Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SMP variability some real data to look at (final with 4 2cpu runs)

Author: Robert Hyatt

Date: 05:48:37 09/06/02

Go up one level in this thread


On September 06, 2002 at 08:01:22, Rolf Tueschen wrote:

>On September 06, 2002 at 01:32:23, Robert Hyatt wrote:
>
>>One quick note.  A couple of typos in last table in the 3rd two cpu column.
>>Got careless as to which file I was looking at and had to re-enter a couple of
>>the numbers.  Plus the 4th 2 cpu run...  Similar to the others, with some
>>better, some worse, speedup is still remaining fairly constant overall.
>>
>>The new "aggregate" is simply the sum of the 2cpu times from one column,
>>divided into the sum of the 1cpu column.  This tends to weight the searches
>>with the longer times more heavily, while the average counts all 24 positions
>>equally...
>>
>>The data:pos        1cpu       2cpu          2cpu          2cpu          2cpu
>> 1         153      89 (1.72)     88 (1.74)     90 (1.70)     89 (1.72)
>> 2         139      90 (1.54)     90 (1.54)     90 (1.54)     90 (1.54)
>> 3         130      82 (1.59)     87 (1.49)     87 (1.49)     82 (1.59)
>> 4         176     100 (1.76)     99 (1.78)    100 (1.76)     99 (1.78)
>> 5         147      95 (1.55)     99 (1.48)     87 (1.69)     95 (1.55)
>> 6         135      77 (1.75)     77 (1.75)     78 (1.73)     77 (1.75)
>> 7          92      50 (1.84)     50 (1.84)     50 (1.84)     50 (1.84)
>> 8         149     104 (1.43)     89 (1.67)     79 (1.89)    103 (1.45)
>> 9          80      60 (1.33)     62 (1.29)     60 (1.33)     50 (1.60)
>>10         155      82 (1.89)     76 (2.04)     75 (2.07)     81 (1.91)
>>11         142      80 (1.77)     80 (1.77)     82 (1.73)     82 (1.73)
>>12         105      76 (1.38)     76 (1.38)     76 (1.38)     75 (1.40)
>>13         149     105 (1.42)    105 (1.42)     87 (1.71)    107 (1.39)
>>14         147     104 (1.41)     87 (1.69)     74 (1.99)     93 (1.58)
>>15         150      86 (1.74)     86 (1.74)     86 (1.74)     86 (1.74)
>>16         159      73 (2.18)     72 (2.21)     72 (2.21)     99 (1.61)
>>17         158      97 (1.63)     97 (1.63)     77 (2.05)    107 (1.48)
>>18          85      42 (2.02)     41 (2.07)     42 (2.02)     40 (2.12)
>>19         125      61 (2.05)     63 (1.98)     73 (1.71)     61 (2.05)
>
>Bob, something is broken here as you can see in number 19. Two times you
>reported 1.21, what I already had questioned. Here suddenly I see 73 (1.71).
>Perhaps a reading error? NB I didn't check the other numbers. I just took the
>usual look for "inconsistencies".
>
>Rolf Tueschen
>
>


I definitely had a couple of typos in the 3rd 2cpu run.  I made a note of
that when I posted the final with 4 runs.  I had two windows open matching
moves/times for run 3 and another open watching run 4.  I got the wrong
window a couple of times.  I will check all 4 again today now that it is
not running.  And if you want to see the raw data to check yourself, feel
free to yell and I'll put the logs somewhere they can be accessed.

Take _all_ of this with a grain of salt until I/you/we check the numbers.  I
did these as they ran last night which makes it easy to screw up at 11pm
or later... :)

I found the errors I found by going back thru the logs, but I really need to
make one more careful pass today (or someone does) to be sure that I didn't
"cross over" on others...




>
>>20         160     112 (1.43)    117 (1.37)    109 (1.47)    107 (1.50)
>>21         156      73 (2.14)     62 (2.52)    103 (1.51)     53 (2.94)
>>22          77      60 (1.28)     73 (1.05)     41 (1.88)     42 (1.83)
>>23         134      93 (1.44)     95 (1.41)     97 (1.38)     97 (1.38)
>>24         131      84 (1.56)     82 (1.60)    115 (1.14)    105 (1.25)
>>  average speedup->    (1.66)        (1.69)        (1.71)        (1.70)
>>aggregate speedup->    (1.64)        (1.66)        (1.68)        (1.64)
>>
>>
>>This gives a pretty good idea of what I meant by "variability".  Some are
>>solid, some vary significantly...
>>
>>More of this tomorrow perhaps...



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.