I use these settings for maximum speed  -aggression=8 -gpugrid=256 -gputhreads=512

Unfortunately we will not see  optimized CUDA miner as OpenCL