[CCB] Cyclone 2.2: 2766

[swami]

Gaultlet continued from GBT 1 tournament.

Score: 81.5/102

Rank: 9

Rating: 2766

Notes: I've not seen any time management trouble as some seem to claim. Overall, this is Ok performance for Cyclone. It's roughly at Glaurung/Fruit's level.

[kranium]

Thanks for testing Swami...
can you give me some more info...

how many cpus were used?
is this number comparable to the CEGT and the CCRL results?
to which version of Fruit and Glaurung are you referring (there is a great disparity)..

also this tournament 'CCB', are the results published? If so can you provide a link?

Thx,
Norm

[swami]

Thanks for testing Swami...
can you give me some more info...

how many cpus were used?
is this number comparable to the CEGT and the CCRL results?
to which version of Fruit and Glaurung are you referring (there is a great disparity)..

also this tournament 'CCB', are the results published? If so can you provide a link?

Thx,
Norm

Hi Norm,

I think Cyclone used default settings with nothing changed, so that is 1 cpu, right?, I was referring to the 051103 version of Fruit and Glaurung 2.1.

Remaining other results apart from Cyclone, is published in this thread:

http://www.talkchess.com/forum/viewtopi ... hlight=gbt

I just added Cyclone to the already finished tournament to see the progress, so essentially it's a gauntlet blitz tournament.

[kranium]

Thanks for testing Swami...
can you give me some more info...

how many cpus were used?
is this number comparable to the CEGT and the CCRL results?
to which version of Fruit and Glaurung are you referring (there is a great disparity)..

also this tournament 'CCB', are the results published? If so can you provide a link?

Thx,
Norm

Hi Norm,

I think Cyclone used default settings with nothing changed, so that is 1 cpu, right?, I was referring to the 051103 version of Fruit and Glaurung 2.1.

Remaining other results apart from Cyclone, is published in this thread:

http://www.talkchess.com/forum/viewtopi ... hlight=gbt

I just added Cyclone to the already finished tournament to see the progress, so essentially it's a gauntlet blitz tournament.

yes, default = 1 cpu
excellent...thx.

[Ovyron]

I was referring to the 051103 version of Fruit

Well, I thought the engine's name was changed to Cyclone because it was a big improvement that required a name change.

[kranium]

I was referring to the 051103 version of Fruit

Well, I thought the engine's name was changed to Cyclone because it was a big improvement that required a name change.

Hi Ulysses-
I'm glad you asked! <:

Comparing to Fruit 051103 is an accurate indication of strength... (don't forget this is just one gauntlet from one person).

from the CCRL (based on 334 40/40 games):
Cyclone 2.2 2924
Toga II 1.4 beta5c 2908
Fruit 051103 2874

New features:
multiPV
dynamic hash
a single executable
4 custom made books
and more.

Although I need give no justification for forking from Toga with a new name, I will to say it was the right choice for me...

1) if you examine the Cyclone code, you'll see that it has many many changes, and is substantially re-written, re-organized. the program flow has changed, the multiCPU routines have been changed, many things have been fixed, code optimized, routines have been removed, more agressive pruning has been implemented. this amount of change would not be practical, or even possible while remaining in the Toga lineage. (this amount of change is atypical, for ex: toga141se (based on 14beta5c) changed just a few lines of code, and piece values.

2) The toga fork is not organized, bit of a mess, IMHO, with many new versions such as 1.5.5se, 1.57se, toga 1.4.7se, etc. etc. (in reality these new version #s are no more than PGO optimized compiles of 141se, (and distributed without code).

3) Lastly, when I started Cyclone in Aug., I immediately contacted the Toga developers group in an effort to colaborate, but this to no avail.
i.e. my request was not well received.

Hence Cyclone was born! (out of the whirlwind and chaos!).

Norm

[Ovyron]

So it's a different engine code-wise, thanks.