Ray tracing by CPU multi-threading and GPU

I wrote a program which does ray tracing by CPU multi-threading and GPGPU. For GPGPU programming, I used NVidia CUDA C. Here are some results.

• Spec
  • CPU : Intel Core i7 CPU
  • GPU : Quadro FX 580 ( 32 CUDA cores )
  • Ray tracing buffer size : 1024 * 1024 pixel

• Results
  • 1 CPU Thread : 1x
  • 2 CPU Thread : 1.18x
  • 4 CPU Thread : 1.72x
  • 8 CPU Thread : 1.96x
  • 1024 * 1024 GPGPU Thread : 10x

From these results, it seems that my program doesn't scale good with multi-core. ( Sorry about it. ) But I think the program can be optimized better.