Where did the Detonator reports go?
It has been a long time since we’ve brought you a Detonator driver report, but it’s not due to an oversight on our part – it has literally been months since NVIDIA officially updated its Detonator driver family, December of last year in fact. Because of this, there hasn’t been much to talk about.
We have received supplemental drivers for use with NVIDIA’s GeForce FX family (which are also compatible with older GeForce products), but these drivers were far from perfect. System stability with these additional drivers was below what we’ve come to expect from an NVIDIA driver, and we ran into multiple visual quirks with GeForce3 and GeForce FX cards in a variety of applications. It goes without saying that there was a reason these drivers were regarded as beta releases, they simply weren’t ready for primetime.
Besides the aforementioned issues (which have been discussed in our GeForce FX preview articles), there have also been suggestions that these drivers were provided merely to improve scores for NVIDIA-based cards in FutureMark’s 3DMark 03 benchmark. The situation has gotten so bad that FutureMark recently decided to discard any scores that were taken with these drivers in its database of user submitted scores on their website.
Fortunately, the problems we noticed with NVIDIA’s 42.6x beta drivers are no longer present in 43.45. We encountered no random rebooting (a problem that was readily present in 42.69) with any of the video cards we tested, nor did we run into any crashes or lockups. From what we can tell so far these drivers are a vast improvement over the releases we received last month, reminiscent of the experiences we’ve had with official Detonator releases we’ve tested with in the past.
In terms of visual quality, we also have no complaints. The texture corruption in 42.72 with GeForce3 has been resolved and we can also report that the quality issues with GeForce4 and 3DMark 03 have also been addressed. Texture quality has been left unchanged, but strangely enough we noticed that anisotropic filtering performance in “aggressive” mode is actually slower than “quality” (by as much as 4%). Visually, both modes look very similar, so we ran all our tests with NVIDIA’s quality setting enabled for this article.