Brutus v7.02 released
2 posters
Page 1 of 1
Brutus v7.02 released
Hi there,
At the eve of the Division 4 tournament, the non-beta version of Brutus is released. To avoid confusion it is named v7.02 including the following features compared to the beta:
In my benches it performs slightly better than v7Beta01. This time, a good win32-compile (Intel-PGO) is included also.
Please make sure you enter the correct EGTB-path! It has to be set manually in the .ini-file. Make sure you delete the old .ini file, as well as any smallegtb-directory that was included in v6! That one is buggy!
Stephan
At the eve of the Division 4 tournament, the non-beta version of Brutus is released. To avoid confusion it is named v7.02 including the following features compared to the beta:
- Some minor bugs fixed
- Altered the king-tropism code of the evaluation function
- altered the margin of lazyEval
In my benches it performs slightly better than v7Beta01. This time, a good win32-compile (Intel-PGO) is included also.
Please make sure you enter the correct EGTB-path! It has to be set manually in the .ini-file. Make sure you delete the old .ini file, as well as any smallegtb-directory that was included in v6! That one is buggy!
Stephan
Stephan Vermeire (Brutus)- Number of posts : 32
Registration date : 2008-05-21
The 64 bit version of Brutus crashes under Arena 2.01
It crashes when playing an engine/engine contest against itself. It may be an artifact of using processes, if they are using the same shared memory objects.
It analyzed EPD problem sets without crashing.
It analyzed EPD problem sets without crashing.
Re: Brutus v7.02 released
Hi Dann,
You are absoluterly right! Due the multiprocesses, I am using shared memory that is equal for each instance of the engine. I am planing to fix that problem in the future so you can play Brutus against itself as long as the two versions are named differently. (I will allocate shared memory according to the name of the engine). Will that suite your needs?
I am quite surprised you found this 'bug' so quickly. I'm impressed!
Best wishes,
Stephan
You are absoluterly right! Due the multiprocesses, I am using shared memory that is equal for each instance of the engine. I am planing to fix that problem in the future so you can play Brutus against itself as long as the two versions are named differently. (I will allocate shared memory according to the name of the engine). Will that suite your needs?
I am quite surprised you found this 'bug' so quickly. I'm impressed!
Best wishes,
Stephan
Stephan Vermeire (Brutus)- Number of posts : 32
Registration date : 2008-05-21
Re: Brutus v7.02 released
Finding the bug quickly was a function of my testing methodology.
I built it from source (64 bit mode) and then I always play a match against itself.
This is a quick way to set up a match under areana.
I do this with the profiler enabled so that I can create a PGO binary.
It seems a logical procedure to me, because I am using the maximum CPU resources for the one given engine and so I can collect data that way more rapidly than any other way.
I built it from source (64 bit mode) and then I always play a match against itself.
This is a quick way to set up a match under areana.
I do this with the profiler enabled so that I can create a PGO binary.
It seems a logical procedure to me, because I am using the maximum CPU resources for the one given engine and so I can collect data that way more rapidly than any other way.
Similar topics
» Brutus version 6.0 released
» Brutus v7Beta01 released
» LittleThought v1.04 released
» Dirty 6J-22 released
» Rotor 0.4 released
» Brutus v7Beta01 released
» LittleThought v1.04 released
» Dirty 6J-22 released
» Rotor 0.4 released
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum
|
|