Slashdot Mirror
Half-Life 2, ATI, NVIDIA, and a Sack of Cash
Latent IT writes "If you're into games, and unless you've been living under a rock for the past few days, you've heard a bit of a rumble from Valve on the relative quality of ATI vs. NVIDIA cards. Starting with articles like this one (previously reported), Valve told the world that the ATI 9800 Pro was nearly three times faster in some cases than the formerly competitive NVIDIA offering, the 5900 Ultra. Curiously, this happened at an ATI sponsored event, "Shader Day". But the story hasn't stopped there. NVidia released this response, essentially claiming that their new drivers, that were available to Valve at the time of their press conference, would make for vast, legitimate performance improvements. An interview with Massive, the creators of the Aquamark 3d benchmark, seems to confirm this opinion - that the NV3x chipset wasn't designed around any certain API very well, and the drivers are critical in achieving good performance. Anandtech writes here about the restrictions Valve placed on what benchmarks could be run. However, the key to this whole story may be this: an article, which I haven't seen get much coverage in all this, seems to make everything a little clearer - Valve stated that their OEM bundling deal with ATI came from the fact that ATI's cards were so superior, and that they were "performance enthusiasts". However, if the Inquirer is to be believed, the bundling deal was a result of an outright auction, on what will probably be the most popular game of the year. Which year that might be, is another issue altogether. Whatever happened to just making hardware, and making games?"
Raw power, eh. That must be why the Radeon 9700 Pro, with a GPU clock of 325MHz, was equivalent to the GeForce FX 5800 Ultra, with a GPU clock of 500MHz. The Radeon 9700 Pro was so focused on raw power that it put out a whopping fifty-seven watts of heat to the 5800 Ultra's mere eighty! The 5800 Ultra had a far more sophisticated cooling system, of course, which consisted of a copper heatsink that stole a PCI slot and a banshee-like fan. Now that's what I call finesse! Do your homework before you post :P
ATI bought the guys who did the chip for the Gamecube, and they did clean paper DX9 design for ATI. ATI went from being a year behind NVidia (DX8 generation) to being a year ahead.
In the R300, ATI decided to do all their calculations in 24 bit floating point: essentially a pure next-gen chip. The NVidia Geforce FX design was based on their DX8 chips, which were far and away industry leaders in fixed-point calculations; NVidia didn't figure that floating-point performance would be very important this generation and tacked it on. What they ended up with was a chip that had a high transistor count, was very good at legacy, fixed-point operations but could not keep up with ATI in floating point. Even then (about a year ago) NVidia's chip might have been competitive but they had process problems that made the chip clock slower than expected and about 9 months late.
ATI's superiority in floating point shaders has been demonstrated by various benchmarks (including some open-source benchmarks, which are the only ones I really take seriously) time and again. NVidia can only be competitive this generation when they 'tweak' their drivers for particular benchmarks. These tweaks sometimes consist of rewriting floating-point shaders to use their legacy fixed-point functionality, and on some occasions of even using pre-generated shadow models to replace the dynamically generated models of benchmarks that run over a known scene.
NVidia's NV3x generation seems weak, compared to ATI, and very weak unless game coders ignore API standards and write custom shaders that do as much as possible in NVidia's legacy hardware. Of course, by historical standards NVidia's NV3x isn't weak at all--they blow away all their competitors and ATI's pre-R300 products. It's just that the design choices made by ATI's new designers allowed them to leapfrog a generation.