If you’re using software to measure the quality of your gaming experience, you’re missing something.

We’ve been working for years to analyze and improve the gaming experience for our customers. What we found is what many customers have long noticed: the most common software tool for measuring a game’s performance – FRAPs – doesn’t always capture what users are seeing.

Particularly when using multiple graphics cards, users are noticing that the action can often pause and stutter even when their software tools are telling them they should be seeing silky-smooth action.

The problem: FRAPs accurately measures frames when they are transferred from a game engine. But a lot happens between that point and when a game gets to the screen. In fact, we found that gamers weren’t imagining things: there can be a big difference between what users see on FRAPs and what they experience.

Two of the problems: what we call ‘drops’ and ‘runts.’ Drops occur when frames that are counted by FRAPs are never displayed. Runt frames, by contrast, are displayed, but only for a few lines of the full 1080p that should be displayed.

FCATA

Sometimes these drops and runts can cause visible ‘tears’ in the scene being put on the screen. Other times they’re too small to see, but like drops they break the smooth flow of the onscreen action, resulting in a stuttery experience.

So, to better understand what was going on behind the scenes and to help make the overall gaming experience as good as possible, we devised a solution that we call Frame Capture Analysis Tools or FCAT for short.

Capturing the data and properly analyzing it isn’t a trivial task. It includes the use of a special capture card, special overlay software that shows a ‘color bar,’ or fixed color sequence on the display, and scripts that help analyze and graph the data.

By using the capture card with the on-screen color bar we are able to compare the content that’s actually shown on the screen with what we know should be there and correlate that to what the real gaming experience is. Simple.

FCATB

We’re proud of the work that we’ve put into this – and we think it can help gamers get the experience they’re paying for. So we’re opening up our FCAT solution, making the scripts and software associated with FCAT freely-modifiable and redistributable. The technical press has already dug in, and the results have been dramatic.

Our hope: that third-party apps can replicate and replace our tools, giving gamers what they need to be sure they’re getting all of the graphics quality they’re paying for.

  • http://www.facebook.com/profile.php?id=1379241255 Francis Paolo Manansala

    basically what pcper guys were doing

  • http://www.facebook.com/profile.php?id=1282903929 Fotiou Alexandros

    Nononononononononooooooooooooo, my life is a lie now!!!!!!!!! 

  • Dominic Sharoo

    The technical press had a much bigger role in developing and testing this test methodolgy than this blog post indicates. This was not just a singular NVIDIA initative but more of a collaborative effort with the press.

    Surely PC Perspective especially could get better recognition and commendation for the hard work they put in to develop this testing than just the nondescript hyperlink in the above blog. As could Tech Report and others who investigated other non traditional types of performance testing.

    Even the commenter below, Francis has recongised this “basically what pcper guys were doing”

  • http://www.facebook.com/alexey.nicolaychuk Alexey Nicolaychuk

    Hi Tom,
    I’ll be glad to add FCAT overlay support to EVGA Precision / MSI Afterburner OSD.

    Alex aka Unwinder

  • Tom Petersen

    Hi Francis, Just to clarify, we have been working with press over the last 2 years to make sure that the tools we develop are useful and clear.  Ryan at PCper has been very helpful.  He has been working with me closely and invested a lot of time and energy to make this possible.

    It is fair to say that though that most of this effort was envisioned, designed and refined by NVIDIA.  The Overlay concept is pure NV, and all of the FCAT parsing scripts are written by NVIDIA.  Ryan is continuing to apply his own ideas and I expect him to continue to improve his own metrics – like he does now with his variation measures.

    For NVIDIA this effort is far from over.  We need to figure out good ways to characterize input latency and also systematically capture some kind of stutter metric.   We will continue to invest because it helps us improve the experience we deliver and it helps our review community better understand the technology we deliver.

  • Luciano Saraiva

    Can everyone download FCAT?
    Unwinder has already written the overlay and it worked just fine for capture.

  • Tom Petersen

    Anyone can download FCAT.   Review sites around the web are posting the tools.  We plan to have posting hosted on NV soon as well.  Of course FCAT requires a pretty high end capture rig where a capture card is used to grab the realtime displayed images.

    Anybody can look at the scripts that post process that captured data though.

  • Tom Petersen

    Nice Job:)  I saw MSI’s beta is available.  very quick.

  • Prowler Prowler

     Agreed!

  • huebie

    Hi Tom,

    i just saw the interview with Ryan from PCper.com and i found it very interesting even if the deep questions are still unanswered. What and when can we expect for future releases of FCAT and upcoming drivers? Will there be a engine-by-engine testing period and updates for the specific profiles. I personally run a Fermi SLi at 120 Hz and did some testing with every single game (e.a. The Witcher 2, Far Cry 3, StarCraft II, Two Worlds II Battlefield 3, TES V Skyrim and so on) to find out which setting gives me the best experience. I noticed that the pre-renderlimit is an important option in combination with a framelimiter.
    What are your (or nVidias) experiences with that option and how (or where) does the prerenderlimit is set? Does the CPU “know” about the limit or is it set after the present call from the engine?
    Whats “Vertical Sync smooth AFR Behaviour” Option in the NVAPI? And why you hadn´t talked about orbmu2k´s nVidia Inspector which is a mighty tool via the NVAPI.

    I would really appreciate an answer and look forward to see you guys dig deeply into the topic to find solutions.

    Greetings from germany.

    Marcel.

  • Phillip Hendrickson

    Tom,

    I just finished watching your discussion with Ryan over at pcper.com ( http://www.pcper.com/news/Editorial/PCPer-Live-Frame-Rating-and-FCAT-Your-Questions-Answered-0 ), and I have a couple of comments, with a question that you may or may not be able to answer.

    First, the comments:

    -Great discussion – the tool that you’ve devoloped in FCAT, which has enabled the work that you guys have done to reduce variance in frame latency/timing, is great.  And it’s also great that you’re making these tools available to the wider community as a way of fostering a community discussion.
    -One of the questions, towards the end of the discussion, asked about what we humans perceive as “smoother” – lower frame latency (ie, higher frame rate), or lower latency variance.  I think that your answer was very good – there’s a lot of space between one extreme end, where you have a high average frame rate with a ton of variance, vs the other extreme, where you have very little variance but a very low average frame rate.  I’ll be interested to see when and whether any consensus comes about where the sweet spot is for average frame rate vs. variance (the assumption here is that pursuing one puts you at a disadvantage with the other, which may not actually be accurate).
    -I’m very interested to see where this discussion goes with regard to multi-monitor gaming.  You, of course, noted that there hasn’t been a lot of detailed discussion around multi-monitor resolutions (5760×1920, etc.), but I look forward to seeing more of that, as I’ve been using a 3-monitor desktop for several years now, and really like my setup.  Side question – are the video capture cards that you and Ryan have been using capable of capturing 5760×1920 pixel images at 60 hz?

    Finally, the question: in the first image in your post, the one that shows the graphics pipeline, from the game engine all the way to the monitor, it’s fairly obvious (to me, at least) that there’s no feedback loop in the pipeline.  From the standpoint of pretty much any biological system (I work in computational neuroscience), a feedback loop is really important from both a control and regulatory standpoint.  I’m guessing that the frame metering technology that Nvidia has implemented in the last couple of years is an effort to add some kind of feedback loop to the graphics pipeline; how much benefit could be derived from adding some kind of feedback loop directly to the game engine?

    Again, great stuff – I look forward to seeing how the discussion continues with more time.

    Phillip

  • ullaulla

    any download link ?