| Card | Score |
| Hercules 3D Prophet III | 5707 |
| Leadtek GF3 | 5700 |
| 32MB GTS: | 3280 |
3DMark2001, of course, is an old standby in revieiwing and benchmarking video cards. And particularly in cases such as this, where all tested cards use technology from a single family, it serves as a good means of measuring relative performance.
It's no surprise that the Hercules and Leadtek GF3's are performing very similarly, as they're both based so closely on the reference design. Equally unsurprising is the fact that the GTS card we threw in there to be a punching bag can't even come to close to matching either GeForce3 video card.
| Card | 640x480 | 800x600 | 1024x768 | 1280x1024 | 1600x1200 |
| Hercules 3D Prophet III | 193.5 | 187.8 | 170.5 | 118.4 | 80.4 |
| Leadtek GF3 | 193.8 | 188 | 170.5 | 118.3 | 80.4 |
| Leadtek 32 MB GTS | 181.5 | 146.2 | 93.6 | 59.5 | 39.7 |
| Card | 640x480 | 800x600 | 1024x768 | 1280x1024 | 1600x1200 |
| Hercules 3D Prophet III | 191.8 | 186.4 | 161.8 | 110.4 | 76.4 |
| Leadtek GF3 | 192.4 | 186.3 | 161.9 | 110.5 | 76.4 |
| Leadtek 32 MB GTS | 180.6 | 141.6 | 89.3 | 56.8 | 37.3 |
In our tests using Quake 3 Arena's Timedemo 2, we see the same pattern in both the 16-bit and 32-bit results as we saw in our 3DMark2001 test. Both GeForce 3 boards offer comparable performance, with the GTS trailing way behind. If we were to run each test on each board 100 times, we'd probably end up with the exact same average scores for both of the GeForce3 video cards.
Having already confirmed the obvious-- that the GF3 is faster than the GTS, we set that previous-generation 32 MB board aside in the rest of our testing.