@CPW contributors

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

@CPW contributors

Post by smatovic »

Heyho, I am not aware of any other way of contact, so...

I am not into the Discord channels, took just a look at the top 20 engines on CCRL, I know only the ~half of engines and people. I can do minimal, incremental updates on existing CPW articles, like version updates of Lc0, SF, etc, but:

- people
- engines
- GUIs
- tournaments
- events
- companies

I am not qualified to write on these topics.

- 6502
- 68k
- Atari 8-bit
- Amiga
- GPU
- OpenCL

I can handle the above topics.

@CPW admins: there is still an active community in the Discord channels, maybe invite some active people, or one representative per Discord, Lc0/SF/OpenBench/EP, so we can divide engines and topics amongst contributors?

***edit***
Or, alternatively, the honest thing would be to freeze and lock current state of CPW with a timestamp.

--
Srdja
User avatar
jshriver
Posts: 1351
Joined: Wed Mar 08, 2006 9:41 pm
Location: Morgantown, WV, USA

Re: @CPW contributors

Post by jshriver »

What information are you wanting? I've wrote code for the 6502 (NES) and m68k (old Sparc IPX and other systems) as well as GPGPU back in the day and OpenCL. Though I highly recommend CUDA for nvidia over OpenCL. Good luck.
smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: @CPW contributors

Post by smatovic »

Some of the top 20 CCRL engines and programmers are not listed on CPW. PlayChess, LiChess, Chess.com have no articles. SPSA could be updated. DAG has no article. GPGPU article in context of training and inference of neural networks could be created, etc......CPW does not reflect current state of computer chess programming.

--
Srdja
smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: @CPW contributors

Post by smatovic »

@CPW-admins: one further wish, the printable .pdf export is pretty much unreadable, cos the links are written as fully URLs, would be nice if this can be changed to embedded links in the .pdf via the MediaWiki backend.
smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: @CPW contributors

Post by smatovic »

Bump.

@CPW-admins:

Last call for action. Pham Hong Nguyen and myself are the only contributors to CPW in the last 3 months. The Engine Programming Discord has now 122 members online with 692 total members, there is an active community out there, invite some active people to CPW?

As mentioned in another post, I myself am out of the loop of the current best practices. I will finish my chess projects over 2024 and then probably stop contributing to CPW.

---
Srdja
AndrewGrant
Posts: 1821
Joined: Tue Apr 19, 2016 6:08 am
Location: U.S.A
Full name: Andrew Grant

Re: @CPW contributors

Post by AndrewGrant »

Pretty sure some Stockfish people are in the process of entirely replacing CPW.
Friendly reminder that stealing is a crime, is wrong, and makes you a thief.
"Those who can't do, clone instead" - Eduard ( A real life friend, not this forum's Eduard )
User avatar
mvanthoor
Posts: 1784
Joined: Wed Jul 03, 2019 4:42 pm
Location: Netherlands
Full name: Marcel Vanthoor

Re: @CPW contributors

Post by mvanthoor »

smatovic wrote: Thu Nov 23, 2023 1:32 pm @CPW admins: there is still an active community in the Discord channels,
Do you have a list of Discord channels? Preferably ones that are not for a single particular top engine? I'm looking more at a Discord-version of Talkchess, a channel that is more general with regard to chess programming (or even, board game programming).

I've never done anything with Discord in the past, so I know nothing about it, honestly.
Author of Rustic, an engine written in Rust.
Releases | Code | Docs | Progress | CCRL
smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: @CPW contributors

Post by smatovic »

mvanthoor wrote: Sat Jan 13, 2024 4:32 pm ...
There you go:

Discord Channels
viewtopic.php?f=2&t=82700

I tried in the past with IRC, Cycosmos, ICQ, and lurked shortly into Discord, but this kind of communication is not mine.

--
Srdja
smatovic
Posts: 2797
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: @CPW contributors

Post by smatovic »

Short version:

I stopped contributing to CPW (seems I was the last one the past three months), did add a note to the main page with link to TC thread with collection of Discord channels.

Longer version:

I have another side project going on and lack the time to work on my Zeta and Eta chess engines, it would be fun to implement NNUE for GPU, to port PeSTO to 6502 processor and to port some simple C engines for AmyBoard GUI on 68k AmigaOS, but, the time...maybe later, when I am in retirement ;)

I engaged in saving TalkChess, but to save CPW single handed is above my capabilities, or alike.

--
Srdja
Viz
Posts: 113
Joined: Tue Apr 09, 2024 6:24 am
Full name: Michael Chaly

Re: @CPW contributors

Post by Viz »

Info on CPW is mostly extremely old and misleading.
The worst thing upcoming chess engine dev can do is to gather info at CPW (well, maybe next to gathering info at talkchess).
Like look at this https://www.chessprogramming.org/Late_Move_Reductions
It's all really wrong, "most" programs don't use any of this conditions (apart from depth one), less common conditions are literally not even a thing, well, some engines use this stuff for killers but it's mostly simplified away as you get stronger.
Also it completely lacks any modern approaches to reductions formula, lacks static adjustment of value for pv nodes, ttpv nodes, history-based adjustment (which is worth like 10 elo alone usually), lacks any mention of recent years improvement of allowing lmr to produce extensions if reductions are negative, as well as do deeper/shallower post lmr searches depending on lmr results.
Also nothing is written about stats assignment for moves that passed LMR by results of their full depth post-lmr search.
Etc, etc, etc.
Implementation that is described there probably will lose 70-80% of elo to modern implementation.
And this is just one article - other ones are also mostly as old and misleading as this one.
Not to mention it's kinda bizarre that there are some people listed as stockfish contributors with like 5 patches but people with 100~ are not listed whatsoever.