Arquivos
rocm-systems/tools/msccl-algorithms/alltoall-8n-0-9kb.xml
T
Pedram Alizadeh c2fc1d6809 msccl algorithms tuning for alltoall on MI300 (#1120)
Co-authored-by: PedramAlizadeh <amd@pmohamma.com>
2024-03-21 20:35:29 -04:00

381 linhas
20 KiB
XML

<!-- Copyright (c) Microsoft Corporation. -->
<!-- Licensed under the MIT License. -->
<algo name="alltoall_llm" proto="LL" nchannels="1" nchunksperloop="8" ngpus="8" coll="alltoall" inplace="0" outofplace="1" minBytes="0" maxBytes="9216">
<gpu id="0" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="0" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="0" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="1" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="1" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="1" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="2" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="2" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="2" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="3" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="3" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="3" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="4" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="4" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="4" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="5" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="5" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="5" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="6" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="6" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="7" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="6" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="7" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="7" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
<gpu id="7" i_chunks="8" o_chunks="8" s_chunks="0">
<tb id="0" send="-1" recv="-1" chan="0">
<step s="0" type="cpy" srcbuf="i" srcoff="7" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="1" send="-1" recv="0" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="0" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="2" send="-1" recv="1" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="1" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="3" send="-1" recv="2" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="2" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="4" send="-1" recv="3" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="3" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="5" send="-1" recv="4" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="4" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="6" send="-1" recv="5" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="5" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="7" send="-1" recv="6" chan="0">
<step s="0" type="r" srcbuf="i" srcoff="7" dstbuf="o" dstoff="6" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="8" send="0" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="0" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="9" send="1" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="1" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="10" send="2" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="2" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="11" send="3" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="3" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="12" send="4" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="4" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="13" send="5" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="5" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
<tb id="14" send="6" recv="-1" chan="0">
<step s="0" type="s" srcbuf="i" srcoff="6" dstbuf="o" dstoff="7" cnt="1" depid="-1" deps="-1" hasdep="0"/>
</tb>
</gpu>
</algo>