Guru3D 970 SLI review

[BrightCandle]

Did you guys check out the GURU3D review on the GTX 980 SLI. It has the same frame pacing problem as the 970 sli. However, as previously mention, the reviewer said it might be their FCAT hardware.

The weird thing about these dropped frames is that they are saying its not impacting the gameplay experience, despite the fact that it was obvious on 7970s. I don't know if the problem is runt frames, it would seem weird for Nvidia to release FCAT to the world and then fail its own test considering that is part of how it tests internally.

 

[Keysplayr]

Has nobody else done FCAT testing on Maxwell SLI?
PCPerspective?
TechReport?

 

[AdamK47]

With 4-way SLI and vsync enabled at 60hz everything plays buttery smooth for me in the dozens of games I've tried with my 980s.

 

[3DVagabond]

Did you guys check out the GURU3D review on the GTX 980 SLI. It has the same frame pacing problem as the 970 sli. However, as previously mention, the reviewer said it might be their FCAT hardware.

When it showed the issues with AMD we were told is was infallible.

The weird thing about these dropped frames is that they are saying its not impacting the gameplay experience, despite the fact that it was obvious on 7970s. I don't know if the problem is runt frames, it would seem weird for Nvidia to release FCAT to the world and then fail its own test considering that is part of how it tests internally.

@ 100+ FPS (or even 75 FPS) it wasn't noticeable with AMD either.

Has nobody else done FCAT testing on Maxwell SLI?
PCPerspective?
TechReport?

No. Not them or anyone else who tests with FCAT, or even FRAPS testing, has done SLI. Only Guru3D. You'd think these other sites would care?

 

[Keysplayr]

When it showed the issues with AMD we were told is was infallible.



@ 100+ FPS (or even 75 FPS) it wasn't noticeable with AMD either.



No. Not them or anyone else who tests with FCAT, or even FRAPS testing, has done SLI. Only Guru3D. You'd think these other sites would care?

I would think that FCAT would now be the defacto participant in ANY new GPU testing. Heck, not even just new but all. It is very weird. TechReport did do frametimes, but only on a single Maxwell. Are they following up with an SLI review? Or is it so very time consuming to conduct FCAT testing?
I dunno. Everyone seems to be saying that they cannot see any evidence of dropped frames or stuttering with Maxwell SLI, so maybe there is something wrong with the FCAT hardware that belongs to GURU3d. I did hear that FCAT was indeed a PITA to setup.

And eventually it would be great if Guru3D could verify if their equipment wasn't behaving correctly or it wasn't calibrated (If it even needs calibrating) correctly. Or even if they tried it again with the newer drivers just because they can.

 

[Madpacket]

Hmm, guess I'll hang on to my dual 290's for now.

 

[wand3r3r]

I would think that FCAT would now be the defacto participant in ANY new GPU testing. Heck, not even just new but all. It is very weird. TechReport did do frametimes, but only on a single Maxwell. Are they following up with an SLI review? Or is it so very time consuming to conduct FCAT testing?
I dunno. Everyone seems to be saying that they cannot see any evidence of dropped frames or stuttering with Maxwell SLI, so maybe there is something wrong with the FCAT hardware that belongs to GURU3d. I did hear that FCAT was indeed a PITA to setup.

And eventually it would be great if Guru3D could verify if their equipment wasn't behaving correctly or it wasn't calibrated (If it even needs calibrating) correctly. Or even if they tried it again with the newer drivers just because they can.

Maybe nvidia doesn't want them to test it? Reviewers guide rules or guidelines? Get a little time to tweak the drivers due to the rush to market since it doesn't appear like they have many cards yet anyway.

I don't recall any benefit of the doubt during the 7970 campaign. I'm inclined to wonder if the fps are high enough to cover it but it should be apparent in 4k.

I'm surprised amd doesn't counter the huge nv campaign, if the alleged runt frames are true.

 

[amenx]

Maybe nvidia doesn't want them to test it? Reviewers guide rules or guidelines?

Here we go again.. veering off into conspiracy theories. Did it not occur to you that the potential blowback from that could be just too much, especially when any reviewer could just spill the beans when given a condition like that?

 

[Keysplayr]

Maybe nvidia doesn't want them to test it? Reviewers guide rules or guidelines? Get a little time to tweak the drivers due to the rush to market since it doesn't appear like they have many cards yet anyway.

I don't recall any benefit of the doubt during the 7970 campaign. I'm inclined to wonder if the fps are high enough to cover it but it should be apparent in 4k.

I'm surprised amd doesn't counter the huge nv campaign, if the alleged runt frames are true.

Benefit of the doubt is being give perhaps because of reports of absolutely smooth gameplay and Guru3D thinks that there is a chance their FCAT hardware is borked somehow. I mean, why WOULDN'T you give the benefit of the doubt with those two pieces of information in your face?
No benefit of the doubt was given to the 7970 because it was blatant and easy to see on screen. No such reports for this Nvidia release.

 

[Abwx]

No benefit of the doubt was given to the 7970 because it was blatant and easy to see on screen. No such reports for this Nvidia release.

Curious that not a single reviewer brought the issue before Nvidia started their viral marketing campaign, so it was obviously not obvious, they (blindly!) assumed that it was obvious because they received some mesurement gear, that is, they relied on softwares to do the measurements that their eyes were incapable of since they saw nothing.

 

[el etro]

Remembering when AMD won against Nvidia in its own FCAT game(Single-GPU only at that time), and them doomed them with the XDMA tech.

Same game, but sides changed.

 

[SolMiester]

Curious that not a single reviewer brought the issue before Nvidia started their viral marketing campaign, so it was obviously not obvious, they (blindly!) assumed that it was obvious because they received some mesurement gear, that is, they relied on softwares to do the measurements that their eyes were incapable of since they saw nothing.

Umm, thats not true at all!, H and a couple of other site were mentioning the stutter before FCAT proved what was happening.

 

[x3sphere]

Umm, thats not true at all!, H and a couple of other site were mentioning the stutter before FCAT proved what was happening.

Yep. And AMD refused to acknowledge it until the FCAT measurements started coming out.

 

[Keysplayr]

Curious that not a single reviewer brought the issue before Nvidia started their viral marketing campaign, so it was obviously not obvious, they (blindly!) assumed that it was obvious because they received some mesurement gear, that is, they relied on softwares to do the measurements that their eyes were incapable of since they saw nothing.

Do you have any sort of problem with a competitor finding and exploiting a weakness of another competitor? Sounds like you think this is being swept under the rug somehow with your word choice of "Curious".

And like SolMeister mentions, H was reporting on this long before FCAT was around.
I could request Kyle to swing by if you need a refresher. Maybe he will consent.
It wasn't obvious until someone thought to check, or at least they did see it and thought nothing of it having no idea frames were being only partially drawn or dropped altogether. They probably chalked it up to simple screen tearing until somebody starting digging deep. Then is was uncovered, brought to public and exploited to the fullest capacity. Prob?

 

[iiiankiii]

Benefit of the doubt is being give perhaps because of reports of absolutely smooth gameplay and Guru3D thinks that there is a chance their FCAT hardware is borked somehow. I mean, why WOULDN'T you give the benefit of the doubt with those two pieces of information in your face?
No benefit of the doubt was given to the 7970 because it was blatant and easy to see on screen. No such reports for this Nvidia release.

Exactly. Just b/c one person is able to detect stutters doesn't mean the next person is sensitive to it. Yet, the data don't lie. The equipment voiding the results? Maybe. Who knows. I can tell you what, though. I doubted the finding from the 7970 frame pacing test b/c I could not tell the difference. Not even at all. Yet, the data was there. So what do you trust? The data, or your eyes. Same situation here. The reviewer is saying it's super smooth. Yet, the data suggested it isn't. Fortunately, I recognized that just because I was insensitive to it doesn't mean others aren't.

Until more tests are done, expect to see doubt regarding frame pacing performance with these gtx 9xx series. Sure, you can have a bunch of people saying it's smooth. Yet, you know there are others out there that are able to detect these micro-stutters. You can't truly trust another's person word on "smoothness". I guess you have to just look at the data and find out what your threshold is for micro-stutter.

Forget about what happened in the past. Frankly and ironically, thanks to Nvidia's FCAT method in an attempt to exploit their competitor's weakness, AMD was able to fix this problem. Competition is great.

 

[Keysplayr]

Exactly. Just b/c one person is able to detect stutters doesn't mean the next person is sensitive to it. Yet, the data don't lie. The equipment voiding the results? Maybe. Who knows. I can tell you what, though. I doubted the finding from the 7970 frame pacing test b/c I could not tell the difference. Not even at all. Yet, the data was there. So what do you trust? The data, or your eyes. Same situation here. The reviewer is saying it's super smooth. Yet, the data suggested it isn't. Fortunately, I recognized that just because I was insensitive to it doesn't mean others aren't.

Until more tests are done, expect to see doubt regarding frame pacing performance with these gtx 9xx series. Sure, you can have a bunch of people saying it's smooth. Yet, you know there are others out there that are able to detect these micro-stutters. You can't truly trust another's person word on "smoothness". I guess you have to just look at the data and find out what your threshold is for micro-stutter.

Forget about what happened in the past. Frankly and ironically, thanks to Nvidia's FCAT method in an attempt to exploit their competitor's weakness, AMD was able to fix this problem. Competition is great.

Actually, the 7970 problem was so bad it was possible for us to see it via videos taken side by side with the 680s. Suffices to say, if you couldn't see it, you were probably near blind. And long before FCAT, H kept mentioning that gameplay seemed to be smoother on Nvidia. Of course we found out later that they were right when FCAT microscopically broke everything down for us. Not because of the data. The data came after. What came first was noticing a difference in smoothness from one vendor to the other and then looking at it and finally saying, hey lets dig in and see what the heck is going on here.

 

[Vesku]

Actually, the 7970 problem was so bad it was possible for us to see it via videos taken side by side with the 680s. Suffices to say, if you couldn't see it, you were probably near blind. And long before FCAT, H kept mentioning that gameplay seemed to be smoother on Nvidia. Of course we found out later that they were right when FCAT microscopically broke everything down for us. Not because of the data. The data came after. What came first was noticing a difference in smoothness from one vendor to the other and then looking at it and finally saying, hey lets dig in and see what the heck is going on here.

This is false, last time I tried multi-GPU I was sensitive to its graphical hitches yet I've never had a "smoothness" problem with my single 7950 even before AMD addressed the issue. Has Guru3D or anyone else run a test with the latest driver to see if it is still dropping frames? Given Nvidia's obsession with such things when it was a jab against their competitor, Nvidia should be harangued about it until they fix it even if it can only be noticed via machine measurements.

 

[iiiankiii]

Actually, the 7970 problem was so bad it was possible for us to see it via videos taken side by side with the 680s. Suffices to say, if you couldn't see it, you were probably near blind. And long before FCAT, H kept mentioning that gameplay seemed to be smoother on Nvidia. Of course we found out later that they were right when FCAT microscopically broke everything down for us. Not because of the data. The data came after. What came first was noticing a difference in smoothness from one vendor to the other and then looking at it and finally saying, hey lets dig in and see what the heck is going on here.

.... Bro, let it go. You're accusing me of being near blind? Really? Why don't you just accept that some people aren't sensitive to such things. I'm not saying others aren't able to see it. There are plenty of people that weren't able to see it. In fact, my most of family and friends weren't able to detect those "micro-stutters". Just accept the fact that there are other people out there that aren't able to detect it without resorting to accusing people of being near blind. Seriously..

edit: I didn't look at any video posted online. So I don't know. I'm just going by games we play.

 

[BrightCandle]

When FCAT was released we basically determined that the stuttering in frame times from fraps was highly correlated with the FCAT results, basically if it was showing wild swings down into the low milliseconds there was a problem. The absolute swing of below about 6ms didn't cause most people any real concerns although many people noticed how much smoother things were with frame pacing even with this level. Just about everyone could spot 12ms swings amd above, so whether you saw it or not really did depended on the games you played and how bad the stutter was.

I complained about my 7970's from the first day crossfire was working (which was about 5 weeks after the cards were released incidentally!). It was immediately apparent to me, but then I am very sensitivity to stutter and frame rates. We had lots of reports from people there was an issue, plenty of people denying it and then FCAT came out and everyone agreed frame pacing was necessary. AMD spent a full year denying there was a problem, closing my bug reports (and presumably Kyle's and everyone else who was having the issue and reporting it) and then the better part of 18 months not fixing it completely. We still aren't "done" on the frame pacing side there are still elements of support missing which presumably will never be completed now.

So I hope its just a bug in Guru 3D's results, but if its really something Nvidia has introduced with these cards well they kind of deserve the talking down they get by the journalists and we need to hold them to fixing it and doing so quicky. I for one think its suspicious that none of the other sites have released SLI results, it does make me think maybe there is a problem and they are working with Nvidia to resolve before they publish. Its not uncommon for reviewers to report such bugs and get alpha fixed code to test. The lack of results either suggests very short review times given by Nvidia or a problem, and by now I am inclined to think its a genuine problem until we see techreport/pcper say otherwise (Does anandtech even do FCAT yet? I think they are still on average FPS not even minimums, the downfall of a tech giant).

 

[x3sphere]

So I hope its just a bug in Guru 3D's results, but if its really something Nvidia has introduced with these cards well they kind of deserve the talking down they get by the journalists and we need to hold them to fixing it and doing so quicky. I for one think its suspicious that none of the other sites have released SLI results, it does make me think maybe there is a problem and they are working with Nvidia to resolve before they publish. Its not uncommon for reviewers to report such bugs and get alpha fixed code to test. The lack of results either suggests very short review times given by Nvidia or a problem, and by now I am inclined to think its a genuine problem until we see techreport/pcper say otherwise (Does anandtech even do FCAT yet? I think they are still on average FPS not even minimums, the downfall of a tech giant).

There was a very short lead time on these cards. According to Kyle over at HardOCP reviewers got them mid-day Monday (15th) and the embargo was the 18th. So, basically only three full days to get a review up.

Anyway, it looks like PCPer does have SLI results up?

http://www.pcper.com/reviews/Graphi...70-GM204-Review-Power-and-Efficiency/Crysis-0

I went through the review, and their data seems to match Guru3D on BioShock Infinite. Some of the other games, like Crysis 3 and Metro Last Light (both games I've personally played on my 980s), show no issue.

So... it may just be an issue with a select number of games, not an indicator of a larger problem. Frame variance is lower on the NV cards in Metro and Skyrim, even at 4K. In the other games PCPer tested they scored worse than AMD or about as good.

 

[Enigmoid]

There was a very short lead time on these cards. According to Kyle over at HardOCP reviewers got them mid-day Monday (15th) and the embargo was the 18th. So, basically only three full days to get a review up.

Anyway, it looks like PCPer does have SLI results up?

http://www.pcper.com/reviews/Graphi...70-GM204-Review-Power-and-Efficiency/Crysis-0

I went through the review, and their data seems to match Guru3D on BioShock Infinite. Some of the other games, like Crysis 3 and Metro Last Light (both games I've personally played on my 980s), show no issue.

So... it may just be an issue with a select number of games, not an indicator of a larger problem. Frame variance is lower on the NV cards in Metro and Skyrim, even at 4K. In the other games PCPer tested they scored worse than AMD or about as good.

I agree. For example,

SLI frame smoothness is horrible. But so is it for a single card. Looks like a driver issue and not a SLI issue.

 

[Erenhardt]

Looks like a driver issue and not a SLI issue.

What's the difference? Anyone still believes in hardware frame pacing?

Refresh, nv released FCAT in end of march, frame pacing beta driver was out in august.

 

[3DVagabond]

Umm, thats not true at all!, H and a couple of other site were mentioning the stutter before FCAT proved what was happening.

And [H] has been reporting that Crossfire is now smoother than SLI, too.

 

[3DVagabond]

I agree. For example,

SLI frame smoothness is horrible. But so is it for a single card. Looks like a driver issue and not a SLI issue.

Bioshock seems to be another game that shows a lot of smoothness issues for both companies, although nVidia is generally faster.

 

[SirPauly]

So what do you trust?

One self! For years and years was vocal about the limitations of AFR and bothered me from an objective context to see single GPU numbers compared to AFR numbers based on they were not the same to me.

From 2008:

I wouldn't mind that option if it would offer superior smoothness because for me it's always the lows that are paramount and highs, can't tell the difference anyway.

Sure, it may not look sexy selling a product but it's not like you're forced to use an option like this; and can be a higher quality setting for superior smoothness attributes while keeping lower frame-rates nice and steady.

I suppose the idea is to sell product after all and understand that but still feel that offering more quality options for the gaming experience would help sell product as well. If a gamer desires or believes he needs more frame-rate -- still could have it.............but why force the ones that desire more consistency here with lows with just offering views into the wind so-to-speak?

Still appreciate AFR though and multi-GPU's - just like to see it improve and evolve and become more seamless. It's not like they don't know and no doubt have plans and ideas to improve this some-how.......just can't wait for these days to come to fruition

http://forum.beyond3d.com/showpost.php?p=1168949&postcount=160


I did bold out because the companies are finally discussing past limitations and seeing this evolve and improve and go beyond just performance numbers. Some gamers have been consistent about quality and improving the gaming experience.

On this nVidia example: More tests, investigations and evaluations are welcomed based on the awareness potentially creates improved experiences for their customers, imho!