—A triangle setup engine (TSE) generates setup parameters for rasterization in 3D graphics hardware. Since the TSE must be flexible for various embedded systems, a programmable TSE is proposed especially optimized for a division-free rasterizer and it shows performance improvement up to 4.76 times derived by 3-way SIMD architecture with special instructions for matrix-vector multiplication. It is integrated with an optimized format converter in 0.13um CMOS technology. Its gate count is 210kgates and the power consumption is 49.1mW at 166MHz operation frequency.