Glxgears is not a real benchmark. The results you get are not really comparable to other systems (it are just plain polygons instead and no textures). Performance depends on color depth, resolution...
Compare performance of real games (quake3, rtcw, ut2003 ..).
And don't compare a GF4MX 440 with a GF4 Ti (likely your friend has such a geforce4). A Geforce4 Ti is a lot faster than a MX. Further a Geforce2 GTS is not much slower than a GF4MX440. In some cases it might even be faster. (4 pipelines vs 2 pipelines)