The middle of ATI’s product line-up has been a little bit of a mess in recent times. The first performance-mainstream part ATI launched in the Radeon X1000-series was the
Radeon X1800XL – a card that was seemingly priced out of the market when ATI launched the
Radeon X1900-series in January, less than four months after its introduction. The Radeon X1800XT 256MB took its place and then ran along side the first
Radeon X1900GT that ATI launched in May.
Since then, ATI decided to change the clock speeds of Radeon X1900GT without really telling the press. Although the changes were made for the best – very similar performance characteristics, lower power consumption and the inclusion of HDCP support – having two Radeon X1900GT’s with vastly different clock speeds would only help to create more confusion in the popular performance mainstream sector.
Less than a month after ‘announcing’ that the Radeon X1900GT specifications had changed, ATI has launched the replacement for Radeon X1900GT. It goes by the name of Radeon X1950 Pro, but don’t mistake this card for using the same R580+ GPU as ATI’s current flagship part, Radeon X1950XTX.
Instead, Radeon X1950 Pro is based on the 80-nanometre RV570 GPU. This GPU sizes up at about 238mm² and is made up of approximately 330 million transistors. RV570 comes with an impressive set of specifications, including 36 pixel shader processors, 12 texture units, 8 vertex shaders and 12 pixel output engines capable of FP16 HDR lighting formats and anti-aliasing at the same time.
Along with simultaneous HDR and anti-aliasing capabilities, RV570 is also capable of utilising ATI’s high quality anisotropic filtering technique that comes as standard on all Radeon X1000-series cards. Not all cards are capable of smooth frame rates in games with it enabled though. We’ll find out whether both of these features are useable on Radeon X1950 Pro a later on.
Another feature that ATI has recently spent a lot of time talking about was its
Stream Computing Initiative. The Radeon X1950 Pro supports the new Folding@Home client that uses the GPU to enhance application performance by a significant amount. While we haven’t measured the performance difference between processing work units on the GPU and in the more traditional manner using CPU cycles, if it’s even half of ATI’s claimed increase (somewhere near 40 times faster), it is still a huge performance enhancement.
Native CrossFire Support:
Finally, ATI has caught up with NVIDIA. It’s been a long time coming, but with Radeon X1950 Pro, virtually all of our concerns with previous CrossFire implementations are wiped off the slate. All that you need to get Radeon X1950 Pro CrossFire running is a pair of Radeon X1950 Pro cards, a pair of CrossFire connectors that look incredibly similar to NVIDIA’s own SLI connector (albeit a little wider) and a motherboard capable of running CrossFire.
ATI has done this by incorporating the CrossFire compositing engine into the GPU, meaning that it’s included on every Radeon X1950 Pro as standard. This alone eliminates the need for one CrossFire Edition card (with the compositing engine) and a CrossFire Ready card along with the dongle to connect everything together. CrossFire of old was a last-minute solution designed to compete with NVIDIA's reborn SLI and there’s no doubting that, in our opinion, the new method of CrossFire is a lot more refined and elegant.
ATI’s all-new CrossFire connector The internal connectors are very similar to NVIDIA’s – in technical terms, it is a pair of 12-bit connectors capable of passing data bi-directionally at up to 350MHz. With NVIDIA’s SLI implementation, data can only be passed over the SLI connector in one direction at a time. This may help ATI’s SuperAA performance against NVIDIA’s SLI AA performance, as it allows the two cards to combine anti-aliasing patterns much faster.
I am glad to see that ATI has made this move (dropping the external CrossFire cable) – it has been a long time coming and I think it is safe to say that CrossFire can now be taken seriously.
Want to comment? Please log in.