Qemu simulate NUMA and parse SRAT
I'm trying to simulate a NUMA system with Qemu/KVM and read the memory from SRAT. I'm starting qemu with qemu-system-x86_64 -boot d -cdrom myiso.iso -m 5G -smp 4 -numa node,nodeid=0 -numa node,nodeid=1 -enable-kvm So there are two regions with two cores each and 5gb memory at the whole system. By parsing the srat, I'll find four entries at the table: Domain: 0 Addr: 0x0 Len: 0xa0000 Flags: 0x1 ====== Domain: 0 Addr: 0x100000 Len: 0x9ff00000 Flags: 0x1 ====== Domain: 1 Addr: 0xa0000000 Len: 0x20000000 Flags: 0x1 ====== Domain: 1 Addr: 0x0 Len: 0x80000000 Flags: 0x1 ====== I don't understand the the last entry. Why does he start at memory address 0?
On x86-64, is the “movnti” instruction atomic?
Where is the X86-64 ABI documented?
Makefile: determine build if build target is amd64, so that I can adapt CFLAGS=-fPIC
How to bypass a 2TB core dump file system limit?
Does a CPU cache entry contains physical or virtual address? [closed]
where is amd64 psABI? [closed]
Carry bit, GAS constraint
Does prefetching a write ever affect single core performance?
MP Specfication for AMD,how to use code to boot multiprocessors as regards to AMD
RPM is not working with i386 (rhel6) but it works with x86_64(rhel5)
How to disassemble an x86-64 .bin file?
Why does setjmp(3) not save all registers on AMD64?
why 'pushb $12' errors in gas?
How to direct gas use a specified encoding form of instructions, for example, MOV?
harmonic series with x86-64 assembly
Unable to read/write mmaped address which is returned by libpci bus under x86_64 linux