Hello, Jiri Hladky, le Tue 04 Jan 2011 23:52:21 +0100, a écrit : > I got a crash course on s390.
(Not sure about the english meaning here: it does not mean hwloc crashed, right?) > They have an access to mainfraim with 18 CPUs and 128 GB of RAM. However, the > mainframe is usually divided into the smaller partitions ( LPAR) on which > virtual machines are running. That's what I've understood from echoes from Debian s390 porters indeed. > In this particular case they have tested hwloc on the virtual machine running > on partition with 2GB of RAM and 4 CPUs of which only 1 was online. This is > certainly not a typical use-case for hwloc. Nevertheless, great news is that > hwloc is running just fine on this box! :-) Indeed: > Machine (P#0 local=2048132KB total=2048132KB Backend=Linux) > Socket L#0 > Core L#0 (P#0) > PU L#0 (P#0) > depth 0: 1 Machine (type #1) > depth 1: 1 Socket (type #3) > depth 2: 1 Core (type #5) > depth 3: 1 PU (type #6) > 63 processors not represented in topology: 0xffffffff,0xfffffffe > 3 processors offline: 0x0000000e > 3 processors allowed but not online: 0x0000000e So sockets and cores were detected, as well as the off-line processors, that's good! (cache are apparently not in the tar, and I can't find code about them in linux/arch/s390 anyway) Samuel
