fMat4fGPUMul proc uses esi edi lpdestination:dword,lpMat4x41:dword,lpMat4x42:dword
invoke glPushMatrix
invoke glMatrixMode,GL_MODELVIEW
; invoke glLoadIdentity
invoke glLoadMatrixf,lpMat4x41
invoke glMultMatrixf,lpMat4x42
invoke glGetFloatv,GL_MODELVIEW_MATRIX,lpdestination
invoke glPopMatrix
ret
fMat4fGPUMul endp
Im Impressed with the capability of my SiS card on multiplying a matrix. It took 3 seconds for calculating 100 millions matrix. Compared to my FPU version of matrix multiplication wich is took 20 minutes, I should reconsider that something was wrong on the software, not the hardware. For a system that clocked at 250 Mhz, it is far superior than a dual core 1.86 Ghz processor.
Now, I think we need a new term for a processor quality. How can a dual core muli hundreds Mega Hertz can be so slow compared to a GPU which is not even a quarter from it.