On 4/22/07, Petter Urkedal <[EMAIL PROTECTED]> wrote:

I find the (src, src, dst)-ordered assembly easier myself, since reading

I've programmed assembly for both types to the point that it doesn't
bother me anymore.  I figure out which order I'm working with, and
somehow, I manage to not get confused.

Imagine being fluent in both Arabic and Japanese.  You don't notice
the opposing word orders.

I can't see a way to do 64 bit adds in less that 4 instructions without
introducing some sort of status register which needs to be saved in case
of async control flow.

IIRC, the way to do a 64-bit add on MIPS involves an add for the lower
half, a compare (subtract), a branch, an add for the upper half in the
delay slot, and another add that the branch may jump over that
increments the upper half in the case of a carry.  So 4 or 5
instructions, depending on the carry.

I've read about the reasoning process the designers went through when
deciding not to include result flags.  It has to do with tracking
instruction dependencies (what we call hazards).  Without flags, the
dependencies are restricted mosly to the register file.  With flags,
you end up having to be too conservative and deal with false
dependencies.

Y'all really need to read up on this stuff.  Patterson and Hennessy.
Computer Architecture: A Quantitative Approach.  Look them up on
Amazon.

--
Timothy Normand Miller
http://www.cse.ohio-state.edu/~millerti
Open Graphics Project
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to