The Road to 3000

Discussion of chess software programming and technical issues.

Moderator: Ras

oriyonay
Posts: 32
Joined: Tue Jun 01, 2021 5:46 am
Full name: ori yonay

The Road to 3000

Post by oriyonay »

Howdy all - I hope you're having a blessed day :)
I'm working on my engine right now and have a good foundation. The engine plays at around 2200 elo level right now, and my current goal is to get it to 3000. My question is, what features should I be implementing (preferably in order) in order to get there? I know Koivisto was able to get there very quickly and want to do the same!

Features I already have in my engine: negamax, qsearch, ID, aspiration windows, check extension, MVV/LVA, null-move pruning, late-move reductions, TTs, killer/history heuristics, PVS, tapered evaluation (PeSTO) + (bishop pair, doubled/isolated/passed pawn, semi/fully open rook/queen files, basic piece mobility and king safety).

Thank you so much - I highly appreciate your help!!
- Ori Yonay
BrokenKeyboard
Posts: 24
Joined: Tue Mar 16, 2021 11:11 pm
Full name: Het Satasiya

Re: The Road to 3000

Post by BrokenKeyboard »

You seem to have a solid TT, I would recommend implementing some form of Lazy SMP once you are confident in your engines performance without multithreading. PSQTs are also useful if you haven't implemented them yet.
So my recommendation is:
1) Make sure that your engine is performant as you can get it
2) Implement PSQTs if you haven't already
3) Implement Lazy SMP, use a lockless hashtable and also have a replacement policy for your hashtable.
Luecx
Posts: 138
Joined: Thu Jun 18, 2020 9:20 pm
Full name: Finn Eggers

Re: The Road to 3000

Post by Luecx »

oriyonay wrote: Sun Aug 15, 2021 2:34 am Howdy all - I hope you're having a blessed day :)
I'm working on my engine right now and have a good foundation. The engine plays at around 2200 elo level right now, and my current goal is to get it to 3000. My question is, what features should I be implementing (preferably in order) in order to get there? I know Koivisto was able to get there very quickly and want to do the same!

Features I already have in my engine: negamax, qsearch, ID, aspiration windows, check extension, MVV/LVA, null-move pruning, late-move reductions, TTs, killer/history heuristics, PVS, tapered evaluation (PeSTO) + (bishop pair, doubled/isolated/passed pawn, semi/fully open rook/queen files, basic piece mobility and king safety).

Thank you so much - I highly appreciate your help!!
- Ori Yonay
When you say mega max, do your refer to the negamax version of pvs or of plain alpha-beta?
The ability to speak does not make you intelligent. https://github.com/Luecx/Koivisto

Image
Luecx
Posts: 138
Joined: Thu Jun 18, 2020 9:20 pm
Full name: Finn Eggers

Re: The Road to 3000

Post by Luecx »

Luecx wrote: Sun Aug 15, 2021 8:38 am
oriyonay wrote: Sun Aug 15, 2021 2:34 am Howdy all - I hope you're having a blessed day :)
I'm working on my engine right now and have a good foundation. The engine plays at around 2200 elo level right now, and my current goal is to get it to 3000. My question is, what features should I be implementing (preferably in order) in order to get there? I know Koivisto was able to get there very quickly and want to do the same!

Features I already have in my engine: negamax, qsearch, ID, aspiration windows, check extension, MVV/LVA, null-move pruning, late-move reductions, TTs, killer/history heuristics, PVS, tapered evaluation (PeSTO) + (bishop pair, doubled/isolated/passed pawn, semi/fully open rook/queen files, basic piece mobility and king safety).

Thank you so much - I highly appreciate your help!!
- Ori Yonay
When you say mega max, do your refer to the negamax version of pvs or of plain alpha-beta?
Ah I am sorry I didnt see you wrote "PVS". In my opinion you should be way higher with the stuff you already have implemented. So I assume you either have a bug in one of those. I think we reached 2200 almost with a vanilla pvs implementation.
The ability to speak does not make you intelligent. https://github.com/Luecx/Koivisto

Image