Test setup

All of today's testing has been run on the following:

- Intel Core 2 Duo E6600
- 2GB Corsair PC6400 DDR2 RAM
- MSI 975X Platinum PowerUP Edition (Socket LGA 775, PCI Express)
- 250GB Western Digital Caviar SE16 hard drive
- Sony DVD-ROM
- NVIDIA GeForce 8800 GTS 640MB
- NVIDIA GeForce 8800GT 512MB
- 1000W Thermaltake Toughpower power supply
- Windows Vista Home Premium (64-bit Edition)

The following drivers were used:

- NVIDIA ForceWare 163.69 WHQL was used on the GeForce 8800 GTS graphics board
- NVIDIA ForceWare 169.04 Beta was used on both the GeForce 8800 GT and GeForce 8800 GTS graphics boards

The drivers High Quality texture filtering mode was enabled on the NVIDIA boards for all testing.

ForceWare 163.69 WHQL performance

To set a performance baseline for at least one of the boards we'll be utilising for our testing, let's begin this section using NVIDIA's current WHQL drivers for the entire GeForce 8 series, ForceWare 163.69.  These drivers were released quite some time before the Crysis single player demo, so we'd expect them to be free of anything in the way of application specific optimisations for this title.

For these tests, we're again using Crysis with all in-game options set to 'High', while utilising a GeForce 8800 GTS 640MB part at both 1280x1024 and 1600x1200.  First up, let's check out performance without anti-aliasing enabled, with the game executable named both crysis.exe (the actual executable name) and driverbug.exe (our made-up, altered name):

As you would expect, no differences in performance despite the change in executable name here.  Now, with 4x anti-aliasing enabled in-game:

Again, no performance changes outside of the usual margin of error.  Everything is absolutely normal with this driver set as anticipated.

ForceWare 169.04 beta performance

We now move on to run the same series of tests using NVIDIA's latest beta ForceWare 169.04 drivers, again utilising the GeForce 8800 GTS 640MB in a similar fashion, but also adding the recently released (and mighty impressive) GeForce 8800 GT into the mix.  So, we've seen the image quality differences created by changing the executable name of the Crysis single-player demo, but what impact does it have on performance.  Again, we start without anti-aliasing enabled, testing at 1280x1024 and 1600x1200, with in-game quality at 'High' - First, on the GeForce 8800 GTS.

We can immediately see the effect of the executable name change to driverbug.exe at both resolutions, as the game returns to its normal behaviour with regard to updating water reflections.  Using Crysis' default GPU rendering timedemo, this equates to just under a 7% drop in performance at both resolutions.

At 1280x1024 with anti-aliasing enabled in-game, we see a slightly larger drop of close to 7.5%, which then shrinks at 1600x1200 as the board's performance beings to become limited by other factors.

Let's now run this same gamut of tests using the GeForce 8800 GT, to see how this board stacks up with both executable names.

Without anti-aliasing, we see a very similar performance drop to that experience with the GeForce 8800 GTS 640MB once the executable name is changed, with moving towards a 7% drop here.

The performance drop actually becomes far less pronounced with this board at 1280x1024 with 4x anti-aliasing, down to around 3.5%, before growing hugely at 1600x1200, although our previous testing with this board has suggested that the GeForce 8800 GT becomes starved of frame buffer space of memory bandwidth at such intensive settings.

But, there you have it - While renaming the Crysis single-player demo executable fixes the image quality issues with all of NVIDIA's ForceWare 169.0x drivers, it also reduces performance by around 7% on average.  It's worth noting that these issues are also experienced in-game, rather than simply in the game's timedemo mode, although due to the slower camera movement of the player walking around rather than the fly-by used when benchmarking, it is normally far less noticeable.