dje-dev

Please count me as another user who thinks Windows Terminal is great, but force word wrapping makes this product unusable - the output becomes unreadable in many situations where it...

Wow, this work looks very interesting and potentially very useful! Basic distributed training/inference (one host, multiple GPU) is currently a gap for Torchsharp and your implementation could be a step...

Yes, error indicates the engine exceeded the maximum number of allowable nodes. Ceres auto-configures this to some reasonable maximum based on the RAM in the computer (it has nothing to...

1. Correct. 2. Actually you have a very good point. It is intended that Ceres will just "stop thinking" when it reaches the configured limit, and I have tested that...

CUDA 11.5 should work fine. Most testing is done on Windows under CUDA 11.5, and Linux under 11.4, but Linux under 11.5 should work. Please indicate the exact version of...

Thank you. Unfortunately this seemingly won't quite work. The engine, if already created, may have already internalized the search parameters and my not reread them (in some cases). Instead I...

The error message stats that the cublas64_11.dll file cannot be found. I do not understand the specifics of how PAD works, but I suggest you locate this file and put...

No problem. If you are sure the line in Ceres.json point to the correct path, e.g. ``` "DirLC0Binaries": "c:\\dev\\lc0\\lc_270rc2", ``` and you are sure that this directory contains a working...

Under cutechess client it is known to be a problem that cutechess will not wait long enough for engine startup. This was previously logged on the cutechess github in conjunction...

There are two issues here. First, as mentioned in the readme, currently Ceres only supports 40b networks on Ampere cards (e.g. 3080), not on the 20x0 series. Even with 30x0...