Top Holiday Games of 2008 Graphics Shootout - Page 1 of 6
Article By Mike Chambers - November 19, 2008
The holiday season is approaching and in typical fashion the game industry has or will be releasing a slew of titles for the occasion. On the personal computer, the two major players in the design of Graphics Processing Units (GPUs), ATI, which was acquired by AMD in 2006, and NVIDIA, both offer an assortment of exciting products.
From the low to high-end, ATI and NVIDIA offer consumer-based GPUs that fit the needs of gamers that range from casual to enthusiast. ATI markets their consumer-based GPUs under the Radeon brand name while NVIDIA uses GeForce. Although nV News is an NVIDIA fan site and primarily reviews NVIDIA-based graphics cards, we occasionally test graphics cards from ATI and their add-in partners.
Top Holiday Games of 2008 Graphics Shootout
Back in June when the Radeon 48XX Series hit the market, ATI forced NVIDIA to re-evaluate retail prices of their GeForce GTX 260 and GeForce GTX 280. The Radeon 4870 HD 1GB has been popular among the enthusiast community while NVIDIA believes that the GeForce GTX 260 Core 216 should be the graphics card of choice in the $200 to $300 price point for games being released this holiday season.
PowerColor Radeon HD 4870 1GB
ATI's most powerful GPU is currently the Radeon HD 4870 X2, which has two Radeon 4870 GPUs working in tandem on a single graphics card. ATI supports multiple graphics card configurations with CrossFire, which works on motherboards that use AMD or Intel chipsets. ATI plans to introduce ATI Stream in an upcoming driver release, which will compete with NVIDIA's CUDA.
The competition between the two companies has always been intense as each considers its products to be superior. NVIDIA recently introduced a more powerful model of their GeForce GTX 260 GPU and distinguished it from the original model by adding Core 216 as a suffix. The GeForce GTX 260 featured 192 general purpose "stream" processors, which increased to 216 on the Core 216. We recently learned that production of the original GeForce GTX 260 has been discontinued in favor of the Core 216.
EVGA GeForce GTX 260 Core 216 Superclocked
NVIDIA's most powerful GPU is currently the GeForce GTX 280. NVIDIA also supports multiple graphics card configurations via their Scalable Link Interface (SLI) and makes use of the GPU for physics special effects using CUDA and PhysX. Although SLI required a motherboard with NVIDIA's nForce chipset, they recently licensed SLI to a number of motherboard manufacturers that are offering motherboards based on Intel's new X58 chipset.
ABOUT THIS ARTICLE
The interesting part of this article is that NVIDIA asked us if we would be willing to conduct a shoot-out of sorts with these high-end GPUs. Although a scheduled vacation was coming up, which limited the amount of time that could be devoted to the project, the offer was accepted. Therefore, keep in mind that this comparison is not intended to be comprehensive as performance was tested in a handful of games that debuted around the 2008 holiday season.
Another important point is that some of the games do not contain an internal benchmarking mechanism so performance, in some cases, was based on gameplay or playing back a cut-scene using the graphics engine. Although we have been measuring gameplay performance since 2002 with Fraps, it does take time and practice to develop a walkthrough in a new game that is close to being repeatable and provides near consistent results.
Although this form of benchmarking is not always an ideal solution, Fraps allows games to be benchmarked that lack an internal benchmark thus allowing a wider variety of games to be tested. Since time was a constraint in getting this article completed, I was unable to provide video clips of the gameplay walkthroughs that were used.
The table below provides key specifications for each of the graphics cards being tested, both of which are overclocked out of the box. Of special interest are the different approaches that ATI and NVIDIA took in the design of their GPUs. On one hand, the Radeon 4870 HD is based on a 55nm fabrication process, which gives the PowerColor Radeon HD 4870 1GB a 28% higher core clock speed than the EVGA GeForce GTX 260 Core 216 Superclocked. The core clock controls the speed of raster operations, which involves reading and writing data between the graphics memory buffers and blending operations.
However, the shader, or stream processor clock speed on the EVGA GeForce GTX 260 Core 216 Superclocked is 69% faster. The stream processor is a general purpose processor that performs calculations on different types of specialized graphics programs (geometry, pixel, vertex, etc.).
Another significant difference is that the Radeon 4870 uses GDDR5 memory, which gives it an effective rate of exchanging data with the GPU at 3700MHz (925MHz x 4) across a 256-bit wide memory bus. The GeForce GTX 260 Core 216 has a much wider 448-bit data bus, but uses traditional GDDR3 memory, which limits the effective data transfer rate to 2106MHz (1053MHz x 2).
Also note the difference in the number of stream processors. This again is due to a difference in architecture as the Radeon 4870 basically has six blocks of stream processors that work on different types of operations, which was designed for optimization purposes.