Elements of the ULTIMATE Chess GUI?

[Dann Corbit]

Many of the ideas you mentioned I would love to see them in a GUI as well but it seems to me the nature of the ideas fit better in a database program like SCID.

The GUI needs to be attached both to engines and to databases.
Otherwise, it is just a thing that scrolls through games.

Do you mean somehow combine these things?
In the sense that when you do an analysis, it is stored in the database and compared with a specific engine, its version, its settings. And when this position is reopened, the analysis will be saved.

What, do you mean like this:

tcec.png

[sovaz1997]

Like, if you analyze a position, then this analysis is stored in a database. Re-opening a position allows you to see the analysis you’ve made. Also, it can be used to fill the engine's hash table (by giving it quick launches of PV moves from the analysis).

[Dann Corbit]

{snip}
Now I understand what you are talking about. Yes, I think there is a definite advantage to having a general-purpose database for this: unlimited expansion options and more flexibility in data management. The ability to make your own requests of any complexity.
But I am afraid that such a database may take up more space and have a lower access speed.

If we use MonetDB, then the data will be accessed at (approximately) memory speed.
And the AMD 7nm Epyc stuff is coming this summer.
$3500 will buy you 18 TB of enterprise class SSD.
A little more, if you want m.2 gumsticks.

Compute power is not the problem.
Utilization of that power for optimal purpose is the problem.

[Fulvio]

If we use MonetDB, then the data will be accessed at (approximately) memory speed.

Do you have some concrete benchmark?
Let's say a 10 million games database: how long it takes to find the games (not just the positions, if I would like to study the ending I need to see the continuation) with for example a king+rook vs king+rook+pawn ending?

[sovaz1997]

{snip}
Now I understand what you are talking about. Yes, I think there is a definite advantage to having a general-purpose database for this: unlimited expansion options and more flexibility in data management. The ability to make your own requests of any complexity.
But I am afraid that such a database may take up more space and have a lower access speed.

If we use MonetDB, then the data will be accessed at (approximately) memory speed.
And the AMD 7nm Epyc stuff is coming this summer.
$3500 will buy you 18 TB of enterprise class SSD.
A little more, if you want m.2 gumsticks.

Compute power is not the problem.
Utilization of that power for optimal purpose is the problem.

In that case, this is a very good idea.

[sovaz1997]

If we use MonetDB, then the data will be accessed at (approximately) memory speed.

Do you have some concrete benchmark?
Let's say a 10 million games database: how long it takes to find the games (not just the positions, if I would like to study the ending I need to see the continuation) with for example a king+rook vs king+rook+pawn ending?

For the ending, there are other solutions - the Nalimov tables, the Syzygy tables. I don't think that the database is suitable for this.

[Fulvio]

For the ending, there are other solutions - the Nalimov tables, the Syzygy tables. I don't think that the database is suitable for this.

Oh... no positional search?
That's a very big limitation for the ULTIMATE Chess GUI...

[Dann Corbit]

If we use MonetDB, then the data will be accessed at (approximately) memory speed.

Do you have some concrete benchmark?
Let's say a 10 million games database: how long it takes to find the games (not just the positions, if I would like to study the ending I need to see the continuation) with for example a king+rook vs king+rook+pawn ending?

MonetDB is a general purpose in-memory transaction based ACID compliant column based SQL store.
https://www.monetdb.org/Home

Independent Benchmark:
https://www.slideshare.net/FENGGEORGEYU ... ncharmarks

Anther interesting choice is omnisci, but it runs on GPU and has some limitations (e.g. GPU memory size).
https://github.com/omnisci

Clickhouse is also interesting:
https://clickhouse.yandex/
But it only runs on Posix platforms, and most of the world is Windoze.

[Fulvio]

If we use MonetDB, then the data will be accessed at (approximately) memory speed.

Do you have some concrete benchmark?
Let's say a 10 million games database: how long it takes to find the games (not just the positions, if I would like to study the ending I need to see the continuation) with for example a king+rook vs king+rook+pawn ending?

MonetDB is a general purpose in-memory transaction based ACID compliant column based SQL store.
https://www.monetdb.org/Home

I probably misunderstood the screenshot you posted.
I thought you already had a SQL database, and I am genuinely interested in some chess benchmark.
The point is that traditionally general purpose SQL engines were too slow, for example:
https://triplehappy.wordpress.com/2016/ ... mentation/

However, as you pointed out, computer will keep becoming faster, and if it wasn't for the performance penalty, general purpose databases would be the obvious choice. And there is the amazing Lichess example, which successfully uses a distributed mongoDB database...

[Dann Corbit]

I do already have a chess database, but it is not general purpose.
The screen shot of the database that I showed was SQL Server.
But I have data in other database systems, also.

Much of what I am saying in this thread is also about things that I want, not things that I have.

I have not done a conversion to MonetDB yet, though I do not think it would be very difficult, since I wrote a real time data synchronization program that will move data from any ODBC data source to any other ODBC data source. But what I have right now works fine for what I use it for. Eventually, I will probably convert it, but I am not in a rush.

I also cannot share my database conveniently, since it has many terabytes of data.