Author: Robert Hyatt
Date: 19:43:39 09/07/03
Go up one level in this thread
On September 07, 2003 at 16:39:41, Bo Persson wrote: >On September 07, 2003 at 11:47:33, Robert Hyatt wrote: > >>On September 06, 2003 at 19:41:00, Gian-Carlo Pascutto wrote: >> >>>On September 06, 2003 at 19:31:40, Michael P. Nance Sr. wrote: >>> >>>>Let Me see if I understand this fully ,and ask the following questions? Isn't >WS 2003 scalable to 64 bit? >>> >>>The operating system itself may very well be, but you cannot get the 64 bit >>>version for Opteron right now. >>> >>>>It should run both 32 and 64 bit Programs right? >>> >>>The normal version will not run 64 bit programs. >> >>How can it _Prevent_ 64 bit instructions? >> >>It can't... > >It can actually, beacuse of the extremely twisted instruction set. > >AMD has redefined some bit patterns so that they behave differently in 32 bit >and 64 bit mode. To use 64 bit registers you need some size-override prefixes >that have a different meaning in 32 bit mode. > >Just like we have 16 bit and 32 bit segments for the IA32, we need 64 bit >segments for the Opteron. That seems to require OS support. > > >Bo Persson When I first looked at this stuff, it appeared to me to be the same issue as that which ocurred with the YMP and then the C90 in the Cray series, which introduced new stuff. A program could use the stuff without the O/S knowing. IE the OS didn't understand that vector registers became twice as long, but that didn't matter with a single application running by itself. I can't imagine that it is not possible on the Opteron to finagle yourself into a 64 bit world without the O/S knowing about it, but I could certainly be wrong.
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.