The processor is being virtualized instead of emulated by Parallels. For the non-geek this means that the processor has hardware level support for virtual machines, which Parallels is using. This lets the VM run directly on the real hardware processor with some limitations so that it can not take over the real machine. I would expect that Parallels is feeding VM's fake information about stepping/model and family so that the VM does not try to use energy-saving features of the processor that it isn't allowed to us in virtulized mode. I would also expect that Parallels lets the VM know about all the extra registers and instructions that are in the processor that is IS allowed to use, hence the funky combination of flags.
For core1 processors, I would go with the core solo version of whatever processor is actually in the Mac you are using.
This list may be of use to you. Its a conversion from the "marketing speak" of Intel to their actual model numbers.
Here is that same list for core2 processors. I'm not sure what to do with these, as there is no core solo versions.
For Mac Pros, tell the compilers that you are running a single Xeon of whatever chip is in there.
Of course, if (maybe when) Parallels get multiprocessor support within a VM working then this will all change.
Important: If you intend to move your VM from one Mac to another then next make sure to compile for generic i686. (i.e. Pentium 4) I could see some program compiled for a Xeon Macpro that wouldnt run on a Core 1 Macbook.
Of course, I could be wrong. My work is with the OS side of things. Apply salt as needed.
