- improved passed-pawn scoring in endgames
- added time to UCI output
- improved time control: x moves in y minutes
- 2 transposition tables: always replace and depth replacement scheme
- made transposition-table-size configurable
- all draw scores are now exactly 0 and stored in the TT
- bugs, again
I just found some topics on this forum regarding questions about 32/64 bit java versions, java.exe, javaw.exe, problems with batch-files and path variables, etc....
Are those questions/topics still relevant? Would be a shame if a java engine is not running optimally
sandermvdb wrote:I just found some topics on this forum regarding questions about 32/64 bit java versions, java.exe, javaw.exe, problems with batch-files and path variables, etc....
Are those questions/topics still relevant? Would be a shame if a java engine is not running optimally
1. I don't know what javaw.exe is, I use only java.exe.
2. Batch files can be used only under Arena.
3. .jar files can be used under WinBoard, the GUI understands them and creates an appropriate command line.
4. You cannot run java engines under Fritz directly. With some of them you can use InBetween but in fact I've had success only with Carballo so far.
5. I have no experience with the Shredder GUI.
One remark:
The JDK compiles the .java files to Java bytecode (.class or .jar). The output of a 64 bit JDK is the same as a 32 bit JDK.
The JRE runs the code and if your OS and CPU are 64 bit, you should install only the 64 bit JRE for optimal performance and less confusion.
The definition of a 64 bit Java engine is that it runs way better (in my case 3 times faster) on a 64 bit JRE because it uses a lot of 64 bit value types, bitboards for instance. (but my engine would also run on a 32 bit cpu with a 32 bit OS and a 32 bit JRE)
Graham Banks wrote:
I'm not sure whether or not you saw my message in TLCV, but Chess22k 1.2 loses on time in roughly 1 out of 7 games.
I didn't saw your message but thanks for the info.
This is something I am trying to fix as we speak. Problem is that I only check if I have enough time left when I have calculated a complete ply and can begin with the next ply. Sometimes however I get a 'node-explosion' in the next ply so I run out of time
