To be quite honest, it wouldn't affect performance. With the OpenGL versions you have the intensive act of simply rendering the damn level (I don't know what will be used, but the only place I've found where my Intel 3000HD graphics outperforms my Radeon 6700M is rendering a .vxl in a STATIC_DRAW VBO, and that is a LOT of cubes) and with the VOXLAP versions (e.g. 0.75) you have the font renderer, and if it weren't for that you'd still have the whole level to draw, depth buffering and all.
After all, each kv6 cube in VOXLAP is rendered as a rectangle which is done something like this:
- Code: Select all
mov ecx, x_position
mov edx, y_position
mov edi, [screen_y_table+edx]
mov eax, block_colour
mov edx, height
lea edi, [edi+ecx*4]
lp_drawbox:
mov ecx, width
rep stosd
add edi, screen_width_minus_width
dec edx
jnz lp_drawbox
(HINT: rep stos is, afaik, the fastest way to fill stretches of memory on an x86 processor)