Koivisto 5.0

[Daniel Shawul]

Scorpio is only a top 10 engine if it uses a GPU. With CPU only, Scorpio is much weaker.

that's completely irrelevant. I'm deeply disgusted with the current state of computer chess,
so much hatred and nonsense is really something.
especially noting the high number of trolls that hijacked this Koivisto release thread just because Luecx dared to express his opinion.
there's nothing else for me here, really.

so, congratulations and good luck to koi guys, I'm outta CC, bye

hijack?? Are you dumb? the guy literally spent more than half of his release accusing others how they are in-original and cut n pasters they are etc... He basically asked for it.

[Madeleine Birchfield]

I blame TCEC for this, as their nnue guidelines are the only reason why so many of the top engines have copied Stockfish code. If TCEC had required the evaluation code to be unique in their guidelines then the likes of Igel, Nemorino, Rubichess et cetera would never have copied Stockfish code and all would have been much more unique than right now, because otherwise they would have been kicked out of TCEC. The state of affairs with copying Stockfish code has given many people the wrong impression that neural networks are just copy and paste.

It's clear TCEC has a huge influence on engine development. TCEC guidelines was the only reason why Igel's author decided to train own nets for Igel instead of using Dietrich Kappe's Night Nurse net. Without current guidelines on training data, Igel would most likely still be using a night nurse net. The same thing would likely happen with nnue if TCEC guidelines required unique evaluation code.

[mvanthoor]

that's completely irrelevant. I'm deeply disgusted with the current state of computer chess,
so much hatred and nonsense is really something.

The top engines all have a team of people around them; as one of the posters says, they don't discuss their stuff here (anymore), because those engines are beyond the basics. What I dislike most, is that much of the discussion seems to be among team members, or maybe between teams, in apps such as Discord.

AFAIK, that information is lost after the chat is done, or is it possible to search Discord chats from the past? The one thing that was invaluable to me while getting up to speed with chess programming was all the historical information in either this forum, Google Groups, or on the internet. Public information about chess programming seems to have started at around +/- 1995. Before that, everything was private between the better chess programmers and teams, and the top engines seem to be going back to that state of affairs. That is what I dislike most. Information that is exchanged which then only exists in the head of the participants of the exchange; it is basically lost for any programmer that comes after.

[Madeleine Birchfield]

AFAIK, that information is lost after the chat is done, or is it possible to search Discord chats from the past?

In most Discord servers it is possible to search for comments made since the beginning of the server. In the Leela discord for example I was able to find comments from mid 2018 regarding the training of the T10 nets. The Openbench discord is the exception I know of, because Andrew Grant deletes all the messages on the server once in a while, so past a certain date there are no messages.

[smatovic]

Come on guys, I advise to relax a little bit, all this NNUE stuff for CPU is new and the community has to figure out in an collective process how to deal with this on multiple levels...

--
Srdja

[Wilson]

Our search has also made a lot of progress. Adding further unique ideas not found in any other engine so far, we gained a large amount of elo inside our search since then.

could you please point me to these "unique ideas not found in any other engine so far" ?

I've taken a look at your search and all I see is stuff copied from Stockfish with some minor modifications so that one can't really say "100%" copied (just like in Ethereal, Booot):

1) same logic and even same formula for LMR initialization

Code: Select all

    for (d = 0; d < 256; d++)
        for (m = 0; m < 256; m++)
            lmrReductions[d][m] = 1.25 + log(d) * log(m) * 100 / LMR_DIV;

anyone interested can read more about this formula copied by everyone in the thread "A question about code similarities" opened by Oliver Brausch in the engine origins forum

2) same "improving" variable to check if the evaluation is improving to do less pruning/reductions

Code: Select all

bool  isImproving = inCheck ? false : sd->isImproving(staticEval, b->getActivePlayer(), ply);

3) same adjusting of the static eval based on tt bounds and score

4) basically the same razoring, static null move pruning, null move pruning and probcut

5) the reduction proposed by Ed Schröder to replace IID but you get a pass on that as Stockfish took it from Schröder but you still call it "Internal Iterative Deepening" which is not anymore. Internal Iterative Reduction as suggested by Ed himself would probably be more accurate

6) basically the same late move pruning, history pruning, see pruning

7) basically the same singular extension

8) basically the same lmr adjustments using history scores

9) same counter move history

10) you have also copied the capture history

As original ideas I see (but I haven't checked Stockfish in a while ):

1) threat pruning
2) kk reductions (which, as you write in the comments, is similar to Stockfish cutnode reduction)

did I miss anything?

care to share how much ELO points are these original ideas worth? and how much the stuff taken from Stockfish?

I honestly cannot bear all these so called chess programmers that come here to lecture the others about what real men should do and they are the first to plunder the hell out of Stockfish, always very careful not to copy same implementation details, to gain places on rating lists they despise so much.

there are no new algorithms in all these engines and since they cannot copy line by line their master, we simply get worse Stockfishes with subpar implementations.

and I really don't understand people like Albert and Norman: writing a weak engine from scratch is really simple. Then you simply copy all the stuff from Stockfish and then you release it under GPL license. As Andrew taught us thanking Stockfish in your sources/home page is not mandatory, commits are enough. Nobody could accuse you of anything

[Damir]

I blame TCEC for this, as their nnue guidelines are the only reason why so many of the top engines have copied Stockfish code. If TCEC had required the evaluation code to be unique in their guidelines then the likes of Igel, Nemorino, Rubichess et cetera would never have copied Stockfish code and all would have been much more unique than right now, because otherwise they would have been kicked out of TCEC. The state of affairs with copying Stockfish code has given many people the wrong impression that neural networks are just copy and paste.

It's clear TCEC has a huge influence on engine development. TCEC guidelines was the only reason why Igel's author decided to train own nets for Igel instead of using Dietrich Kappe's Night Nurse net. Without current guidelines on training data, Igel would most likely still be using a night nurse net. The same thing would likely happen with nnue if TCEC guidelines required unique evaluation code.

You should say the same about Lelenstein Alliestein, Scorpio, Stoofvlees as well than, since they copy things from Lc0 too...

[Madeleine Birchfield]

I blame TCEC for this, as their nnue guidelines are the only reason why so many of the top engines have copied Stockfish code. If TCEC had required the evaluation code to be unique in their guidelines then the likes of Igel, Nemorino, Rubichess et cetera would never have copied Stockfish code and all would have been much more unique than right now, because otherwise they would have been kicked out of TCEC. The state of affairs with copying Stockfish code has given many people the wrong impression that neural networks are just copy and paste.

It's clear TCEC has a huge influence on engine development. TCEC guidelines was the only reason why Igel's author decided to train own nets for Igel instead of using Dietrich Kappe's Night Nurse net. Without current guidelines on training data, Igel would most likely still be using a night nurse net. The same thing would likely happen with nnue if TCEC guidelines required unique evaluation code.

You should say the same about Lelenstein Alliestein, Scorpio, Stoofvlees as well than, since they copy things from Lc0 too...

And I have. In fact one of my first posts on this forum was about Allie reusing Leela's cudnn backends.

[Twipply]

You should say the same about Lelenstein Alliestein, Scorpio, Stoofvlees as well than, since they copy things from Lc0 too...

I don't think you know if Stoofvlees has copied anything or to what extent. Besides that, note that the author of Stoofvlees is also the author of Leela Zero - from which Leela Chess Zero was initially forked.

[AdminX]

You should say the same about Lelenstein Alliestein, Scorpio, Stoofvlees as well than, since they copy things from Lc0 too...

I don't think you know if Stoofvlees has copied anything or to what extent. Besides that, note that the author of Stoofvlees is also the author of Leela Zero - from which Leela Chess Zero was initially forked.

Don't forget he is also the author of Sjeng