## 载体 x[8:16, :]:数值检查通过 { s0: sadd.s32 s9, 16, s0 ; { s0: @p0 dma.simple [vmem:s10], [hbm:s9], length=16, dst_flag=[sflag:52] } { misc: @p0 vwait.ge [sflag:52], 16 ; { s0: @p0 dma.simple [hbm:s1], [vmem:s10], length=16, dst_flag=[sflag:52] } { misc: @p0 vwait.ge [sflag:52], 16 ; { misc: vwait.ge [sflag:45], 1 ; ## 改写后:原 DMA 之前插入 4 个 bundle,原 DMA 换成第二段 { s0: sadd.s32 s11, -13, s9 ; s1: simm.s32 s12, 5 } { s0: simm.s32 s13, 8 ; s1: sadd.s32 s14, 5, s10 } { s1: simm.s32 s15, 3 } { s0: @p0 dma.strided [vmem:s10], [hbm:s11], length=10, dst_stride=s13, src_stride=s13, elements_per_stride=s12, dst_flag=[sflag:52] } { s0: @p0 dma.strided [vmem:s14], [hbm:s9], length=6, dst_stride=s13, src_stride=s13, elements_per_stride=s15, dst_flag=[sflag:52] } 结果等于 x[3:11, :],数值检查通过