Well, today my new video card arrived. It’s the eVGA 6800GS CO (397) with 256MB of DDR3 ram. It’s replacing my ASUS 5900FX card with 128M of ram, and I’m hoping for a pretty significant power boost from this upgrade. In case you haven’t noticed, I’m an NVidia Fanboy. My system is AGP, so PCI Express wasn’t an option. Although there is the new BFG 7800 AGP card, it’s out of my price range.

So before I popped it in, I downloaded Futuremark‘s 3dMark 2006 and tried it out. The performance was abysmal on my 5900FX, less than 1fps on the first test. I quickly aborted it, and figured that since even my new card is last year’s model, I should give it a break and use last year’s test. An hour later I had 3DMark 2005 setup, and commenced with the testing. For those of you interested, I did all these tests immediately after a fresh reboot. I have an AMD AthlonXP 3200 and 1Gig of system memory. I ran all my tests with a “standard operating system”, meaning that I had my antivirus running and all my system try icons idling away happily. I’m running the official 81.98 Drivers from NVidia, and Windows XP. I had hoped to use the same drivers on the new card, but they wouldn’t recognize my card. The install CD that comes with the card has the 82.12 drivers, so I wound up using those for the 6800 tests.

3d Mark 2006
3dMark 2005 Total SM2 CPU SM3/HDR
ASUS NVidia 5900FX 1040
 324
 151 795 N/A
eVGA 6800GS CO 4828
 2279
 975 800 894

The difference was immediately evident, as the tests went from “Seconds per frame” to “Frames per second” on almost all the tests. 3DMark06 had the most visible improvements, as the HDR & SM3 tests weren’t even run on the 5900 card. The CPU results stayed the same (as they should) so you can see the different is pretty much exclusively in the graphics. Six months ago I wouldn’t have believed you if you said that a simple graphics card upgrade could give you this kind of improvement. As predicted by the Video Card Stability Test & Benchmark Results, I got the 3x-5x improvement easily.

But how do these numbers translate into “real world” performance. Well, I still had FEAR laying around and it was pretty sluggish before. The default settings for my hardware were to set the System to “Medium”, and the Graphics to “Custom”. What is “Custom” you ask? For me, 1 notch below “Low”: 640×480, No FSAA, No Shadows at all, No Volumetric Lighting, and everything else set to Minimum except Water Detail which is for some reason set to Medium. For this test I loaded up FEAR and let it autodetect, and then used the “Test Settings” option to get the Minimum, Average, and Maximum framerates for the test sequence. I then switched to Low, Medium, and High and ran the test sequence again. (There is another setting “Maximum”, but seeing how poorly my 5900 did in High I decided to pass on that).

Low Medium High Default
5900 61/131/249 11/18/33 4/9/21 33/52/80
6800 89/162/298 46/93/222 34/58/121 36/59/121

The default settings for the 6800 were 1024×768 with everything set to “Maximum”, except FSAA and Soft Shadows disabled. Shadow detail was still on & set to Maximum, but the soft edges were turned off. Again, six months ago I would never have believed this kind of result. Effectively, using the default settings I get the exact same framerate as before, but now I get the benefits of double the resolution, shadows, more effects, more details, and more of everything that makes it beautiful to stare at.

So, I’m happy. No, happy isn’t quite the right word. Ecstatic is closer. For $200 I breathed new life into my aging system and easily bought me another year, hopefully two, of gaming goodness.
[tag:nvidia] [tag:geforce] [tag:benchmark] [tag:computer]

Be Sociable, Share!