|NVIDIA GeForce GTX 200 series technology preview|
|NVIDIA's GT200 architecture&heading=NVIDIA GeForce GTX 200 series technology preview|
|PhysX, CUDA, Medusa|
|GeForce GTX 200 series parts, Conclusions|
NVIDIA GeForce GTX 200 series technology preview
It might not feel like it, but it's been over eighteen months since we last welcomed a new generation of NVIDIA graphics boards, courtesy of their first DirectX 10 offering in the form of the GeForce 8800. After so much speculation surrounding NVIDIA not moving to a unified shader architecture, we were more than a little surprised to see them do just that, offering up their G80-codenamed core with 128 Stream Processors and both performance and a feature set that deeply impressed us.
From that point forward, NVIDIA have barely looked back, last year refreshing and eventually replacing huge 90 nanometre G80 core with the 65 nanometre G92 part, enabling them to release the fantastic GeForce 8800 GT, a part which is still hugely popular even now.
While he haven't heard the last of G92 by a long stretch, as it will continue to power numerous parts in NVIDIA's line-up for some time to come, it's time for something new at the high-end of the graphics giant's product range. Enter the GT200 architecture and the product line which it powers as of today, the GTX 200 series. While NVIDIA's second-generation unified shader architecture is certainly a case of evolution over revolution in general, there's still plenty to discuss and consider. Despite attending NVIDIA's UK event to celebrate the launch of this board, we've been left without any samples or even hard copies of any slides, architecture diagrams and the like, but nonetheless we shall battle on and provide you with some insights into both the GT200's core architecture, as well as the exact make-up of the two GTX 200 series parts being launched today. So, read on to get to grips with just what NVIDIA is serving up this time around.
NVIDIA's new GeForce nomenclature
Before we go anywhere, it's probably sensible to start with a brief explanation of NVIDIA's new graphic board naming scheme from this launch onwards, as some of you are doubtless what happened to the GeForce 9900 or the GeForce 10 series or whatever you were expecting to come next.
After plenty of feedback regarding how confusing NVIDIA's graphics product lines were becoming, and somewhat in line with AMD's recent change in graphics board naming, NVIDIA have decided that it's time for a change, to simplify their nomenclature and (hopefully) make it much more readily apparent where a particular product fits into their line-up.
Today sees the launch of the GeForce GTX 260 and 280, so let's just break these product names down. Firstly, the GTX moniker is now going to be used for any high-end, enthusiast level boards - That means no more use of the Ultra name, nor will we be seeing the GTS name used against a high-end part. Of course, AIB partners are still free to add weird and wonderful names to their products, but the basic rule is that GTX will stand for high-end boards and high-end boards alone. As time goes on and this naming scheme gets fleshed out, we should see further names like this introduced to differentiate mid-range and low-end parts, which should make for a big improvement over seeing GT, GTS and GTX names used in each separate product range.
This leaves us with the three digit number, which is equally as simply to decipher. The first number denotes the generation of the architecture - In this case NVIDIA considers this their second-generation unified shader architecture, hence the '2'. The other digits are used to denote performance within the band of graphics board concerned - For example, today's launch takes in two high-end GTX parts with different performance characteristics, thus these are afforded the numbers '60' and '80' to denote relative performance to one another.
Of course, the proof in any naming scheme is how it holds up once we see a full top to bottom range of products using it, so we'll have to wait and see how it handles the usual flood of differently specified parts that hit the market, but this certainly seems like a step in the right direction and appears to be far more logical than the often-confusing names we've seen from some recent NVIDIA product launches.