My recollection is that the switchover to referring to them as a "GPU" wasn't integrating 2D and 3D in the same card, but the point where we offloaded MUCH more

My recollection is that the switchover to referring to them as a "GPU" wasn't integrating 2D and 3D in the same card, but the point where we offloaded MUCH more computation to the graphics card itself. So we're talking specifically about when NVidia launched the Geforce 256 - a couple of generations after the TNT