Something wrong with Dragon 3.11-2?

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 27 April 2000 11:16:33:
FQ-2 tournament result: Bringer 1.6 - Dragon 3.11-2 9 - 1 !!
For me, this is somewhat surprising. In my GS-5 tournament (played under very similar conditions) Bringer 1.6 has 2 points, Dragon 3.11 (the old version with only 4 MB hash) has 3.5 (both out of 8 games). This seems to be controversial. Can it be that Dragon 3.11-2 is significantly weaker than its predecessor?
I myself can't try 3.11-2 since it does not run at all on my PC.

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 27 April 2000 12:45:22:
Als Antwort auf: / As an answer to: Something wrong with Dragon 3.11-2? geschrieben von: / posted by: Gabor Szots at 27 April 2000 11:16:33:

FQ-2 tournament result: Bringer 1.6 - Dragon 3.11-2 9 - 1 !!
For me, this is somewhat surprising. In my GS-5 tournament (played under very similar conditions) Bringer 1.6 has 2 points, Dragon 3.11 (the old version with only 4 MB hash) has 3.5 (both out of 8 games). This seems to be controversial. Can it be that Dragon 3.11-2 is significantly weaker than its predecessor?
I myself can't try 3.11-2 since it does not run at all on my PC.

Some time ago I raised the same question, but absolutely nothing happened. I don't know if Dragon 3.11-2 has a problem with the increase in hash size, but I think it might be the case. I cannot make it acknowledge the presence of tablebases either. I recommend using the Dragon 3.11-1 for testing, because it behaves more consistently than new version.
Best wishes...
Mogens

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 27 April 2000 13:34:32:
Als Antwort auf: / As an answer to: Re: Something wrong with Dragon 3.11-2? geschrieben von: / posted by: Mogens Larsen at 27 April 2000 12:45:22:

FQ-2 tournament result: Bringer 1.6 - Dragon 3.11-2 9 - 1 !!
For me, this is somewhat surprising. In my GS-5 tournament (played under very similar conditions) Bringer 1.6 has 2 points, Dragon 3.11 (the old version with only 4 MB hash) has 3.5 (both out of 8 games). This seems to be controversial. Can it be that Dragon 3.11-2 is significantly weaker than its predecessor?
I myself can't try 3.11-2 since it does not run at all on my PC.

Some time ago I raised the same question, but absolutely nothing happened. I don't know if Dragon 3.11-2 has a problem with the increase in hash size, but I think it might be the case. I cannot make it acknowledge the presence of tablebases either. I recommend using the Dragon 3.11-1 for testing, because it behaves more consistently than new version.
Best wishes...
Mogens

Hello Mogens,
I kind of expected that you'd answer since I remembered you had tested Dragon 3.11-2 against various opponents.
I have just finished a 10-game 5-minute blitz between Bringer 1.6 and Dragon 3.11-2, the score was the same 9-1 as Frank's. Now this Pentium 166 machine has 32 MB RAM, of which I used 4 MB for hash. The machine on which Dragon 3.11-2 doesn't want to run is a Celeron 433 with 128 MB RAM, 32 MB for hash. This seems to confirm your suspicion about some kind of memory problem.
Best regards,
Gábor

[Mogens Larsen]

Geschrieben von: / Posted by: Mogens Larsen at 27 April 2000 15:44:05:
Als Antwort auf: / As an answer to: Re: Something wrong with Dragon 3.11-2? geschrieben von: / posted by: Gabor Szots at 27 April 2000 13:34:32:

Hello Mogens,
I kind of expected that you'd answer since I remembered you had tested Dragon 3.11-2 against various opponents.
I have just finished a 10-game 5-minute blitz between Bringer 1.6 and Dragon 3.11-2, the score was the same 9-1 as Frank's. Now this Pentium 166 machine has 32 MB RAM, of which I used 4 MB for hash. The machine on which Dragon 3.11-2 doesn't want to run is a Celeron 433 with 128 MB RAM, 32 MB for hash. This seems to confirm your suspicion about some kind of memory problem.

Well, I answer all the questions I know anything about and then some ). In one of my tournaments Dragon 3.11-1 did very well. When I compared the same opponents to the new version of Dragon, the results were completely different. It didn't perform well in my tournament ML-8 either.
Difficult to say, but I'll use the Dragon 3.11-1 until the new Dragon 4.xx arrives.
Best wishes...
Mogens

[Gabor Szots]

Geschrieben von: / Posted by: Gabor Szots at 28 April 2000 07:45:06:
Als Antwort auf: / As an answer to: Re: Something wrong with Dragon 3.11-2? geschrieben von: / posted by: Mogens Larsen at 27 April 2000 15:44:05:

Hello Mogens,
I kind of expected that you'd answer since I remembered you had tested Dragon 3.11-2 against various opponents.
I have just finished a 10-game 5-minute blitz between Bringer 1.6 and Dragon 3.11-2, the score was the same 9-1 as Frank's. Now this Pentium 166 machine has 32 MB RAM, of which I used 4 MB for hash. The machine on which Dragon 3.11-2 doesn't want to run is a Celeron 433 with 128 MB RAM, 32 MB for hash. This seems to confirm your suspicion about some kind of memory problem.

Well, I answer all the questions I know anything about and then some ). In one of my tournaments Dragon 3.11-1 did very well. When I compared the same opponents to the new version of Dragon, the results were completely different. It didn't perform well in my tournament ML-8 either.
Difficult to say, but I'll use the Dragon 3.11-1 until the new Dragon 4.xx arrives.
Best wishes...
Mogens

This all is too much to be merely coincidental.
So do I.
Best regards,
Gábor