Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Here is some test data

Author: Robert Hyatt

Date: 14:29:39 09/05/03

Go up one level in this thread


On September 04, 2003 at 22:12:59, Dezhi Zhao wrote:

>On September 04, 2003 at 21:16:50, Jeremiah Penery wrote:
>
>>On September 04, 2003 at 20:07:09, Dezhi Zhao wrote:
>>
>>>If you use more registers than necessary, the CPU has to do less register
>>>renaming work.
>>
>>The CPU renames the registers for every instruction, therefore doing the same
>>amount of renaming work in every case.
>
>I read a document somewhere in the Intel site that says if you use more
>registers you have less registers for renaming. It comes from my memory. I
>forgot the exact document.
>
>>
>>For more, see
>>http://www.realworldtech.com/forums/index.cfm?action=detail&PostNum=1018&Thread=21&entryID=11428&roomID=11
>>
>>Read the whole thread if you're interested, it's good stuff.
>
>thanks. i will read it.


The problem is, _every_ time you write to a register, you do a rename, so that
the original value can be kept for OOO execution of previous instructions.  The
new value in a new (renamed) register can be used for instructions following
the instruction just executed.

The more you write to registers, the more renames you do...



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.