How Additional Aerodynamic Drag Helped Make GTA III Work On PS2

The PlayStation 2 was a revelation when it hit the market in 2000, and yet by modern standards, it’s almost hopelessly weak. In fact, it’s so under-powered, Rockstar developers had to pull every trick in the book to make Grand Theft Auto III even work on the platform.

The story comes to us from developer [Obbe Vermeij]. He explains that the PlayStation 2 couldn’t keep the entire open-world game map in its tiny 32 MB of RAM. Instead, models had to be streamed from the DVD drive as the player moved around the world. However, even the DVD drive wasn’t fast enough. If the player moved too quickly, they would outpace the system’s ability to load new assets, and the world would fall apart. Roads would vanish, buildings simply wouldn’t appear before the player passed by them.

According to [Obbe], getting around this challenge was the job of one [Adam Fowler]. He notes that even optimizing the layout of data on the DVD wasn’t enough to help. Nifty hacks had to be employed to slow the player down. Road networks were changed to stop the player speeding towards areas that needed lots of new models. In other areas, vehicles in the game would experience a nearly-imperceptible 5% increase in air drag to dull their speed. This was chosen as a more invisible solution; cutting engine power directly was audible to players as the audio changed.

It shows you just how hard developers had to work back when resources were far more constrained than they are today!

17 thoughts on “How Additional Aerodynamic Drag Helped Make GTA III Work On PS2

  1. If there’s one thing that’s been a constant over the decades I’ve been playing with consoles and working with computers it’s that devs will always use more resource than is available, in 25 years time I don’t have any doubt we’ll be seeing articles just like this about today’s games.

      1. It’s mostly about getting it working as fast as possible and worrying about performance later.

        But then the managers come in and see it working, and say it’s done – go sell it.

        1. There is a battle between perfection and reality. If you ship a bad game then you might struggle financially, if you don’t ship a game until it’s perfect then you will struggle financially.

          1. Or you can go the Bethesda way and pre-sell the game to the point of breakeven by name and hype alone, then release it buggy and let the player-modders fix it for you.

    1. I don’t think we will have the same sort of thing nearly as much in current games – they are almost all built on top of handful of engines that do most of the work and also restrict many of the tricks you might try to pull to increase performance. Plus hardware now is so insanely powerful as a baseline you can get very very playable performance that looks really good out of very cheap low power chips without much effort beyond having a lower quality texture or polycount models in the same ‘universal’ game engine so there isn’t much need to look for ways to push the boundary of performance further.

      I’m sure there will be exceptions but for most games there really won’t be a need to pull cunning tricks for performance to nearly the same extent there used to be now even low end hardware is so impressive and human senses are not getting any faster/sharper. Though if/when VR type gaming kicks off in a bigger way…

      1. Ragdolls in Postal 2 were bad in a funny way, ragdolls in Max Payne 2 were decent… but those are both 2003 games. 20 years later and in AAA game you get to roll owls down the stairs as if it’s still Postal 2:

        Actual collision model is probably a single cylinder. Owl’s collision-less ragdoll keeps ragdolling while its collision cylinder rolls down the stairs.

        It’s such a shame we don’t get in-game physics like in GTA IV. It was so much fun to play with car crashes and pedestrians. Only BeamNG comes close.

        1. That is all design choice and lack of interest in making the collision models and rag doll physics though. It is not a limitation on performance that requires cunning tweaks to keep the game functional because the hardware can’t hack it, just ‘lazy’ developers making the entirely rational decisions to not bother as its not important for this game.

          You can have very good ragdolls, even very good deformation and destructive terrain physics if your game designer wants to put in the work to make all the visual models actually map more properly to the collision models, and then have a mechanism for them to break when they should. Its just not worth it for most games. Take Space Engineers or Kerbal Space Program for instance, not 100% realistic physics, but they go far enough to have impactful and destructive collisions, and both are rather old games now.

          Though that sort of in depth physics modelling is something I was thinking of as a potential exception needing cunning tweaks to get around limited hardware, but in practice more often than not its not a limitation at all. Just something the developer didn’t want to bother with, usually because there really isn’t a point. (With other options being related to the VR space – as that is computationally very expensive and the ‘AI’ brains behind the game)

        2. Rolling an owl is actually pretty hard, since it’s mostly a fluffy ball of feathers that can go every which way, so it has no definite shape. It’s one thing you would expect them to skip over, since it requires such intricate modeling of the object to get the collision box “just right”. Getting it half-way right will still look just as bad, so it’s better to just ignore the problem than spend inordinate amount of developing time optimizing for this one case.

      2. Uh…this is still 100% the case with modern consoles. Take the Switch for instance, there are countless ports to it that literally have to use tricks to run. Zelda, RDR, etc. Take a look at Modern Vintage Gamer’s Switch videos for proof (he has a great channel btw)

        1. The switch is a very odd case, being a chip that was basically obsolete even before the console launched in performance and its now 7 years older than that, well on the way to retired. Plus games on it, even ones designed exclusively for it don’t actually tend to run well no matter the tricks they pull as the cooling on a switch unit is so insufficient the chip can’t even perform to its best, on top of the chip being pretty garbage…

          In effect it shouldn’t count as representative of modern as it isn’t anyway, and you get actually decent performance emulating a Switch game on a super cheap end PC than you do on the switch too, even with the emulation overheads. Plus I did say ‘nearly as much in current games’, not never at all! The early days of 3d gaming had lots of tricks required and everything was pulling some of them, now…

          1. You say: “it’s not used today,” I provide you with a common use today.

            You say “doesn’t count,” I say Nintendo consoles are ALWAYS under powered with obsolete chips that *need* tricks just like the Switch.

            You sound very young without much real world experience.

          2. At this point Sword the cheapest crappiest chip the Big N could plausibly put in the next generation without having to start up a new production line of extra obsolete designs just to ‘always be underpowered’ will just play modern PC games at useable frame rate with no major tricks to make it work. It is going to be the exact same game the giant gaming PC tower with its 500w GPU and 200 odd Watt CPU would run, just with the lower quality textures etc – no extra tricks for performance required… Case in point the hordes of handheld gaming PC like the Steamdeck – which is also running effectively obsolete silicon, but much less so. Meaning it has the performance to run almost any modern PC game at 30+FPS very stably on its screen (even in many cases at 30-60fps at 1080P and 2K resolutions)…

            Also Nintendo consoles are not ALWAYS underpowered, often especially in really recent history but not always, they are just at least since the 8 bit era always weirder, the N64 and even gamecube for instance are very capable for their era, more held back by other weirdness than outright performance. Now since the Wii I’d agree they have always been compared to the competition underpowered, though it is still more the weirdness that makes them different…

          3. Oh also I didn’t say the tricks would never be needed I said “nearly as much in current games” and “I’m sure there will be exceptions but for most”.
            The Switch can be argued to be an exception, but for me I’d argue its not ‘current’ instead.

  2. Jak and Daxter had a similar but more obvious solution, Jak would trip to buy the system more time to load in adjacent levels.

    But it’d only do it once during that event, which means if it didn’t catch up, the ground in those levels will be black, and will still have its kill box.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.