Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Level 2 cache sizes and the rights of Tom Kerrigan

Author: Robert Hyatt

Date: 07:53:23 08/26/04

Go up one level in this thread


On August 26, 2004 at 10:34:13, Vincent Diepeveen wrote:

>On August 26, 2004 at 10:14:18, Robert Hyatt wrote:
>
>>On August 25, 2004 at 09:56:00, Vincent Diepeveen wrote:
>>
>>>On August 25, 2004 at 05:41:09, Daniel Clausen wrote:
>>>
>>>>On August 25, 2004 at 05:04:00, Graham Laight wrote:
>>>>
>>>>>On August 24, 2004 at 20:54:33, Christopher Conkie wrote:
>>>>>
>>>>>>Hi Eduard,
>>>>>>
>>>>>>It's all very interesting these differing opinions. I'm just wondering whether
>>>>>>all the users who want to discuss about computer chess would like to go to
>>>>>>USENET. I'm not sure that Steve thought it was a good idea either.
>>>>>>
>>>>>>You have got me facinated however. Where is this moderate, frugal, limited,
>>>>>>ordinary place we should all go to talk about computer chess?
>>>>>
>>>>>http://groups.google.co.uk/groups?hl=en&lr=&ie=UTF-8&c2coff=1&group=rec.games.chess.computer
>>>>
>>>>I'm neither pro or contra USENET here, but keep in mind, that the discussion
>>>>style would change slightly. In USENET it can take up to a day until everyone
>>>>can see the post. It's also possible that you see the answer to something w/o
>>>>seeing the original question yet. Nothing wrong with that, but it's something
>>>>different to how CCC works at the moment.
>>>
>>>A big problem of CCC is that it moves too fast and that searching in archives
>>>never works and after a day or 2 all postings you did you can't read any reply
>>>from anymore as they are gone. Usenet gets saved everywhere.
>>>
>>>A good example is that a few years ago here at CCC hyatt posted that he had
>>>tested quad xeons with 1MB L2 cache versus Xeons with several MB's L2 cache and
>>>saw zero difference in speed.
>>
>>Grow up.  There were _never_ xeons with "several MB's L2 cache".  I specifically
>>tested 512K 1024K and 2048K.  The results I posted a few years back simply said
>>"the larger cache is not worth the much larger cost."
>
>>Feel free to make up whatever you want, of course.  You always do...
>
>To be exact you posted there was 0% difference in speed between the different L2
>cache sized Xeons.
>

Please post the quote, with context.  "zero percent difference with respect to
cost" is one thing.  I _might_ have said that as 17% improvement for 4x the cost
is not very good.  But I'll wait for your quote.  Particularly since you also
claimed I wrote a JICCA paper about Crafty that you could never cite.
Particularly since you claimed that I had said my speedup formula worked for any
number of processors when I had actually claimed it was an estimate that ha only
been tested through 8/16 processors.


You are good at posting things you _claim_ others wrote.  But you _never_ back
them up with facts.  In the case of the JICCA article, you can't get away with
blaming CCC archives.  the JICCA has a complete list of all articles available
on the web.  What is your excuse there?

Fraud?

The only cache question I ever remember answering here was the "Is the larger
price of the bigger xeon L2 cache worth it?"  My answer was "no".  My quad
chassis cost me about $5500 when I bought it.  The 512K L2 xeons were about
$1250 each.  Total cost was 10,500 plus disk drives.  1024K xeons were 2500
bucks a pop.  Price of machine would have climbed to 15,500.  2048K xeons were
about $5500.  Price would have climbed to $27,500.  Was that worth the 17%
improvement?  Not by any reasoning I could think of.



>Which renders your entire discussion with Tom Kerrigan as complete nonsense from
>your side.

Yes it was.  I just ran the test with a well-known program to see what happened,
and posted the results here as I always do.  When I make a statement I generally
post data to back it up.  When someone else posts something, you _always_ demand
data.  And I usually supply it.

Where is _your_ data to support _your_ speedup claims for _your_ program?
Remember the days of "my speedup is > 2.0 for two processors and I can proof why
that is reasonable with good hashing"???

Where _is_ the data?  Where is the WCCC data?  Where is _any_ data to support
your crap?  Only in your head???

I've made all of mine available.   Which is _far_ more than you have done.

And you talk about dishonesty.

Check out the closest mirror.

So in reality, _any_ discussion with _You_ is a pointless discussion.  You make
things up.  You even make things up and then claim that I (or others) wrote them
to support your own made-up stuff.

Marvelous strategy.

>
>>
>>>
>>>Now we have a big thread here where he denies it.
>>>
>>>At RGCC such an idiocy would not happen. You search his old posting and dang, he
>>>has to shut up.
>>>
>>>Now there have been tens of postings with Kerrigan in all his rights exploding
>>>and Hyatt keeping posting nonsense.
>>>
>>>Even worse is that in CCC this nonsense gets posted within a few minutes. In
>>>RGCC it takes at least 1 minute for your posting to show up and a few minutes
>>>for world wide distribution.
>>>
>>
>>
>>Shows your ignorance of usenet news.  It all depends on how your local newsfeed
>>is configured.  We send and receive usenet news articles _continuously_ here, so
>>that there is little delay on our end.  If your local newsfeed only uploads
>>every 15 minutes, you will see a 15 minute delay before your articles get out.
>>But that isn't the case for general usenet news.
>>
>>
>>
>>
>>
>>>Note that if you want to pay for a message board system that will cost you the
>>>big amount of 12 dollar a month, if you want your own root server with 1.2
>>>terabyte bandwidth a month it goes up to 50 dollar a month soon.
>>>
>>>Starting a member paid message board system is the biggest idiocy i ever read
>>>about.
>>>
>>>CCC is already reaching so little persons.
>>>
>>>Someone interested in computerchess will not soon find CCC. In fact they never
>>>find it as you need a username and a password to login.
>>>
>>>I do not know a single computerchess enthusiast who found himself this forum.
>>>
>>>Now let alone people who are interested in computerchess but are not very
>>>fanatical searching for information.
>>>
>>>>Also, messages which only make sense for a few minutes or hours (like
>>>>game-number of a certain game in ICC) won't be possible anymore. Some people
>>>>will consider this a good thing, others a bad thing.
>>>>
>>>>Another point is that moderation as it works here won't work there. You either
>>>>have no moderation or you have a moderated group, where each post has to be
>>>>approved first before it gets forwarded to the group. (at least I think that's
>>>>how it works, please correct me if I'm wrong)
>>>>
>>>>Sargon



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.