On Wed, 2008-12-03 at 19:34 +0100, Giuseppe Falsetti wrote: > Hollis Blanchard ha scritto: > > On Mon, 2008-12-01 at 19:30 +0100, Giuseppe Falsetti wrote: > > > >> Hi Hollis, > >> no I don't have a 440, I have only many 970 :-| > >> So I can't hope that KVM will support 970 cpus? Qemu already support > >> this, why for KVM is so difficoult? > >> > > > > Implementing KVM for 970 would require emulating the 970 core in the > > kernel (probably the MMU would be the most difficult). I won't say this > > is easier or harder than qemu, but it's hard in different ways. > > > Sorry, but, for example if I want to emulate an x86 system on a 970 > Host, I need to emulate an x86 core and not the 970. > Or not? Obviously the emulation is slowly with and without 970 features > optimizations. KVM (and hypervisors in general) only execute guests of the same instruction set as the host. The reason this is faster than pure emulation is that the majority of instructions execute natively on the processor, with only infrequent intervention required by software (the hypervisor). So KVM won't let you run an x86 system on a 970 processor; that's what qemu is for. Unlike KVM, qemu actually translates the x86 instructions into PowerPC ones. -- Hollis Blanchard IBM Linux Technology Center -- To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html