Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: working set is < 384k (another correction)

Author: Robert Hyatt

Date: 21:00:12 08/21/04

Go up one level in this thread


On August 21, 2004 at 23:35:08, Robert Hyatt wrote:

>On August 21, 2004 at 20:49:26, Jeremiah Penery wrote:
>
>>On August 21, 2004 at 17:31:27, Robert Hyatt wrote:
>>
>>>
>>>
>>>BTW 384K is wrong.  If something is in L1, it is most likely in L2 as well.  L1
>>>is a most-recently-used subset of L2 just as L2 is a most-recently-used subset
>>>of main memory...
>>
>>In Intel processors, L2 cache is inclusive of L1 cache.  Not so in AMD
>>processors.
>
>
>How so?  How do you get something into L1 without it also first coming into L2?


OK.  Had forgotten about AMD's "exclusive L1/L2 cache policy".  So the
comparison is further flawed since the AMD with 128/256 is really 384kb of
cache.  and the 128/512 is bigger than the equivalent Intel.

Not that any of this makes any real difference to the discussion at hand, of
course...




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.