Designing An Efficient Kernel-level and User-level Hybrid Approach for MPI Intra-node Communication on Multi-core Systems L. Chai, P. Lai, H. Jin, D. Panda International Conference on Parallel Processing 2008, Sep 2008.