Author: Robert Hyatt
Date: 10:27:19 01/17/03
Go up one level in this thread
On January 17, 2003 at 11:13:42, Matt Taylor wrote: >On January 17, 2003 at 03:46:10, Bas Hamstra wrote: > >>On January 17, 2003 at 01:01:32, Matt Taylor wrote: >> >>>On January 16, 2003 at 00:57:20, Walter Faxon wrote: >>> >>>>On January 15, 2003 at 02:24:42, Matt Taylor wrote: >>>> >>>><snip> >>>>> >>>>>MMX does hold one card though that seems to be mocking me. MMX has 8 64-bit >>>>>registers. Integer has 8 32-bit (or 4 64-bit) registers. Hmm...I really hate >>>>>doing this because it's -ugly-, but I think I'm going to use the stack pointer >>>>>as an additional GPR. I will need it. >>>>> >>>>>-Matt >>>> >>>> >>>>Surely you mean the base pointer (ebp) ?? >>>> >>>>-- Walter >>> >>>No, I'll need the base pointer too. It's legal to use the stack pointer, just >>>ugly. Intel already thought of this -- different privilege levels use different >>>stacks so an exception in a user program can't kill the OS. >> >>I guess this will suspend interrupts during the time the routine runs? Yes, why >>not. You will get famous if you find a way to outperform rotated bitboards, you >>know. Gerd was already close, he is a real pioneer. >> >>Bas. > >Nope. There is no way to do that unless you convince the operating system to do >it for you. Debatable as to whether or not that's a good thing... > >Cleanly outperforming rotated bitboards will be quite difficult. That is not my >goal right now; it's too big of a project. We'll see. > >-Matt The way to whip rotated bitboards is to get the hardware people to define a 64 bit register for occupied squares. Plus three more 64 bit registers that are simply rerouted bits from the first register, ie they do the rotating on the fly by simply moving the bits to where they are needed, at zero cost. That would be why a real hardware bitboard program would/could be blazingly fast, the rotated updates would not be needed.
This page took 0 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.