Well boys and girl, nVidia’s done it again. Less than a month and a half after the launch of the GeForce 7800GTX today sees the release of its ‘little’ brother – the GeForce 7800 GT. As with the 6800 GT this is likely to be a big hit for nVidia, offering much of the power of the 7800 GTX at a lower cost.
But whereas the 6800 GT was, bar clock speeds, the same as the Ultra, the 7800 GT does have some architectural differences to the GTX. The major one is that the GT has 20 pixel pipelines, rather than the 24 of the GTX and seven vertex shaders instead of eight. Clock speeds have been reduced too, with the core running at 400MHz compared to 430MHz for the GTX and the memory at 1,000MHz compared to 1,200MHz. That said the two retail GTX boards we’ve looked at from XFX and AOpen were both running at 450/1250MHz for core and memory so it’s not unreasonable to expect many retail GT boards to also offer a ‘free’ boost.
Aside from this the GT is the same as the GTX, based on the same 0.11 micron process. Fully present and correct is full support for Shader Model 3, High Dynamic Range and SLI – the trio that nVidia’s marketing people like to refer to as ‘the Power of Three’. Right…
In terms of line-up the 7800GT appears to be equivalent to the 6800 vanilla, which likewise offered reduced pipelines (12 instead of 16). This means that the GTX is occupying the same space at the 6800GT did leaving space for an ‘Ultra’ type card to be released above the GTX, presumably when nVidia feels that the time is right. We can also predict that nVidia will produce 16-pipeline versions of the 7800 series when it becomes more economical to do so than producing the 16-pipeline 6800 series. These would certainly be faster than the 16-pipeline 6800 cards due to the architectural improvements that nVidia has made with the 7800 series, so it would be faster performance at a lower cost - good news for consumers. That said, this is all speculation at this juncture.
What is hard fact though is that the 7800GT is here, and from the scores it’s clear that nVidia has a winner on its hands. As you can see from the picture the PCB itself on the reference GT itself is actually shorter than that of the GTX. With the reduction of the pipelines the power requirements have dropped so that only a 300W PSU is required compared to 350W for the GTX. The heatsink on the reference card has been tweaked and there’s no heatsink needed over the power regulator. A power connector still needs to be plugged in to the back of the cards, dual DVI is present and VIVO is supported too.