Arasan 24.2

[jdart]

Arasan 24.2 is available from https://www.arasanches.org.

Changes in Arasan 24.2:
1) Various Makefile fixes/enhancements, including support for compilation on
Windows with clang-cl
2) Performance improvements to PGN reading code
3) Remove log file support
4) Make position learning work in UCI mode, make default learning file
location a writeable directory
5) Add option to set learn file location
6) Read arasan.rc from the home directory on Windows, if present.
7) Re-organize NNUE SIMD code for feature transformer, make it work properly
when accumulator width is not an exact multiple of the register count
8) Protocol fixes: in UCI mode, ignore illegal moves sent from interface
(Arena can do this). Ignore move sent after game end in CECP mode.
9) Build executables with clang
10) Support compilation on ARM Linux with GCC (aarch64 architecture)

Note position learning is off by default: it can be enabled by setting the appropriate CECP or UCI option. It is disabled when low strength settings are in place (<50% full strength).

I don't expect much playing strength difference over version 24.1, except possibly on Windows, where the new clang-built executables are significantly faster.

[pkumar]

The link works as below:
https://www.arasanchess.org/

[peter]

Thanks for the new version, Jon!
Position learning now works in Shredder (13) too (in Arena installed as Winboard- engine it did as 24.1 already) as UCI- engine, but .lrn has to be created in the given AppData- directory, and in .rc- file it has to be set to true as well as in UCI- options.
In Fritz (19) it doesn't work, regards

[Graham Banks]

Thanks, Jon.

[Graham Banks]

Hi Jon,

keeps losing on time on my 5950x.

I emailed you a couple of debug files.

[Graham Banks]

Posted the issue in github as well.

[jdart]

Posted the issue in github as well.

Will take a look.

[Graham Banks]

Fixed exes coming soon. Thanks, Jon.

[jdart]

I have released a patched version, available at https://www.arasanchess.org.

This includes a fix for an issue that could cause the program to crash, which would cause some GUIs to report games lost on time. I have only seen this behavior on Windows, but it could possibly occur elsewhere. I have also increased the default move overhead value, as another way to reduce losses on time.

This may not be a complete fix: Graham was saying he still had issues. But I'm releasing it because it does fix a real problem.

[peter]

Thanks, Jon!
So far no more disconnections in Cutechess, which happened practically at once after starting a new match before. Shredder had been better (at least disconnection was counted "only" as time loss but tournament went on, in Cutechess match got stopped by single disconnection) but still about 20% of games.
Will try Shredder next, if no more time losses occur there now with new version neither anymore, no extra- mention will follow from my side here. Just as for position- learning, in Cutechess it doesn't work, in Shredder it still does, regards