Most likely it was a Glide Wrapper that emulated Glide in Software.Are you sure about that? I was under the impression that that wasn't the case - normally you end up having to use a glide wrapper.
The biggest problem for 3dfx was that they did not have T&L, they thougth 16bit was good enough and that they wanted to make all cards themselves causing huge delays so by the time the V5 5500 was released it was slower than the competition (and more expensive)