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

Discussion in 'Video Cards and Graphics' started by Keysplayr, Dec 5, 2012.

  1. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    Yes, I do. Erm, that's why I made this statement.

    "Even [H] stated that stutter is only an issue on CF systems".
     
  2. Rvenger

    Rvenger Elite Member <br> Super Moderator <br> Video Cards
    Super Moderator

    Joined:
    Apr 6, 2004
    Messages:
    6,293
    Likes Received:
    4


    I am going remind everyone who is in doubt of this quote. I have owned both cards and on single GPU configurations, I do not notice a single difference. Both Nvidia and AMD provided me with a great gaming experience. While all the Nvidia Fanboys can make a huge case of this, I will enjoy gaming on my 7970.
     
  3. ZeroRift

    ZeroRift Member

    Joined:
    Apr 13, 2005
    Messages:
    177
    Likes Received:
    0
    If one company is making a measurable, positive contribution to the consumer's exerpience, they should most definitely be "thrown a bone".

    That aside, this is not about giving credit to one company or another (though that may be part of the outcome of the investigation). It is about knowing what is best for you: the consumer. More information is always valuable in this regard.
     
  4. Avalon

    Avalon Diamond Member

    Joined:
    Jul 16, 2001
    Messages:
    7,406
    Likes Received:
    10
    I think it's a perfectly fine topic to investigate. It could be a problem, or it could not be. If it is, I'd like some light shone on it so that it can be improved for others.

    A little bit of curiosity never hurt. Except when it killed the cat :O
     
  5. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    Nvidia have been a better choice for Multi GPU for a few years now. This is a well known fact and the new Kepler series has frame metering to reduce stutter even further.

    The OP is calling for an investigation based on the TR review done using a Windows 8 system in single GPU configuration. When it comes to single GPU there is no discernible difference between AMD or Nvidia when it comes to stuttering. So any investigation is most likely not going to find anything different to what we already know. If there were issues with stuttering in single AMD GPU configurations it would have been noticed and reported by now.

    My take on the TR review is that the AMD Windows 8 drivers are not mature yet. Or are we to totally discount every other review TR did before this one, where frame latency on the 7950 was never an issue?
     
    #180 ICDP, Dec 7, 2012
    Last edited: Dec 7, 2012
  6. Rvenger

    Rvenger Elite Member <br> Super Moderator <br> Video Cards
    Super Moderator

    Joined:
    Apr 6, 2004
    Messages:
    6,293
    Likes Received:
    4

    I think a lot of the stuttering on the 7950 is due to it's low clock speed. I noticed Heaven was very choppy until I overclocked it to around 950-1000mhz.
     
  7. ZeroRift

    ZeroRift Member

    Joined:
    Apr 13, 2005
    Messages:
    177
    Likes Received:
    0
    The bold part is unproven. Objective evidence to support or condemn conflicting subjective results is all that is being requested.

    Besides, tinkering with hardware and software is what we enthusiasts are so enthusiastic about, right? :D
     
  8. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    Sorry about that, I re-read my post and realised I was making a bold statement so edited my post to state "most likely not going to".

    Sorry again :)
     
  9. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    Good point, but the previous TR reviews didn't show any issues with the stock 7950 in this regard. If it was the core of the issue all of their reviews would show the same symptoms.
     
  10. Keysplayr

    Keysplayr Elite Member

    Joined:
    Jan 16, 2003
    Messages:
    21,200
    Likes Received:
    48
    I think as more and more review sites catch on, they'll want to do an investigation of their own.
    "most likely not going to" is also just opinion and unproven. As you can see, MANY members here would absolutely love to see this examined by more review sites.
     
  11. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    Catch on to what exactly? Don't you think if there was something to catch on to the review sites and AMD GPU owners would have done so already? If single GPU microstutter on AMD was such a problem everyone would have noticed it by now. Those of us who own both AMD and Nvidia cards have stated they get no more stutter than they get with Nvidia hardware? I and others have stated stutters are not a problem on our AMD GPU hardware. Ryan, the very person you called out to carry out an investigation has also stated he sees no difference between AMD and Nvidia in this regard.

    We all know that in Multi GPU Nvidia handles microstutter much better. This microstutter is not as common in single GPU config and in fact is no different between AMD or Nvidia.

    One single review site sees stutter in single GPU config when using Windows 8, and now there has to be an investigation? Even the [H] reviews show that in single GPU config AMD cards don't suffer any more than Nvidia cards. If you are referring to the Far Cry 3 performance review [H] state that it is only the scripted cut scenes that are affected by FPS stutters.

    I can assure you, if my HD 7970 system was giving microstutter problems compared to my GTX 680 system I would have noticed and kicked up one hell of a stink.
     
    #186 ICDP, Dec 7, 2012
    Last edited: Dec 7, 2012
  12. ZeroRift

    ZeroRift Member

    Joined:
    Apr 13, 2005
    Messages:
    177
    Likes Received:
    0
    I think the real question is: why are you opposed to it? Seeking knowledge is one of the driving forces of civilization. Many people have posted in this thread in the same pursuit. Enthusiasts want to know because they are enthusiastic.

    "Not a problem" is not a quantity that has meaning for those who are not directly making the observation. It is also subjective.

    Who is "We"? I think this is, in fact, what people want objective information in order to determine.

    Far more petty things have started killing wars. It is not surprising that enthusiasts seek to understand that which they are interested in. Furthermore, the simple act of collecting the information for use in investigation can be edifying in itself.

    What are you implying with this?

    What's wrong with wanting to know simply for curiosity's sake?

    Additionally, if there are significant findings from the investigation, wouldn't you want to know them regardless of what card you currently own?
     
  13. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    I'm not opposed to it, if such a test was done it would no doubt help people form a purchasing decision. My opposition is based on the request being made based on one test done on a Windows 8 system. The review site that posted this comparison has posted many reviews in the past showing AMD cards to have no serious problems with latency in single GPU configurations. I know from my own experience that in single GPU use AMD and Nvidia are no more or less prone to stutters. If multiple AMD GPU owners state they get no stutter problems and one chart says "yes you do" who are we to believe? It would be nice to see the results of this test out of curiosity, but the "big wars" alluded to by the OP is pure hyperbole.

    A person who has experience with something will trump someone basing their opinion on a chart every single time. Do you have an AMD GPU, if so do you notice stutter? If you don't own one then why are the opinions of those who do not objective enough for you?

    How on earth would seeing a chart on a web site tell you AMD cards suffer from stutter any more than the opinions of those who have experienced current AMD GPUs first hand? Would you tell someone who said "I have no stutters on my AMD GPU", that they are wrong and then point them to a meaningless chart?
     
    #188 ICDP, Dec 7, 2012
    Last edited: Dec 7, 2012
  14. RussianSensation

    RussianSensation Elite Member

    Joined:
    Sep 5, 2003
    Messages:
    19,459
    Likes Received:
    703
    This is 100% false and has been shown to be a problem for both AMD and NV multi-GPU systems.

    Micro-stutter gets worse as games get more GPU demanding and the GPU can no longer provide as high frames per second. This is why last generation multi-GPU cards exhibit more micro-stutter in modern games than current GPUs. A multi-GPU setup needs to render more framerates than a single GPU to feel as smooth (i.e., 60 fps on a multi-GPU card may feel only as smooth as 45 fps on a single GPU card. This was already investigated by Tom's Hardware and other websites before). This applies equally to NV.

    [​IMG]
    [​IMG]

    When the GPU power is there, GTX690 is much smoother compared to GTX590, but rest assured the minute there is a single GPU that can deliver GTX690 level of performance, it'll be smoother in Crysis 3 than GTX690, even though their performance will be similar in theory. GTX690 is still not as smooth compared to a single card, even in today's games.

    [​IMG]

    [​IMG]

    I have no problems at all if analysis of micro-stutter is investigated but for it to be great these 2 things have to apply:

    1) Please use Windows 7 x64. How many people are on Windows 8? Windows 8 can be tested in 2-3 years when adoption picks up to 50%. Right now it's irrelevant. Websites have also reported bugs, game crashing and driver issues for both NV and AMD in W8.

    2) The only way to fairly test NV vs. AMD is to use Radeon Pro for AMD. This is because NV incorporates Adaptive Vsync into their driver already but you need a 3rd party tool for AMD. Without using Radeon Pro for AMD CFX setups, the "smoothness" cannot possibly be measured objectively.

    Here is why:

    Nvidia's stock solution:
    [​IMG]

    Without Radeon Pro - micro-stutter clearly evident
    [​IMG]

    With Radeon Pro
    [​IMG]

    Actually, with even more manual tinkering, with Radeon Pro you can make micro-stutter less than a single HD7970.

    [​IMG]

    "Aside from a few dropped frames and a handful of spikes when the test changes scenes, our dual-Tahiti card enjoys much smoother sailing. In fact, the end result is often better than what you'd see from a single graphics card, with virtually no micro-stuttering left."
    http://www.tomshardware.com/reviews/radeon-hd-7990-devil13-7970-x2,3329-11.html

    If reviewers want to ignore the ability to enable dynamic Vsync on AMD's CFX setups, they will inevitably arrive at a foregone conclusion.

    I understand why a review may not use Radeon Pro since they want to measure out-of-the-box gaming experience but if anyone is spending $1000 on HD7970 CFX setup, chances are they are an enthusiast and at that point anything goes, just like enthusiasts use MSI Afterburner for custom fan profiles/overclocking, SweetFX for enabling a new level of IQ enhancements or using NV Inspector for injecting SMAA or improving image quality. Why wouldn't someone with $800+ of GPUs want to download a 3rd party tool that enhances their gaming experience?
     
    #189 RussianSensation, Dec 7, 2012
    Last edited: Dec 7, 2012
  15. ICDP

    ICDP Senior member

    Joined:
    Nov 15, 2012
    Messages:
    709
    Likes Received:
    0
    My apologies for that, I meant that stutters are only an issue with CF/SLI. Sorry.
     
  16. omeds

    omeds Senior member

    Joined:
    Dec 14, 2011
    Messages:
    637
    Likes Received:
    0
    Thats not fair at all. If radeon pro is used, then inspector should be too. The authors terminology is wrong, as has been discussed on many forums. Its not radeon pros adaptive vsync that fixes it, its the frame cap feature.

    If both CFX and SLI use a frame cap, resuts are on par. I've been saying this and using this method for almost 5 years on each technology, with excellent resuts.
     
    #191 omeds, Dec 7, 2012
    Last edited: Dec 7, 2012
  17. Siberian

    Siberian Senior member

    Joined:
    Jul 10, 2012
    Messages:
    258
    Likes Received:
    0
    I think smoothness is as relevant or more than fps. Very, very few people will watch a fps meter while they play. I bet most will notice a stutter while playing. Two sites have done a good job of answering the question. I don't see a lot of room for debate.
     
  18. Will Robinson

    Will Robinson Golden Member

    Joined:
    Dec 19, 2009
    Messages:
    1,408
    Likes Received:
    0
    This thread is hilarious.
    So now we have a new performance metric to add to video card tests?
    Smoothness/per watt :rolleyes:

    <Hint for the future> Watch "smoothness" sail out the window next time NVDA manages to get a frames per second lead.:whiste:
     
  19. Vesku

    Vesku Diamond Member

    Joined:
    Aug 25, 2005
    Messages:
    3,731
    Likes Received:
    21
    I don't think it reflects that well to assume there is a real "smoothness" issue with single GPU until more testing is actually done. Too much like the "coffee is good for you", "coffee is bad for you" tabloidish news articles, meanwhile science community has to repeat "nothing conclusive at this time" or "the papers conclusions are more nuanced".
     
  20. Siberian

    Siberian Senior member

    Joined:
    Jul 10, 2012
    Messages:
    258
    Likes Received:
    0
    You don't want your video card to run games smoothly?
     
  21. AnandThenMan

    AnandThenMan Diamond Member

    Joined:
    Nov 11, 2004
    Messages:
    3,777
    Likes Received:
    31
    Your posts are getting more and more ridiculous. I actually look forward to seeing how much more desperate your posts are going to become, especially if AMD comes out with the 8000 series before Nvidia has a refresh of their own. :D But really why do you want an investigation at all, AMD is going to be out of business very shortly according to you, by the end of the year is what you consider very likely. As to what I bold'd, YOU are paid Nvidia marketer, why should anyone take anything you post as objective.

    Now me personally, I don't notice any difference between my 670 and 7970 outside of the 7970 being obviously the faster card. This thread is completely pointless and a pathetic attempt to try and mitigate the fact that AMD currently has the single GPU crown.
     
  22. KingFatty

    KingFatty Diamond Member

    Joined:
    Dec 29, 2010
    Messages:
    3,025
    Likes Received:
    1
    I guess the issue here is some people believe/don't believe that smoothness is a real thing.

    I think it's a real metric that could potentially be measured, but so far none of the tech blogs have been able to do it (but I think they demonstrated that the typical frame-measuring tools aren't capable).
     
  23. bononos

    bononos Diamond Member

    Joined:
    Aug 21, 2011
    Messages:
    3,314
    Likes Received:
    22
    Because the problem of rendering each single frame cannot be fragmented across all gpu resources in an sli/xfire setup (the lowest latency of each frame is still dependent on the computing power of each single card) we have the problem of microstutter.

    So its not a 'new' metric per se since the objective of measuring framerates was to be a marker for smoothness/realism. Since there is a better marker for smoothness/realism, why not use it?

    The question now is whether the latency measurements using fraps or whatever is what the end user actually experiences which is why some posters have suggested using camera setups to measure real world on-screen latencies.
     
  24. Vesku

    Vesku Diamond Member

    Joined:
    Aug 25, 2005
    Messages:
    3,731
    Likes Received:
    21
    We know it can't be completely extrapolated to the actual experience because FRAPs can only report on part of the game processing path. So far it is that single Techreport article that shows some noisiness in the part of the path FRAPs measures. They have yet to post a more detailed investigation of their unusual results.

    Delving more objectively into this would be interesting. But I find the tint Keysplayr has put on this a bit over the top.

    I personally wouldn't take a position on something which requires a lot more examination.

    Big wars, really? Perhaps it's a reflection of Keysplayr's intentions?

    If by so much talk he means his own...

    My recollection could be completely shot but CF vs SLI microstutter has been discussed for quite some time.

    I'm noticing a trend about where 'all this "smoothness" talk' is coming from.

    Quite a strong statement to base on the one Techreport article.

    ---

    I'm actually as interested as Keysplayr is in seeing this looked into more thoroughly. But perhaps the hyperbole could be turned down a notch?
     
    #199 Vesku, Dec 7, 2012
    Last edited: Dec 7, 2012
  25. 3DVagabond

    3DVagabond Lifer

    Joined:
    Aug 10, 2009
    Messages:
    11,762
    Likes Received:
    57
    Objective is the key word. Let me know when we have a test done that is objective, but measuring something and then trying to attribute an effect to it that nobody who uses the hardware ever sees, isn't objective. It's finding a measurement that nVidia beats AMD in and trying to make it appear that the fact AMD cards are faster and better value is irrelevant because of this one statistic. Until we have a double blind comparison that identifies there is a problem, we have absolutely nothing to base these conclusions on.

    Back in the day (mid 80's) I was involved in double blind comparisons of Audio equipment. Until you've done something like that you can't understand how biased we all really are. You need to recreate the natural environment. An investigative environment often adds stress to the subject and interferes with one's ability to sense (hear, see, taste) differences. It's really not as simple as A/B comparisons.