作者: Zoltán Ádám Mann
DOI: 10.5072/ZENODO.29348
关键词:
摘要: With the recent development of high-performance graphical processing units (GPUs), capable performing general-purpose computation (GPGPU: on GPU), a new platform is emerging. It consists central unit (CPU), which very fast in sequential execution, and GPU, exhibits high degree parallelism thus performance certain types computations. Optimally leveraging advantages this challenging practice. We spotlight analogy between GPGPU hardware/software co-design (HSCD), more mature design paradigm, to derive process for GPGPU. This process, with appropriate tool support automation, will ease significantly. Identifying challenges associated establishing can serve as roadmap future field.