High Performance Matrix Inversion on a Multi-core Platform with Several GPUs2011-02-010 PRE AI DOI:10.1109/pdp.2011.66原文链接原文求助分享收藏摘要 En