|
Radeon 9000 products, as with the card we review in this article, are based on ATI's latest RV250 Core. ATI typically denote a value card by the addition of a "V" in the product name, for instance the VE solutions, equating to Value Edition Cards. This addition of a "V" into the RV250 Core codename indicates that this solution is indeed a value part. The addition of the "50" gives the impression that the card is in fact a higher performing solution than the R200 Parts, i.e. Radeon 8500, yet with the aforementioned "V" it's quite confusing as to exactly what to expect from the RV250. One could assume ATI's implication with the codename is that the RV250 is in some situations a better performing part than the R200 based products, but not entirely so.
Aesthetic reasoning aside, exactly what differs in terms of design between the R200 and RV250 one might ask? Well, unlike NVIDIA's solution on how to address the value market, that is remove the DirectX 8 programmability from the GPU for use with their Geforce4 MX based products, ATI chose a far more suitable route.
The original R200 Core had a total of four pixel rendering pipelines, with two texture units per pipeline, and as such, maintains a somewhat high production price. Cards based on the RV250 while maintaining all four rendering pipelines, adopt only a single texture unit per pipeline. The benefit of this architecture is that ATI are able to significantly reduce the die size of the RV250 in comparison to the R200, keeping costs low, yet maintain a full DirectX 8 Programmable Part.
One of the more beneficial improvements integrated into the RV250 is an upgraded triangle setup engine, boasting many of the optimisations reminiscent of the setup found in the R300. The texture units on the R200 can each deal with 3 textures at any given time, whereas those found on the RV250 can deal with 6 textures at any given time. This means that in some situations, such as single texture fill rate benchmarks, the RV250 should out perform solutions based on the R200. This should also result in improved vertex shader performance through the parts single vertex pipeline.
This design results in the RV250 looking to maintain a competitive level of performance in scenarios, more so games, that only utilise one or two textures at any given time.
The disadvantages to these architectural modifications are that in situations that make extensive use of a higher number of textures, such as the recently released Unreal Tournament 2003 (up to 4 textures in a scene simultaneously), the RV250 will not perform as well as an architecture such as the R200.
So, although in some synthetic benchmarks, the RV250 may well outperform the R200, considering most recent games make use of multitexturing, it is likely the R200 based solutions will remain the higher performing part in gaming scenarios.
Moving away from the architectural changes and laying focus on the hardware features, the RV250 remains largely unchanged from solutions based on the R200 Core.
The RV250 maintains full DirectX 8.1 Compliant hardware, so features such as the pixel shading effects found in the latest game releases are fully supported. The Anisotropic Filtering and Anti Aliasing (SuperSampling) engines remain unchanged also, and we'll touch upon exactly what level of performance reduction each feature impedes upon the RV250 later in the review. ATI's HyperZ II Technology, a bandwidth management technology, also remains unmodified from that of the R200.
|
