Time　consistency　of　each　node　and　each　core　is　one　of　the　kernel　problems　that　must　be　addressed　to　promote　the　development　of　many-core　cluster　system　that　keeps　the　distributed　system　in　step,　and　it　is　the　basis　of　performance　evaluation.　In　this　study,　a　new　model　of　hybrid　time　synchronization　architecture　was　established　to　improve　the　precision　of　internal　clock　synchronization　of　a　many-core　cluster　and　the　accuracy　of　time.　First,　some　clock　synchronization　method　was　compared　for　clock　synchronization　model　decision.　Second,　a　high　precision　time　acquire　method　based　on　was　explored.　Third,　a　hybrid　clock　synchronization　and　time　acquire　architecture　on　cluster　system　was　designed　to　enhance　the　time　precision　and　internal　clock　consistency.　Finally,　the　architecture　was　implemented　on　a　Zynq　SoC　and　Epiphany　many-core　processor　combined　platform.　An　experiment　was　conducted　to　evaluate　the　precision　and　accuracy　of　this　architecture.　The　result　shows　that　the　hybrid　architecture　achieved　a　higher　level　time　precision　and　internal　clock　consistency　than　common　clock　synchronization　method　like　NTP.　And　result　demonstrates　the　promising　application　of　the　proposed　architecture　in　study　many-core　cluster　system　and　in　supplying　reliable　time　reference　for　the　evaluation　of　distributed　system　performance.