On Sun, 2006-06-11 at 22:28 +0200, Anders Johansson wrote:
It looks to me like a variation of the standard Intel segmentation thing
It's _not_. Page tables and windowing is very, very different. Case-in-point: AMD 52-bit PAE uses no less than a 7-layer logic to provide 20-bit Real86/Virtual86, 32-bit Protected386 and 36-bit Processor Address Extension (PAE) modes. It's far, far more complicated and involved than what I've seen discussed here. Regardless of the "base API model" used, the actual "application binary interface" (ABI) of the unliked as well as linked object code is the kicker. All _current_ x86 implementations, including x86-64, have _never_ broken the _hard_ 48-bit/256TiB address limitation of the _legacy_ i386 16-bit segment + 32-bit offset registers -- although, actually, it is really a byproduct of the i486 TLB (translation lookaside buffer). This _hard_ 48-bit limitation is at the foundation of why AMD calls the _current_ x86-64 "Long Mode" (16-bit segment + 32-bit offset) and uses a 52-bit PAE approach to be compatible with the i686 (Pentium Pro) PAE (which is only 32-bit). And it will probably be _only_ solved by virtualization (which will also solve some serious portability issues with Win32 as well -- I'm sure Microsoft is kissing AMD/Intel butt right now with the nexgen of 'Visor solutions). -- Bryan P.S. Don't get me started on Intel's AGTL+ "bus" versus AMD EV6 "crossbar" and HyperTransport "partial mesh" when it comes to cache and I/O coherency -- let alone AMD offering a _real_ I/O MMU and Intel not. And that also leads back into why AGP/PCIe software/driver support is so proprietary, because of Intel's _lack_ of addressing I/O coherency in hardware. -- Bryan J. Smith Professional, technical annoyance mailto:b.j.smith@ieee.org http://thebs413.blogspot.com ------------------------------------------------------- Illegal Immigration = "Representation Without Taxation" -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com