发明名称 Providing data to registers between execution stages
摘要 In some implementations, a processor is provided having a buffer to store one or more instructions, a decoder configured to decode the one or more instructions and generate one or more decoded instructions, a processor register file to store one or more operands, and a plurality of execution units. Each execution unit includes a plurality of execution stages and a plurality of registers. The plurality of execution stages is configured to execute one or more decoded instructions using the one or more operands. The plurality of registers is positioned between the plurality of execution stages to latch data between the plurality of execution stages.
申请公布号 US8909903(B1) 申请公布日期 2014.12.09
申请号 US201113311184 申请日期 2011.12.05
申请人 Marvell International Ltd. 发明人 Chen Hong-Yi;Tjeng Jensen
分类号 G06F9/30 主分类号 G06F9/30
代理机构 代理人
主权项 1. A processing device comprising: a buffer to store one or more instructions; a decoder configured to decode the one or more instructions and generate one or more decoded instructions; a processor register file to store one or more operands; and a plurality of execution units, each of the plurality of execution units having: a plurality of execution stages configured to execute the one or more decoded instructions using the one or more operands, anda plurality of registers to latch data between the plurality of execution stages, the plurality of registers positioned between the plurality of execution stages to latch data between the plurality of execution stages, wherein a register of the plurality of registers positioned between a pair of execution stages of the plurality of execution stages is configured to provide data directly to a preceding register of the plurality of registers positioned between a different pair of execution stages of the plurality of execution stages.
地址 Hamilton BM