Tech-Hounds.com

NVIDIA Forceware Revisited

Graphics card drivers
PCI Express x16 graphics cards
Graphics card NVIDIA GeForce 6600GT

After doing our round up of GeForce 6600GT, we came to the conclusion that there's a lot of homework to be done by NVIDIA's driver developers. Since then, NVIDIA has release the latest Forceware version, the 78.01. We recently received the Gigabyte GeForce 6600GT Silent Pipe (which you can read the review here), but we used Forceware 77.72 for evaluation purposes in that review. Of course, we're aching to see whether or not NVIDIA fixed the many problems we encountered in Forceware 78.01. What better way to test it but by using additional games? We spent some time with everyone's favorite these past months - Battlefield 2.



We only spent a short time with Battlefield 2, so we won't be using it in testing (for now). This is more of a quick and dirty test, we choose the above settings since it's still very playable on the GeForce 6600GT. You'll also notice that the 100 fps cap is still active. We really like the custom quality options, since it allows us to troubleshoot graphical glitches and problems. Unlike Battlefield 1942 and Battlefield: Vietnam, Battlefield 2 uses shaders extensively and each time we change video settings, the game will automatically 'optimize' the shaders for best performance. Even with fixed settings, the game will optimize the shaders if you change graphics card drivers. We ran test with three drivers - Forceware 66.93, 77.72 and 78.01. There are some graphical differences as you can see below.





While the images are not exactly the same frame, you can clearly see quality differences. We didn't use a Battle Record, opting for a screenshot in game, that's why they don't exactly match up. With Forceware 66.93, the (dynamic) shadows will render incorrectly, resulting in the artifact you can see clearly on the first image. You can however, turn off the lighting (although not completely) to avoid this bug. Forceware 77.72 and 78.01 renders the shadows correctly.





However, Forceware 77.72 has a graphical bug that renders some parts (the sandbags) a little 'funky'. It's even more annoying to look at if you're moving the camera. This artifact is not present on Forceware 66.93 and 78.01. So, at least compatibility wise, the latest Forceware 78.01 is the best one to use with Battlefield 2. How about performance?

	Forceware 66.93	Forceware 77.72	Forceware 78.01
Minimum fps	52.25	50.85	57
Average fps	69.18	70.39	72.30
Maximum fps	82.25	84.57	85

These are average results from five runs, with 5 to 10 % variance between each run. As you can see, the maximum fps never reach the 100 fps with the settings we used on the GeForce 6600GT. However, long extensive gameplay does indicate we can see much lower fps than these (especially with lots of smokes and explosions), down to the 25 to 30 fps range. It's clear that upping the settings will tax the graphics card beyond its capability. You really have to use at least a GeForce 6800GT if you want to use the 'High' settings.

Our incursion with Battlefield 2 shows that again you need newer drivers if you want to play newer games on NVIDIA hardware. Sadly, with past driver versions, this mean sacrificing some performance and quality with older games. But it looks like things are getting better with Forceware 78.01. Let's see whether or not that's true for all games, new and old. Forceware 66.93 is very similar to 66.81 and you can look at the quality differences in our past article (which you can read here and here). There's no need to post the 78.01 screenshots, since they're basically the same as with 77.72, so let's discuss performance.

Performance

Performance testing are done with either the built-in tools in the game or by playing a replay or a cut scene that's close to actual gameplay and measuring the frame rate per second with FRAPS. Just like Battlefield 2, in Brothers in Arms we actually played a level in a certain, repeatable way, so there are some variances with the results. We ran the game three times and average the results, for each setting and resolution. Even without averaging, the raw scores are very close to one another, usually within 5 %. We haven't done extensive testing with Battlefield 2 yet, so no Battlefield 2 scores (for now).

Our test setup
AMD Athlon 64 3000+ socket 939
2 x 256 MB Kingston KVR 3-3-3 PC3200 DDR-SDRAM
MSI K8N SLI Platinum Socket 939 PCI Express motherboard
Gigabyte GeForce 6600GT Silent Pipe graphics card
Maxtor DiamondMaxPlus9 80 GBs Serial ATA 8 MB buffer
ASUS E-616 DVD-ROM
450 watts ATX power supply

Windows XP Professional with Service Pack 2 installed
NVIDIA NForce 4 6.66 reference driver
Creative SoundBlaster Live! 24 bit 5.12.1.512 driver.
DirectX 9.0c
all respected games used for benchmarks have been updated to their latest, final builds

The results:

Looks like we're off to a good start. Performance with the latest Forceware 78.01 is basically the same as with Forceware 77.72. Forceware 66.93 is slightly faster (particularly on 1600 x 1200 with AA and AF), but for all intent and purposes, the three drivers offer the same level of performance.


Looks like the shader 'bug' we've seen on the Forceware 71.22 up to 77.72 is still present in the newest drivers. Unlike previous test results, we now see quite a significant difference between the 66.93 and 77.72 / 78.01. Instead of 1 or 2 fps, it's now 4 to 12 fps. Mind you, these results came from the Gigabyte GeForce 6600GT Silent Pipe which has a slightly higher clock than the reference specification. This might explain the performance difference between these and previous results, but it doesn't explain why we're seeing this much performance difference by using a different drivers (particularly since the image output is not the same).


Looks like NVIDIA still have no fix for the shadow rendering bug in this game. With shadows set to 'Full', we're still seeing very low performance. So, it looks like you still have to sacrifice the shadows in this game if you want a fluid, entertaining gameplay.


Thankfully, nothing's wrong in Nascar 2003. We're still seeing 2 to 3 fps difference, with the newer drivers being slower. But you don't have to worry, the difference is not noticeable. Of course, there are some image quality compromises, but everything should clear up when AA and AF are enabled.


Like we said before, newer drivers targets newer games. While we're not seeing the same increase going from 77.72 to 78.01 as going from 66.93 to 77.72, we're still seeing a little performance boost (around 2 to 3 fps) across the board. Of course, you'd be more happy to know that the 78.01 also will not crash you system when you quit the game (or alt-tabbing).


Performance in Homeworld 2 are identical between Forceware 77.72 and 78.01. While that might seem good, it's actually not. Forceware 66.93 is still the fastest driver. Quite surprising, since Homeworld 2 is an OpenGL game, where NVIDIA should excel. Thankfully, patching the game to the latest version now gives you real anti aliasing and anisotropic filtering with the game (regardless of driver versions).


Just like Nascar 2003 and Homeworld 2, SW: KOTOR doesn't get any noticeable performance improvement with newer drivers. You're getting the same level of performance with both Forceware 77.72 and 78.01.

Conclusion:

Driver bugs can be very annoying, both for developers and users alike. Squashing them out takes a lot of time and effort. With the number of games out there, it just get harder It's perfectly understandable driver developers target the newest, popular games first - Battlefield 2 is a perfect example of this. However, this does not excuse the (re)introduction of old or new bugs on old games. Just like with Forceware 77.72, we're still seeing quality image problems and significant performance slowdown with Forceware 78.01 compared to Forceware 66.93.

For the time being, NVIDIA users must still make some compromises on a game to game basis. We'll be updating our driver choice of testing to Forceware 78.01, however for testing with F1 Career Challenge and Splinter Cel, we'll fall back to Forceware 66.93. Again, the reason for this is that this should give us a clearer view, sans driver bugs (and shader replacement), of what the GeForce series can do. Hopefully, within the next week or so, we will have another graphics card round up, this time with more new games (at least Battlefield 2).

Go to top