AMD to Tweak GCN Latency Issues with Driver Updates

[Gregster]

You do realise they picked random games?

3 were gaming evolved titles
2 were twimtbp
2 were indi

Well, you got one fact right, and that's that we were attacked, from reddit to YouTube to a host of tech forums across the web, a strangely cohesive group of posters said many of the same things about us. They weren't true, but you repeated them enough they created an impression, which I suppose was the point.

What's amazing is how every single one of the stated "reasons" you attacked us is false.

Here is the article in question:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited

*You claim we tested "over half" TWIMTBP games. The games we tested are:

-Borderlands 2 (TWIMTBP)
-Guild Wars 2
-Sleeping Dogs (Gaming Evolved)
-Assassin's Creed III (TWIMTBP)
-Hitman: Absolution (Gaming Evolved)
-Medal of Honor: Warfighter (Gaming Evolved)
-Skyrim

Gaming Evolved games: 3
TWIMTBP games: 2
Non-affiliated games: 2

Setting aside the fact that we prefer to test the most notable games regardless of whose label is on them, your claim is false.

*You claim the GTX 660 card we tested is rare and only for the Asian market. Yet it is a US-market card that remains to this very moment in stock at Newegg right here:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814500269

This card cost $20 *less* than the Sapphire 7950 we tested, and *both* cards tested were the highest-clocked variants of their type available at Newegg at the time we published.

*You claim we don't show the settings at which we tested, although we listed them exhaustively with screenshots of the in-game menus:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/3

*You claim we didn't show the in-game locations where we tested and asked for raw data, including video. Yet we *embedded video in the very review* showing our testing session for each game. Again, see here:

http://techreport.com/review/23981/radeon-hd-7950-vs-geforce-gtx-660-ti-revisited/3

We then posted video recorded at 120 and 240 FPS showing our Skyrim test session side-by-side on the GeForce and Radeon to illustrate the differences. See here:

http://techreport.com/review/24051/geforce-versus-radeon-captured-on-high-speed-video

*You repeated the trope that we saw "vastly different" results from other sites, yet none of them test like we do. Whatever you were expecting, time to re-think.

*You ask when PC gamers started caring about "slight" latency discrepancies. The answer, of course, is at the beginning of things, back when 3D gaming began. We've fought the battle for fluidity for years. Only in the past year have we used the language of frame latencies to describe it, but we've always wanted fluid gameplay. You are welcome to pretend PC gamers don't care about in-the-moment performance, just FPS averages, but you would once again be wrong on the facts.

Now that we've established you were factually wrong on every "reason" why TR was "attacked," one question remains: why *was* TR attacked?

Kinda makes you go: hmm.

A reply from Damage (of TR) to some of the people who can't see anything but TR attacking AMD.

 

[humbug]

You two are like an old married couple. I know you only stay together for the kids benefit.

oh.... no thanks, i don't want his babies, what a horrible thought.

What would the world be like with more little humbug and gregsters running around...

peace gregster

 

[Rusty0611]

The question is, can you see the difference?

Multi-card I could... just. AMD was visibly less smooth. But it wasn't enough to really distract me for more than a couple of seconds when I first fired BF3 up and then I didn't notice it.

Single card there's no perceivable difference for me; irrespective of what a graph illustrates to you it's only what you can actually perceive which is important.

 

[tommybhoy]

The question is, can you see the difference?

This.

The video Gregg posted although shows what's up, is shot at essentially 4x@60Hz or 2x120Hz(that's the way I see it anyway) and slowed way down on a barren landscape, what happens when both AMD/Nvidia get into a stramash ingame with a few ncp's battling it out?

Do they both still stay the same or would there be no dramatization and less PR(either way)@TR?


Happy New Year Rusty and to Humbug too if I missed you out-I'm loosing track now that auld age is taking over.

 

[humbug]

Happy new year tommy

 

[bru]

I suppose the bottom line is that whether we as individuals have noticed this issue or not AMD have acknowledged it, and they are going to take steps to correct/minimise it, which has to be a good thing in the long run.

 

[mmj_uk]

Over 12 months to admit/realise that the issue exists is bad even by AMD's standards.

It's good to see reviewers focussing on other areas of gaming experience rather than simply the raw frame rate.

 

[Rroff]

The question is, can you see the difference?

Multi-card I could... just. AMD was visibly less smooth. But it wasn't enough to really distract me for more than a couple of seconds when I first fired BF3 up and then I didn't notice it.

Single card there's no perceivable difference for me; irrespective of what a graph illustrates to you it's only what you can actually perceive which is important.

Yeah the underlying effect is much pronounced with multi GPU but I've rarely been able to notice the difference from single cards. (and I'm very sensitive to it and been running frametime tests long before there was mainstream awareness of frametimes).

When I saw the frametime post Final8y posted couple of days back or so I was pretty sure they either got the results switched around or some other user error - just didn't match with anything I've ever seen or the result any review site that does frametime testing has shown and I guess this now proves it.

Good to see AMD apparently taking the issue seriously and putting effort into fixing it, its been one of the biggest issues for me with AMD that they've traditionally been slow to respond let alone fix issues like this.

Over 12 months to admit/realise that the issue exists is bad even by AMD's standards.

It's good to see reviewers focussing on other areas of gaming experience rather than simply the raw frame rate.

Given the new architecture and the compleixties that can come with dealing with frametimes I'm inclinded to be lenient on that aspect - aslong as they actually release fixes in a half decent time interval rather than just make a noise about it and hope it all goes away.

 

[humbug]

Over 12 months to admit/realise that the issue exists is bad even by AMD's standards.

It's good to see reviewers focussing on other areas of gaming experience rather than simply the raw frame rate.

The Never settle bundle and its Drivers are not "over 12 Months" old, maybe 6 weeks old.

They are (on-going development) BETA drivers, they will just add this to the development list and fix it along with whatever other work is still to be done with those drivers before they are finished.
They may have already been aware of it before techreport.

 

[Sin_Chase]

The question is, can you see the difference?

Single card there's no perceivable difference for me; irrespective of what a graph illustrates to you it's only what you can actually perceive which is important.

I entirely dis-agree.

I loathe the "That will do" mentality. Whether or not you notice is not the point, if you do not have the mindset of 'Anything less than perfect is not good enough' then it shows elsewhere in your product also.

 

[dave_beast]

This applies to both Red and Green, you can plow through a bunch of games and find latency issues in some games on the Red side and with other games on the Green side.

+1 this.

The games I play, Arma2, BF3 etc do not suffer from this latency. You could say in those games that it does effect AMD cards, that at least colours are more vibrant on them as opposed to Nvidia. It's all swings and roundabouts and fanboism veiled under an "article" Some people assume that because Nvidia gpu's cost more that they're superior

 

[Rusty0611]

I entirely dis-agree.

I loathe the "That will do" mentality. Whether or not you notice is not the point, if you do not have the mindset of 'Anything less than perfect is not good enough' then it shows elsewhere in your product also.

I think you're missing the point - it's not a case of "that will do"; it's a case of "I don't notice any difference" and if it doesn't impact the experience then not bothered. If you can't notice the difference then why in your eyes, is it not already perfect? Because a graph tells you it's not perfect, even though you're unable to perceive any difference, you should take up the pitchforks to demand improvement? There's no harm in asking of course but if you can't perceive (that word again ) any difference then I would rather they work on other things.

That said, multi GPU does need improving as you can tell the difference. I'm sensitive - although not overly so - to all kinds of things like this: latency; response times; low FPS; tearing etc; and although it's not massive there is definite room for improvement.

So I think it is the point really. If you can't notice it why even bother caring? If you can then fair enough it is affecting you and therefore should be resolved.

 

[Sin_Chase]

I think you're missing the point - it's not a case of "that will do"; it's a case of "I don't notice any difference" and if it doesn't impact the experience then not bothered. If you can't notice the difference then why in your eyes, is it not already perfect? Because a graph tells you it's not perfect, even though you're unable to perceive any difference, you should take up the pitchforks to demand improvement? There's no harm in asking of course but if you can't perceive (that word again ) any difference then I would rather they work on other things.

That said, multi GPU does need improving as you can tell the difference. I'm sensitive - although not overly so - to all kinds of things like this: latency; response times; low FPS; tearing etc; and although it's not massive there is definite room for improvement.

So I think it is the point really. If you can't notice it why even bother caring? If you can then fair enough it is affecting you and therefore should be resolved.

All you did was re-iterate the post I quoted entirely neglecting what I said in mine

A lot of people do not get the idea of having pride in your work/craft and are happy with "That will do" because "It does not matter/Will not be noticed". It does matter, noticed or not.

In-fact there is a really really good Windows 8 UI video which re-iterates the same mentality with regards to UI design. Pretty much the same mind set "Nobody else will notice, but it has to be perfect"

 

[Rusty0611]

All you did was re-iterate the post I quoted entirely neglecting what I said in mine

A lot of people do not get the idea of having pride in your work/craft and are happy with "That will do" because "It does not matter/Will not be noticed". It does matter, noticed or not.

In-fact there is a really really good Windows 8 UI video which re-iterates the same mentality with regards to UI design. Pretty much the same mind set "Nobody else will notice, but it has to be perfect"

Not really.

My point is: what difference does it make to you personally other than a perusal of graphs for comment if you're unable to perceive the difference? Why does it matter to you if it's not "perfect"? I get your point on that the mentality of "that'll do" isn't good enough but there's no real evidence on either side of that kind of mentality creeping in to affect something actually perceivable.

It's a good point but too loose a generalisation to apply in reality.