NVIDIA’s ForceWare 53.03 release arrived with little fanfare earlier this month. This was due in part because the driver had already been available for quite some time on the Internet; a handful of NVIDIA’s board partners had already provided ForceWare 53.03 for release on their respective websites. No PR statements or announcements were made at the time the driver was originally released by these manufacturers, but it’s essentially the same driver that has provided on NVIDIA's website (with the sole addition being WHQL certification on the official NVIDIA driver).
As a result, these drivers have been thoroughly dissected in newsgroups and message boards throughout the ‘net, but we still wanted to take a look at them.
NVIDIA’s Release Highlights for ForceWare 53.03 mention “Increased graphics performance when using nForce3 systems.” But we actually found some performance improvements on our 875P test bed. Of course, the other major addition was official support for the GeForce FX 5900 XT, which we reviewed earlier this month in the form of eVGA’s e-GeForce FX 5900 SE.
However if you look a little further in ForceWare 53.03’s documents, you’ll find a few more interesting tidbits in the v53.03 Release Notes. For instance, on page 5 NVIDIA notes it has “re-enabled NVIDIA’s compiler technology for 3DMark 03 Patch 340 performance” as an “issue resolved” in v53.03. If you recall, FutureMark released a patch for 3DMark 03 (Patch 340), that, according to FutureMark’s Tero Sarkkinen, was released in order to “make sure that our customers can get an objective 3DMark03 performance comparison with the latest hardware and drivers. Our customers will be able to perform apples-to-apples performance comparisons between the various IHVs’ graphics cards.” FutureMark went on to state that:
“Parts of the program code have been changed so that possible 3DMark03 specific optimizations in current drivers will not work. 3DMark03 specific optimizations in drivers are against run rules of 3DMark03, because they invalidate the performance measurement results and thus make it impossible to compare performances of different hardware. A list of drivers that have been tested - and confirmed to produce valid 3DMark03 scores - has been published on Futuremark’s website.”
Of course, Patch 340 has only been approved for use with ForceWare 52.16. If you look over FutureMark’s list of approved drivers, you’ll see that ForceWare 53.03 has not been approved by FutureMark, while ATI’s CATALYST 3.9 and CATALYST 3.10 have, despite the fact that CATALYST 3.10 was released over a week after ForceWare 53.03.
With NVIDIA’s suggestion that Patch 340 disables their compiler, and FutureMark’s counter that it doesn’t, we probably haven’t seen the end of this story. Whether or not either company takes this issue public or not is another story however, we’ve already seen the two sides quarreling over this issue earlier this year. This is a subject that may possibly go well into 2004 before both sides come to an agreement, if at all.
Other than this topic, the other big issue is Battlefield 1942, specifically the popular Desert Combat mod. NVIDIA has resolved the rendering issues present in 52.16, but the problems with Splinter Cell’s shadow buffer mode we first mentioned in our Detonator 52.16 driver report are still present. We run all of our Splinter Cell tests in “projector” mode in order to run apples-to-apples comparisons with ATI cards, so shadows and dynamic lights render correctly for our benchmarking tests.
A growing number of NVIDIA users are complaining about this issue however, as the shadow buffer mode is offered exclusively to NVIDIA card users, but currently doesn’t work.