Author: James B. Shearer
Date: 10:35:04 01/16/00
Go up one level in this thread
On January 16, 2000 at 10:27:18, Robert Hyatt wrote:
>On January 16, 2000 at 02:17:21, James B. Shearer wrote:
>
<snip>
>> Hsu's projections in the 1990 Scientific American article for the next
>>generation machine due in 1992 were 1000 chess chips, 3 million positions per
>>second per chip, 1 billion positions per second overall and 3400 rating. The
>>actual results for DB-2 which arrived in 1997 (5 years late) were 256 chess
>>chips, 2-2.5 million positions per second per chip, 200 million positions per
>>second overall and rating 2900. So he missed rather badly.
>> James B. Shearer
>
>
>DB-2 used 480 chess chips, and peaked at 1B nodes per second. I didn't see the
>3200 rating, and wouldn't have believed _that_ had I seen it, which predicts
>a 3:1 win/lose ratio vs a 2800 player. But for his hardware, I think he hit
>it pretty accurately, except for the time-frame. When you read his book, you
>will see that at least 2 years were lost in some of the P/R things IBM had him
>do, such as playing in Hong Kong, where they had to cobble something together
>with old hardware to play. IBM wanted visibility _all_ along the way...
The IBM web site says 256 chips (32 processors * 8 chips per) but I
see Hsu says 480 (30 processors * 16 chips per) in his IEEE paper so I suppose
that is correct. Still less than 1000.
As for speed if you want to compare peak rate then the projected peak
rate was 3 billion (1000 chips * 3 million per) so he is still short.
The projected rating was 3400 not 3200.
James B. Shearer
PS: When is Hsu's book coming out?
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.