Using Drivers Also for GPU access

Hi,
I've been reading up on ATI ISAs and learning things about how to make shaders. But it might be easier to use the format required to use largely distributed drivers. U guys think if I supplied all of the pixel coordinates and such in code, there is a way to learn how to use some of these common drivers very efficiently? Here's some of what I've been doing from the Assembly syntax documents:

Code:

ShaderType = 3
TargetChip = b
NumClauseTemps = 4

00 ALU: ADDR(1) CNT(5) 
       0       x: MOV      T1, MEM*
    1       x: MOV      R0.x, MEM*
          y: MOV      R0.y, MEM*
          z: MOV      R0.z, MEM*
          w: MOV      R0.w, 0
01 VTX: ADDR(2) CNT(4)
       2 VFETCH R1.x, R0.x
 3 VFETCH R1.y, R0.y
 4 VFETCH R1.z, R0.z
 5 VFETCH R1.w, R0.w
02 TEX: ADDR(3) CNT(3)
   6 FETCH4 R2.x, R0.x, T1
     7 FETCH4 R2.y, R0.y, T1
     8 FETCH4 R2.z, R0.y, T1
03 EXP: POS0, R1.x
04 EXP: POS1, R1.y
05 EXP: POS2, R1.z
06 EXP: POS3, R1.w
07 EXP: POS4, R2.x
08 EXP: POS5, R2.y
09 EXP: POS6, R2.z
END_OF_PROGRAM

ShaderType = 1

00 ALU: ADDR(1) CNT(5) 
    0       x: MOV      T1, MEM*
    1       x: MOV      R0.x, MEM*
          y: MOV      R0.y, MEM*
          z: MOV      R0.z, MEM*
          w: MOV      R0.w, 0
01 EXP: PIX0, R0.x
02 EXP: PIX1, R0.y
03 EXP: PIX2, R0.z
04 EXP: PIX3, R0.w
END_OF_PROGRAM

Trying to get down the memory reads and writes which I think are broken up between scratch buffers, reduction buffers, and scatter reads and writes.

What would be a good way to learn how to handle a graphics driver? Writing some shaders for a bunch of cards would take long. Ok.

Cheers,
Denat