Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Still wrong

Author: Robert Hyatt

Date: 18:19:11 10/26/01

Go up one level in this thread


On October 26, 2001 at 17:12:17, Tom Kerrigan wrote:

>Almost all architectures have separate int/fp.
>
>Sun's MJAC, or MAJC, or whatever the heck they call it is hailed as somewhat
>innovative and definitely controversial for combining the two.
>
>http://www.cs.earlham.edu/~mutioke/cs63/sparc.htm
>
>"The floating point unit has 32 32-bit non windowed registers, which must be
>saved on a per-context basis"

OK.. You are clearly right here.  Guess I have been too Jaded over the years
in writing asm for the xerox and then the Cray machines.  There the registers
are general purpose.

Memory fails as age increases, apparently.  :)


>
>You seem to recognize that the FP registers on x86s are separate from integer
>registers, so your comment in an earlier post about 64-bit FP values choking the
>scalar datapath (or whatever you said exactly) makes no sense. The FP data
>doesn't have to ever be on the scalar datapath unless you explicitly do some
>converting.
>


There is only _one_ data path _into_ the CPU.  I was originally talking about
the 64 bit chunks that can flow into the cpu from outside.  And that is a
real bottleneck on Intel boxes, still.  IE you can't possible load
instructions, int data, and fp data, fast enough if you have to use memory.
And the classic SPEC benchmarks tend to stream data like crazy...






>-Tom



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.