(forwarding a private email to the list, with the author's approval) Thanks to Dave Love, we now know that hwloc reports HTX devices and their location properly. Brice
Le 30/01/2012 15:26, Dave Love a écrit : > I just looked at the hwloc mail archive and saw your post about this. I > don't know if the results below are right, but I can provide more info > if it helps, e.g. with a different hwloc, or even check the MB > physically on a dead one of these. I might have some questions in > return, though :-). > > This is with hwloc 1.3 under Red Hat 5 on Supermicro H8DCE-HTE with > InfiniPath_QHT7040 (ipath0): > > Machine (7562MB) > NUMANode L#0 (P#0 3522MB) > Socket L#0 > L2 L#0 (1024KB) + L1 L#0 (64KB) + Core L#0 + PU L#0 (P#0) > L2 L#1 (1024KB) + L1 L#1 (64KB) + Core L#1 + PU L#1 (P#1) > HostBridge L#0 > PCI 10de:0053 > Block L#0 "hdc" > PCI 10de:0054 > PCI 10de:0055 > PCIBridge > PCI 1002:4752 > PCIBridge > PCI 14e4:1659 > Net L#1 "eth0" > PCIBridge > PCI 14e4:1659 > Net L#2 "eth1" > HostBridge L#4 > PCI 1fc1:000d > OpenFabrics L#3 "ipath0" > NUMANode L#1 (P#1 4040MB) > Socket L#1 > L2 L#2 (1024KB) + L1 L#2 (64KB) + Core L#2 + PU L#2 (P#2) > L2 L#3 (1024KB) + L1 L#3 (64KB) + Core L#3 + PU L#3 (P#3) > HostBridge L#5 > PCI 10de:0054 > PCI 10de:0055 > > and likewise, but Supermicro H8DSL/InfiniPath_QHT7140: > > Machine (7946MB) > NUMANode L#0 (P#0 3906MB) > Socket L#0 > L2 L#0 (1024KB) + L1 L#0 (64KB) + Core L#0 + PU L#0 (P#0) > L2 L#1 (1024KB) + L1 L#1 (64KB) + Core L#1 + PU L#1 (P#1) > HostBridge L#0 > PCIBridge > PCIBridge > PCI 14e4:1648 > Net L#0 "eth0" > PCI 14e4:1648 > Net L#1 "eth1" > PCI 1166:024b > PCI 1166:024b > PCI 1166:0214 > Block L#2 "hda" > PCI 1002:4752 > HostBridge L#3 > PCI 1fc1:000d > OpenFabrics L#3 "ipath0" > NUMANode L#1 (P#1 4040MB) + Socket L#1 > L2 L#2 (1024KB) + L1 L#2 (64KB) + Core L#2 + PU L#2 (P#2) > L2 L#3 (1024KB) + L1 L#3 (64KB) + Core L#3 + PU L#3 (P#3) > > I also have Sun x4100, which I think has HTX, but with nothing in it.
