AMD's ATI Catalyst 10.3a Driver Performance

AMD’s new "performance" driver is out in the form of the non-WHQL Catalyst 10.3a. This driver provides performance improvements compared to 10.2 WHQL. We run apples-to-apples test in the latest games to see just how much faster it is in real-world gaming environments.

Introduction

AMD is on their game this generation, delivering gaming hardware that has impressed, and now driver features and performance that equally impress. If you are not at all familiar with what Catalyst 10.2 and 10.3 are bringing to the table, we highly suggest you start by reading our Catalyst 10.2 and Catalyst 10.3 Preview article first. Then we suggest you check out our ATI 10.3a Catalyst Driver Update were we discuss what the new Catalyst 10.3 driver will bring to the table in terms of performance improvements. There are some impressive claims being spouted, and in this brief article we aim to see if they hold water in real-world gaming, no timedemos or benchmarks here.

Quite simply, we are going to compare the latest officially released driver, Catalyst 10.2 WHQL to the newly released ATI Catalyst 10.3a Preview driver. We will be running five games in an apples-to-apples scenario, meaning all in-game settings, resolution, AA and AF are set at the exact same settings. We will be using the latest patched versions of: DiRT 2, Aliens vs. Predator, Assassin’s Creed 2, Battlefield: Bad Company 2, and an oldie but goodie, Crysis: Warhead.

Our system configuration is as follows: Gigabyte EX58-UD5, Intel Core i7 920 @ 3.6GHz, 6GB DDR3, Win7 x64. The video card we are using for all testing will remain the same, a 1GB AMD Radeon HD 5870.

The only difference in testing will be the two driver versions, which you can see below.

Article Image

Apples-to-Apples Graphs

We are simply going to list each graph down the page, and provide some commentary at the end. In all situations, the highest possible in-game settings are used, this means DX10/DX11 and every in-game option enabled and at their highest values in every game. We are testing at 2560x1600 with 2X AA in most games, except in Assassin’s Creed 2 where 8X AA was more than playable and in Crysis: Warhead we are using a lower resolution with No AA.

Article Image

Article Image

Article Image

Article Image

Article Image

AP2AP Summary

Obviously the biggest game to improve here is DiRT 2. AMD claimed an "up to" 20% improvement on Radeon HD 5800 series video cards in this game. We are seeing here a near 13% improvement, which is very good for a driver update to provide that kind of a boost. Unfortunately this was the only game to provide such a boost, in Aliens vs. Predator we are seeing around a 2% increase, which is a bit lower than AMD’s claim of a 5% performance increase, which is all rather small to begin with.

In Assassin’s Creed 2 this game wasn’t specifically mentioned in the list of games receiving improvements, and we can see why. Even at 2560x1600 with 8X AA the game is flat lined at the game cap of 64 FPS, it just runs that fast on the Radeon HD 5870. This game certainly doesn’t need any driver performance improvements, there is nowhere for it to go since it is game capped. If you want to get picky, it does have a higher minimum framerate with the new 10.3a Preview driver.

Battlefield: Bad Company 2 was also not on the list, but we did end up seeing a 5% performance increase on average. More impressive though was the improvement in the minimum framerate, which brought us up over 30 FPS at these settings.

Crysis: Warhead was on the list, and AMD claimed a 6% performance improvement. It seems here we are only seeing half that, close to 3%.


The Bottom Line

We generally saw performance improvements in every game we tested; there is no question about that. AMD should be commended for staying vigilant on improving performance in games. We know these guys work hard at it. However, it does have to be said that most of these improvements don’t add up to changing the gameplay experience. Anything under a 10% performance improvement can pretty much be written off in gamer’s eyes. To gamers, these small increases in performance are meaningless if they don’t let you enable an extra in-game option or raise graphics settings.

It should be noted also that this was just a handful of games tested, and AMD is claiming performance improvements in many more games. Therefore, the bottom line is that this driver is all around a good thing, and if you are running on Radeon HD 4800 series or the new Radeon HD 5000 series, you should grab this driver right now for possible performance improvements along with all the new features contained within.

The one question on everyone’s mind is of course, are these performance improvements introduced in Catalyst 10.3 going to be enough to edge out Fermi, and for that, we shall soon find out.

Discussion