Myracle GUI

[mar]

If the following conditions are set:

- I have set an "Override own book" in the engine, say in Stockfish.
- I start "New Tournament" and check "Disable own book".

Stockfish will still use the book set in its engine options. No problem for me, but my question is, is that something that slipped through? Or can't you prevent it at that stage?

you're mistaken, it works as intended (just checked). you probably forgot to unset the global tour book

[mar]

build 83 is up:
- can change engine overrides in edit tournament dialog
- can enable "simple movelist" in settings/appearance to force line break before each white move (=cute-style movelist)
- can hide individual tabbars as part of layout config (in view menu - so that we don't get Krzysztof permabanned)

[mar]

build 84 is up:
- fix opening multiple pgn/epd files at once
- can pass command line args to open pgn/epd files

the web now also contains a (highly experimental) Linux version for Ubuntu-compatible distros; minimal only (no bundle, no engines included)

[Wilhelm]

Today I tested the init delay in run epd . I set 3 sec. but all 20 concurrency started immediately. I had expected all 3 sec. one. Is this correct?

William

[mar]

Today I tested the init delay in run epd . I set 3 sec. but all 20 concurrency started immediately. I had expected all 3 sec. one. Is this correct?

yes the init delay is additional startup delay given to the engine, so it starts all at once, waits for 3 secs and starts searching;
probably not what you want

so in the next build I'll include an additional parameter: delay rnd, in seconds, so the init delay will be base
and up to "delay rnd" extra seconds, randomly per engine instance.
so now if you set base init delay to 0 and delay rnd to 3, they's start crunching randomly, which is what you want I guess.

it's still unclear to me why you'd want this, but it's a simple change, so no problem...

[mar]

build 85 is up:
- better NAG handling in moves
- fix color burn table overflow for custom board themes
- fix highlight color not being updated properly when switching themes
- additional extra random delay (delay rnd) in run epd suite
- refuse to parse utf-16/BOM pgn files

[emadsen]

Thank you Martin for sharing your engine tournament GUI. Very impressive. Running multiple tournaments concurrently, with the ability to pause and resume individual tournaments, all subject to an overarching CPU budget, is a fantastic feature!

I did notice two issues I suspect are bugs.

1) While running my MadChess engine in a gauntlet tournament, I opened the scratchpad, pasted in an interesting position I found online, and started MadChess as an analysis engine. I wanted to see multiple lines from my engine. I didn't notice a +/- button in the GUI to quickly toggle it from 1 to 4 MultiPV lines. So I double-clicked the engine name in the Analysis panel at the bottom of the GUI. I changed MultiPV to 4 in the Engine Options dialog. The GUI correctly showed 4 lines of analysis. However, when I switched to a tournament tab, I noticed MadChess was handicapped. It was playing the tournament games with MultiPV = 4 (evident in the Analysis panel of the tournament games). I assumed my Engine Option change in the scratchpad was a local setting, but it appears to be a global setting.

Would it be possible to add +/- buttons to increase MultiPV for analysis engines only? Scoping it to analysis only and not requiring navigation into the Engine Options dialog is simpler and more intuitive for the user, in my opinion.

2) Even with MultiPV = 1, I've noticed analysis engines (all engines, not just mine) repeatedly stop and start if run while a tournament is running. If the tournament is paused and not using any CPU, analysis works fine. But if analysis is competing with a tournament for CPU, I see the following behavior in the Analysis > Debug panel:

Code: Select all

setoption name UCI_AnalyseMode value false
setoption name UCI_AnalyseMode value true
position startpos moves...
go infinite
[many info depth... lines from engine]
stop
bestmove e4g3
setoption name UCI_AnalyseMode value false
setoption name UCI_AnalyseMode value true
position startpos moves...
go infinite
[many info depth... lines from engine]
stop
[this repeats]

Not sure if it's relevant that I received a "cannot start analysis: out of scheduler budget" error when I first attempted to start analysis. I then paused the tournament and waited for a few CPUs to free up (though many tournament games still were running), then ran analysis and observed the repeated stop and start behavior.