Basically, the current state is that softgpu will either completely fail to render anything properly or that it will render the game close to perfect, which I'd say is a sign of a solid implementation of most features but there are some critical bugs that I haven't debugged, yet. It's obviously painfully slow still, and I likely won't be the one who optimizes it to fullspeed.
The GLES backend has a number of ridiculous limitations which can't be solved at all or only with a huge effort and performance penalty. With softgpu, we basically get free fixes for any bugs that arise from this.
The GLES backend has a number of ridiculous limitations which can't be solved at all or only with a huge effort and performance penalty. With softgpu, we basically get free fixes for any bugs that arise from this.