Dateien
Mustafa Abduljabbar b33b5755f6 Support gfx950 in topo_expl and resolve dependency on FMT (#1829)
* Support gfx950 in topo_expl

* Fix dependencies and fetch fmt from sources

* Remove third_party folder in make clean

* Add empty target when fmt is found

* Add MI350 example

* Update README.md

---------

Co-authored-by: isaki001 <ioannissakiotis@gmail.com>

[ROCm/rccl commit: dfad51e3c9]
2025-08-26 10:11:38 -04:00

168 Zeilen
12 KiB
XML

<system version="2">
<cpu host_hash="0x1b2f53a636e182bb" numaid="0" affinity="00000000,00000000,00000000,ffffffff,ffffffff,ffffffff,00000000,00000000,00000000,ffffffff,ffffffff,ffffffff" arch="x86_64" vendor="AuthenticAMD" familyid="191" modelid="2">
<pci busid="0000:01:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc03" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:03:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re0" dev="0" latency="0" speed="400000" port="1" guid="0x28303efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:06:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:08:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="0" sm="256" gcn="gfx950" arch="38911" rank="0" gdr="1">
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:11:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc03" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:13:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re1" dev="1" latency="0" speed="400000" port="1" guid="0x18483efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:16:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:18:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="1" sm="256" gcn="gfx950" arch="38911" rank="1" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:61:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc02" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:63:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re2" dev="2" latency="0" speed="400000" port="1" guid="0x78333efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:66:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:68:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="2" sm="256" gcn="gfx950" arch="38911" rank="2" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:71:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc02" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:73:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re3" dev="3" latency="0" speed="400000" port="1" guid="0x981e3efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:76:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:78:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="3" sm="256" gcn="gfx950" arch="38911" rank="3" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
</cpu>
<cpu host_hash="0x1b2f53a636e182bb" numaid="1" affinity="ffffffff,ffffffff,ffffffff,00000000,00000000,00000000,ffffffff,ffffffff,ffffffff,00000000,00000000,00000000" arch="x86_64" vendor="AuthenticAMD" familyid="191" modelid="2">
<pci busid="0000:81:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc01" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:83:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re4" dev="4" latency="0" speed="400000" port="1" guid="0x80963ffeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:86:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:88:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="4" sm="256" gcn="gfx950" arch="38911" rank="4" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:91:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc01" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:93:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re5" dev="5" latency="0" speed="400000" port="1" guid="0x681e3efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:96:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:98:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="5" sm="256" gcn="gfx950" arch="38911" rank="5" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:e1:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc00" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:e3:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re6" dev="6" latency="0" speed="400000" port="1" guid="0xd0373efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:e6:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:e8:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="6" sm="256" gcn="gfx950" arch="38911" rank="6" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:f9:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
<pci busid="0000:f1:00.0" class="0x060400" vendor="0x1000" device="0xc030" subsystem_vendor="0x1d9b" subsystem_device="0xcc00" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:f3:00.0" class="0x020000" vendor="0x14e4" device="0x1760" subsystem_vendor="0x14e4" subsystem_device="0xf323" link_speed="32.0 GT/s PCIe" link_width="16">
<nic>
<net name="bnxt_re7" dev="7" latency="0" speed="400000" port="1" guid="0xe84a3efeffe604d6" maxconn="131073" gdr="1"/>
</nic>
</pci>
<pci busid="0000:f7:00.0" class="0x060400" vendor="0x1022" device="0x1500" subsystem_vendor="0x1022" subsystem_device="0x1500" link_speed="32.0 GT/s PCIe" link_width="16">
<pci busid="0000:f9:00.0" class="0x120000" vendor="0x1002" device="0x75a0" subsystem_vendor="0x1002" subsystem_device="0x75a0" link_speed="32.0 GT/s PCIe" link_width="16">
<gpu dev="7" sm="256" gcn="gfx950" arch="38911" rank="7" gdr="1">
<xgmi target="0000:08:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:18:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:68:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:78:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:88:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:98:00.0" count="1" tclass="0x120000"/>
<xgmi target="0000:e8:00.0" count="1" tclass="0x120000"/>
</gpu>
</pci>
</pci>
</pci>
</cpu>
</system>