Author: Günther Simon
Date: 12:05:40 02/06/06
Go up one level in this thread
On February 06, 2006 at 14:33:13, chandler yergin wrote:
>On February 06, 2006 at 14:14:33, Robert Hyatt wrote:
>
>>On February 06, 2006 at 14:03:46, chandler yergin wrote:
>>
>><crap snipped>
>>
>>And this has _what_ to do with "huge hash" vs "small hash"???
>Here are 'middle game' positions with "huge hash" and how "small hash solves
>them in record time!
...
>{'Hiarcs 7.32-64MB/450 MHz Pentium II' 91s / 9, 'Fritz 5.32-111MB/450 MHz
How much total physical ram has your historical P2 450?
I bet you never really listened about the _first premise_ for
all we discussed here. It simply says never exceed your physical ram.
If swapping takes place you can never compare solution times
(and engine games would be useless at all).
I guess your old machine probably has 128MB and WIN98 (or may be it is even
WIN95?) already needs around 70MB IIRC (it's so long since that times)
<quote>
{'Hiarcs 7.32-64MB/450 MHz Pentium II' 91s / 9, 'Fritz 5.32-111MB/450 MHz
Pentium II' 3s / 10, 'Junior 5.0-111MB/450 MHz Pentium II' 516s / 17, 'Crafty
16.15-111MB/450 MHz Pentium II' 1s / 9 BEST, 'Fritz 6.00-128MB/ 450 MHz
Pentium 2' 3s / 9, 'Junior 6.0-128MB/450 MHz Pentium 2' 231s / 18, 'Nimzo 7.
32-127MB/450 MHz Pentium 2' 1s / 7, 'Little Goliath 2000 V2. 5-128MB/450 MHz
Pentium 2' 1s / 8} 1. Rxe5 {Fritz 5.32 solves in 3 secs} ({Fritz 8:} 1. Rxe5
Rg8 2. Re3 Bf8 3. Rh1 Qd8 4. Reh3 Ra8 5. Ne4 Bg7 6. Qh5 Bf8 {[%eval 478,14]})
1... fxe5 2. f6 exf6 3. Rxf6 Kg8 4. Rxh6 Bxh6 5. Qxh6 Rd7 6. Ne4 Qf8 7. Nf6+
Qxf6 8. Qxf6 b5 9. Qg5+ Kf8 10. Qf5+ Kg8 11. exd7 Rd6 12. Qh7+ Kf8 13. Bg6 Rxd7
14. Qxd7 Kg8 15. Qf7+ Kh8 16. Qh7# 1-0
</quote>
Anyway, if you look above you'll see also, that you have already screwed up
your 'test' conditions. You cannot compare _different_ programs solutions
with different hash sizes (even if you had no swapping).
This simply tells nothing about the influence of hash at all.
You need of course to test the _same_ programs with different hash sizes,
but this should have been self evident...
... but I see:
Hiarcs 7.32 64MB
Fritz 5.32 111MB
Junior 6.0 128MB
Fritz 6.0 128MB
Crafty 128MB
LGoliath 128MB
Nimzo 128MB
you have never tested the same program with more than one different
hash setting, thus actually you compared nothing.
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.