been paying attention, just not long ago GTX680 lost in smoothness factor to the HD7970Ghz on the same site and on many occasions GTX570 exhibited more stuttering than HD6970 did.

Were where NV users when GTX560Ti/GTX570 stuttered in BF3 despite performing faster than HD6870/6970, respectively?

Why wasn't Scott this critical when NV cards stuttered in those other reviews? It just happened that AMD's 7950 stuttered more in that specific selection of games under AMD's latest beta drivers. Everyone is up in arms but stuttering across NV and AMD depending on the title has been a general occurrence at TR based on many of their other reviews. That's why it's best to buy your cards based on the games you play, not only based on average benchmarks if say you mainly play BF3 or Skyrim. Hopefully AMD will address any driver issues they are having but the notion that AMD's single GPU cards suddenly stutter more ignores all those other reviews and other games where NV cards stuttered more. We can't just ignore those other TR reviews and games but only focus entirely on 1 review, not to mention ignoring conflicting conclusions by other websites that show HD7950 boost delivering smoother/faster gaming experience than 660Ti.

TR's results also don't explain why the FPS in Skyrim and MOH:W don't agree with just about any other website online that did recent testing in those games at 2560x1440/1600 with AA.

Perhaps if 2-3 more other websites started tracking micro-stutter with a high-speed camera and tested other games, and other GPUs like HD7970Ghz vs. GTX680, we would get a much more well-rounded picture.

 

[RussianSensation]

It already IS a big deal if it proves that TR and H are accurate..

As I said before, ppl seem to be confused and mixing and matching SLI/CF micro-stutter with single-GPU micro-stutter. You can't just do that, and you did.

If you read HardOCP's conclusions for single GPUs, it's completely opposite of TR's. Once again in case you missed it:

"This pricing band was far less competitive as the Radeon HD 7950 with Boost simply demolished the GTX 660 Ti across the board with regards to raw frame rates and overall game play experience across our suite of testing."
http://www.hardocp.com/article/2012/11/12/fall_2012_gpu_driver_comparison_roundup/8

HardOCP's latest GTX680 SLI vs. HD7970Ghz CF made no mention whatsoever that a single GTX680 felt smoother or played faster than a single HD7970Ghz. They said the opposite:

"We even found, that on the whole, single-card HD 7970 GHz Edition video card provided better performance in triple-display gaming. The GeForce GTX 680 struggled a bit in some games at those high 5040x1050 resolutions."
http://www.hardocp.com/article/2012/12/04/gtx_680_vs_radeon_hd_7970_multidisplay_showdown/7

A videocard can't provide a better gaming experience when "it struggles." That's just plain English.

You keep mixing and matching TR & HardOCP to suit your viewpoint, instead of objectively looking at these 2 reviews separately:

1) HardOCP only concluded that SLI >>> CFX in smoothness, which hardly anyone on our forum disputes. They made no such assertions for GTX680 vs. HD7970Ghz or HD7950 Boost vs. GTX660Ti, where they time after time gave the advantage to the Radeons. Perhaps you forgot this review as well by them?

2) TR concluded that GTX660Ti felt smoother than HD7950 but it tells us nothing about HD7970Ghz vs. GTX680 or how GTX660Ti vs. HD7950 would fare in other games.

What you did instead is clump both of those reviews under 1, without reading the actual details of what they discuss. TR's own testing of HD7970Ghz showed it beating GTX680, or it is opposite of HD7950 vs. GTX660Ti review. That also escaped you? You seem to be very eager to want to jump to conclusions that AMD cards are stutter infested, which I suppose is understandable from an NV Focus Group member who ignored all those other TR reviews where NV cards stuttered more.

 

[Jaydip]

@ RS

read my earlier posts.I never said myself(680) or my brother(7950 DC II) encountered any ms.But none of us have very good eyesight honestly so it may be down to that.MS is a real issue for some and that is why we should find an objective method to identify that.Just because many of us don't encounter it doesn't invalidate its existence.My only ms issue so far has been with 4870X2.

 

[Keysplayr]

In other words you assume and don't know for sure.

Any luck having Rage 3d investigate?

Probably won't have much luck getting Rage3d to participate. But you never know.

 

[NIGELG]

Probably won't have much luck getting Rage3d to participate. But you never know.

Caveman Jim is a very reasonable guy.IF there's something worth investigating he just might.

 

[ICDP]

Hitman Absolution frametimes. Same level and settings as the TR article. I even matched the path they took through the level.

Intel i7 2600K @ 4.7 GHz
16 GB DDR 1886 RAM
HD 7970 VTX3D X-Edition underclocked to 900/1250 to match 7950 in TR article (950/1250).
Windows 7 64 bit Pro
12.11 b11 with Cap 2
Res: 1920x1080
RadeonPro not running at all.
All in game settings the same as in TR article. CCC settings = vsync off and all others at default.

I have to say that my results differ massively as those on TR. No excessive stutters at all.