30FPS Vs 60FPS Is “Negligible” Says Need For Speed Dev

30FPS Vs 60FPS Is “Negligible” Says Need For Speed Dev

NFSSHmultiSCRNprint07That’s the view of Need For Speed: Shift producer Jesse Abney, whose “authentic/” racer displays at 30 frames per second. Abney claims there’s plenty more under the hood that determines a racer’s performance.

I was in a Forza Motorsport 3 demo at E3 this year. During his presentation, Turn 10’s Dan Greenawalt said, “If it’s not 60 frames, it’s not a racing sim.” Greenawalt was placing emphasis on his game’s refresh rate, which he believes gives it an edge over its rivals in the field.

When chatting with Abney recently, I brought up Greenawalt’s remark, knowing that Shift runs at 30 frames. I asked Abney why, and what the difference is. His answer revealed a lot more about the development process than I expected.

“It’s funny hearing that. For one thing, these machines [pointing at the PS3 he had Shift running on]have multi-threaded architecture, meaning they can operate at different speeds; the control I/O, the driving physics, the collision physics, etc… all those things can run at entirely different speeds depending on which thread they’re on.

“And then the render thread, the speed of that and the refresh rate, which is really what 60hz versus 30hz relates to, is the end result showing up on screen.”

Abney points instead towards controller latency as a major factor in whether a racing game can feel like it’s running slower than it ought to be. As the Digital Foundry blog examined yesterday, this is the time it takes between a button press and on-screen action.

“In the past the clock cycles with which the simulation was gathering controller data was often slower than the input. So when users are doing this [waggles the analogue stick] , but the game was only updating every 1/5th of a frame, it would look like it was sluggish and going slower than it was.

“But now with multi-threaded architecture and a number of other improvements in pulling rates and radio signals and all that, especially to the speed of the processors, there is now virtually no lag. Really, as long as you’re running your controller I/O and your simulation thread at full speed, the perception of 30 versus 60 is negligible.”

My time with Shift suggests there’s merit to this argument. Combined with the fully articulated driver model and “Driver Vision” effects, there was little to suggest the game was running slow. Of course, the proof will be when we see Forza 3 and Shift running side-by-side. Not long to go now.


  • I like your trickling SHIFT coverage. This game is high on my ‘want it now now now’ list.

    By the way, I’ve spotted a typo in the last quote box:

    “Really, as long as your running your controller I/O and your simulation…”

    That first ‘your’ should be ‘you’re’.

  • Yeah… no.

    That is simply outrageous marketing spin, trying to cover the fact they’re stuck with a poorly implemented game.

    • And is this ‘Shift argument’ going to be the new Sony excuse for why games run worse on the PS3 than the 360?

    • “They can say what they like but 60 is always easier on the eyes than 30 :)”

      That’s only true if 60fps is maintained. If frame rate varies wildly, then you’ve got big problems in terms of player perception.

      In my books, a rock-solid 30fps rate is better than a varying, nominal 60fps.

      Besides, you can do a LOT more, visually, at 30fps than you can at 60, at a given hardware spec. You can add more filtering (including better AA), optical effects, better lighting, etc.

      I prefer a game to be more intricately detailed, and running at 30fps, than one that can run at with details missing at 60fps.

  • I dont mind if a game runs at 30fps as long as its consistently 30. Its worse in games like GTAIV on 360 where indoors it could be like 40-50, then step outside and you can see it clearly drops to well below 30 once the action heats up. If the frame rate is consistent you stop noticing it and its not an issue.

  • Burnout Paradise runs at 60fps, 720p. I personally just can’t imagine a racing game that didn’t run at that speed.

    First-person shooters, yes. RPGs, sure. Real-time strategy, no problem. But a racing game?

    And the argument about input lag doesn’t make any sense. If you press a button just as frame N is being displayed, you have to wait until frame N+1 is displayed before you can see it – regardless of how fast you process input in other threads. That’s 33ms at 30FPS or only 16ms at 60FPS.

    • “That’s 33ms at 30FPS or only 16ms at 60FPS.”

      That’s a truly insignificant gap, even in a racing game. What kind of change is the user truly expecting in such a time frame? A barely-discernable change in the rotation of the steering wheel? A slight change of camera orientation?

      If you’re talking about a sprite-based shooter, sure, the frame-rate can be an issue. But given the level of detail in a modern racer, and the mount of post-processing going on (motion blur, anyone?), 60fps is not an automatic win, especially for the sheer amount of processing power it requires. Processing power that can go in to accurate motion blur or lighting, for instance.

      • That’s 33ms assuming your input loop runs at “infinite” frequency. If it runs in time with the render loop (which it traditionally does), then it’s even worse – then it works out to between 1.5 and 2.5 frames of lag (depending on where in the render cycle you press the button).

        I’d say it’s more important for a racer than any other game to get the input lag down to a minimum. You’re typically moving so fast that 100-200ms delay can be enough to see you smash into an obstacle and when your input is introducing a 50ms delay on top of your reaction time, it can be significant.

        For a technical discussion of the different, see here: http://realtimecollisiondetection.net/blog/?p=30

        Anyway, I’m not so much concerned about the theory and if they get it to play fine at 30fps, I don’t really care at the end of the day 🙂

    • GTA4 natively runs at lower than 720p, rendered at 640p if i’m correct(?), so there’s always upscaling.

      So.. try changing your console’s settings to 720p for this game and just watch the smoooooooth GTA goodness unfold! A huge difference in the gaming experience, no texture popup and no controller lag. And don’t even think about going online using a higher res!

    • “At 30fps, when there’s action, it obviously drops and you notice this drop. GTA4 I could not play because of this.”

      But a solid 30fps frame rate allows you to safely to maintain the framerate by dropping features such as AA and other post-processing effects on a per-frame basis.

      It’s MUCH more likely that a 60fps game will drop, rather than a 30fps game will. And when it does drop, there’s not as many graphical options you can disable per-frame, in order to save the rate, because they’ve already been disabled in order to get the 60fps in the first place.

      The short version: a 30fps gives you much more scope for stability _and_ image quality.

  • I refuse to play any racing game that run at 30fps, it is terrible. Take good old sega rally 2 on dreamcast, the pal version was 30fps and the ntsc was 60fps, and it was like chalk and cheese between the two. Games like forza2, pgr3, Gt5prologue all run amazingly smoothly and realisticly at 60fps, 30fps, like the awful sega rally released last year on 360/ps3 and all the need for speed games are just hopeless. It’s these low quality racing games that keep selling to people who don’t care about any sort of realism or actual motion quality. Playing any game at half the framerate is a complete cop out with this generation of consoles.

  • I’ve done one experiment with Shift (PC, Driving Force GT, Geforec 8800GT, Intel E8400):
    Having the wheel centered, I moved it fast to the left (almost to the end of configured 270 degree). I noticed a small delay until the ingame wheel STARTED to turn (I would estimate 200-300ms). How can I drive in such conditions?
    I even tried turning off vsync (game), and tripple buffering (nvidia control panel). Tonigh, I’ll try turning off anti-aliasing to see if that affects anything and also use FRAPS to see my frame-rate. I’m even thinking of changing the “max pre-render frames” from nvcpl. If nothing changes, there’s no way I’ll continue with Shift.

    • I agree with you Mathew. 30fps is crap for racing games. I was testing Shift vs Race Driver Grid on my PC. Playing at 60fps is quite a bitg jump in playability. You just seem to be able to ‘feel’ the driving far more realistically. At 30fps it’s too choppy. Really it’s all about the minimum framerate – that’s what causes problems, not the maximum framerate. Is 30fps the minimum or maximum FPS for the console NFS Shift? I hope it’s the minumum. I find the demo of Shift on PC to be unoptimised compared to Grid. I just can’t get it running smoothly, even with everything on low settings (aside from on the straights, around corner it drops heaps). Where as with the same hardware Grid it wont go any lower than 55fps, with 4x anti aliasing on. (Quadcore Q6600 @ 2.3, 8gb ram, 8800GTX 768mb, 1920×1200 res)

      On another note, Shift seemly very ‘sluggish’ in responisivness… Why? I tried turning V-sync, and triple buffering off (D3DOverider), but there is still a lag. Aside from these performance problems it seems like a good game… But if they make it through to the final release, I wont be buying it – it’s just not enjoyable.

      Also, don’t underestimate the importance of Anti Aliasing in racing games – with so many sharp lines on screen, 4x AA does wonders for improving the clarity. It’s even worth dropping other settings down to low or medium, like textures and shadows. Grid with 4x AA is graphically superior to Shift with no AA, despite Shift’s new depth of field effects. (Shift’s lens flare sure looks pretty though!)

      • *update*

        I updated my Nvidia drivers, and now it runs very nicely on medium settings with 4x antialiasing. drops no lower than about 35 fps, and a solid 60fps on the straights.

        Can’t yet work work out why the cars feel so sluggish though… they just seem so less responsive, like there’s a half second lag on the controls. Weird.

Show more comments

Log in to comment on this story!