On Wed, 2011-06-22 at 14:13 +0800, Chia-Hung Lin wrote: > Hi, > > I don't know if this post is appropriately for this board. (If not, > which forum should I post to?) > > I am wondering if my machine is configured right. > My machine equipped with two Intel Xeon E5640 CPUs (each has 4 cores). > The red hat version is Red Hat Enterprise 5.5 and linux kernel is > 2.6.18-194.el5 > My question is when I check /sys/devices/system/node, there is only > one node0 directory. > The information displayed by "numactl --show" is > policy: default > preferred node: current > physcpubind: 0 1 2 3 4 5 6 7 > cpubind: 0 > nodebind: 0 > membind: 0 > > Why there is only one node instead of two? Is there any thing wrong > with my machine? > Also, I found a message in dmesg : > No NUMA configuration found > Faking a node at 0000000000000000-0000000318000000 > > What should be a NUMA configuration look like? > Another thing I think strange is the BIOS setting. Its default is SMP. David: in SMP mode, the bios will interleave/stripe the memory of the 2 sockets [assuming you have memory on both sockets?] into a single "node" and will [I think] omit the ACPI numa topology table [SRAT]. This is consistent with the "No NUMA configuration found" message you're seeing. > When I try to change it to NUMA, I can only see one CPU after re-booting. > Is it normal? (the machine model is Dell T5500) Don't know about that. Doesn't sound right. Do you still see "No NUMA configuration..."? What does "numactl --hardware" show in this case? You might try booting with 'debug' [instead of 'quiet' -- default in RHEL 5.5] and see whether the console log or dmesg file give any clues. > It will be very helpful if you can give me some direction. > > Thanks a lot. Regards, Lee -- To unsubscribe from this list: send the line "unsubscribe linux-numa" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
